Skip to content
v10.0.0-beta.171MIT

The Complete AI Development Toolkit — 110 skills, 36 agents, 162 hooks for full-stack development.

Changelog

All notable changes to the OrchestKit Claude Code Plugin will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

10.0.0-beta.171 (2026-10-11)

Features

  • hooks: add claude-haiku-5-5 to the model vocab and pricing (#4660) (4621212)
  • models: advance the haiku alias to claude-haiku-5-5 (#4661) (5c4c961)

Bug Fixes

  • agents: drop skill references that name no existing skill (#4709) (0e2bcab)
  • ci: five GA-blocking CI and test bugs, red-first (#4687) (fc17326)
  • drop the system-reminder wrapper; isolation only on 2 git agents (#4695) (1894035)
  • git-hooks: src guard accepts a counterpart changed on the branch (#4710) (f59bc76)
  • hooks: drop continueOnBlock from 4 command hooks for Desktop (#4685) (8c426ee)
  • hooks: four GA-blocking hook bugs, red-first (#4688) (59093c3)
  • hooks: ork cursor pins and #4688 follow-ups (#4713) (f65f016)
  • review-pr: post only when the user types --post (#4678) (d2a3a8f)
  • skills: inline fan-out skills, hook-free forks, shared target rule (#4693) (8a4716e)

10.0.0-beta.170 (2026-10-10)

Bug Fixes

  • hooks: tldr-summary skips a Read with allow_large (#4700) (13a6c2a)

10.0.0-beta.169 (2026-10-10)

Documentation

  • skills: MCP description cap is 4,096 characters since CC 2.1.296 (#4699) (9fae329)

10.0.0-beta.168 (2026-10-10)

Bug Fixes

  • hooks: price Sonnet 5.5 cache reads at $0.10 per MTok (#4698) (7abe08a)

10.0.0-beta.167 (2026-10-10)

Features

  • cc: adopt Claude Code 2.1.289 through 2.1.296 (#4702) (c5705af)

10.0.0-beta.166 (2026-10-09)

Features

  • review-pr: standards pass with a review-only standards file (#4671) (cfbd187)

Bug Fixes

  • ci: cache and bound the docs Playwright install (#4667) (379e1e7)
  • review-pr: run inline so its frontmatter hooks run (#4682) (1713ea0)
  • tests: restore exec bit on git-isolate.sh (#4681) (7626a41)

Documentation

  • doctor: CC 2.1.289 to 2.1.293 matrix rows, Haiku 5.5 (#4662) (732f9a5)
  • lab: playground for the Playwright install hang (#4665) (#4669) (eda91e5)

10.0.0-beta.165 (2026-10-07)

Features

  • prompt-focus: where your attention went, from your own prompts (#4657) (0ef71d2)

10.0.0-beta.164 (2026-10-07)

Bug Fixes

  • hooks: display-lint hint is shell-aware, model-only, once per session (#4655) (dc0b811)
  • skills: reference skills load inline; ork:auto routes e2e asks (#4656) (480e358)

10.0.0-beta.163 (2026-10-07)

Features

  • glyph: output mode dials, done sign-off, and a try-it page (#4647) (f917aac)

Miscellaneous

  • deps-dev: bump the vitest group across 1 directory with 2 updates (#4612) (9246bca)

10.0.0-beta.162 (2026-10-07)

Bug Fixes

  • deps: sharp 0.35.5 in docs/site for advisory 1241331 (#4653) (fb6742b)

10.0.0-beta.161 (2026-10-06)

Miscellaneous

  • deps-dev: bump vitest (#4611) (ef6ff6b)
  • deps: bump the npm-minor-patch group across 1 directory with 2 updates (#4613) (4f298b0)

10.0.0-beta.160 (2026-10-06)

Miscellaneous

10.0.0-beta.159 (2026-10-06)

Bug Fixes

  • deps: lock source-map-js 1.2.2 and proxy-addr 2.0.8 in all trees (#4642) (badfc82)

10.0.0-beta.158 (2026-10-05)

Bug Fixes

  • test: nested pre-push hook runs in full mode (#4635) (ee47e33)

10.0.0-beta.157 (2026-10-05)

Bug Fixes

  • lesson-cards: home via env, /lessons before corpus load (#4625) (8cbcb1d)

10.0.0-beta.156 (2026-10-05)

Bug Fixes

  • promote-lights: /lights takes a bare watch target (#4624) (fe0fb75)

10.0.0-beta.155 (2026-10-05)

Bug Fixes

  • lesson-cards: do not read process when HOME is unset (#4608) (1a2da0b)

Miscellaneous

  • deps-dev: bump the vitest group (#4618) (ede8569)
  • deps: bump the npm-minor-patch group in /docs/site with 8 updates (#4616) (0b71bfc)
  • deps: bump the remotion group (#4617) (166fe79)

CI/CD

10.0.0-beta.154 (2026-10-04)

Features

  • site: advertise Actions MCP and add try-safely sandbox page (#4596) (8f5c732)

10.0.0-beta.153 (2026-10-04)

Bug Fixes

  • ci: close docs.yml gate gaps on filter paths and cancel-in-progress (#4591) (7316a1d)

10.0.0-beta.152 (2026-10-04)

Bug Fixes

  • brainstorm: scope the turn-end survival claim (#4593) (c0cd49d)

10.0.0-beta.151 (2026-10-04)

Bug Fixes

  • test: retain dispatcher stderr in the session_id quarantine probe (#4592) (7586f63)

10.0.0-beta.150 (2026-10-04)

Bug Fixes

  • expect: scope browser close to the run's own session (#4594) (b3e6925)

10.0.0-beta.149 (2026-10-04)

Bug Fixes

  • test: make careful-guard link-to-temp case platform-stable (#4590) (75f0141)

10.0.0-beta.148 (2026-10-04)

Features

Bug Fixes

  • mods: register memory-lens in the marketplace, gate missing entries (#4587) (3d35ab4)

10.0.0-beta.147 (2026-10-03)

Features

  • cc: adopt Claude Code 2.1.285 through 2.1.288 (#4588) (742663e)

10.0.0-beta.146 (2026-10-02)

Bug Fixes

  • glyph: triage gate never falls back to the real Google Chrome (#4584) (9c41934)

10.0.0-beta.145 (2026-10-02)

Bug Fixes

  • hooks: do not spawn the session namer under test (#4569) (05e329d)

10.0.0-beta.144 (2026-10-02)

Bug Fixes

10.0.0-beta.143 (2026-10-02)

Bug Fixes

  • glyph: avoid macOS Chrome app launches in sandboxes (#4580) (23e2c69)

10.0.0-beta.142 (2026-10-02)

Miscellaneous

  • deps: bump dompurify from 3.4.14 to 3.4.16 in /docs/site (#4578) (1317df7)

10.0.0-beta.141 (2026-10-01)

Features

  • skills: careful and freeze, on-demand guard hooks in a skill (#4576) (7ce7810)

10.0.0-beta.140 (2026-10-01)

Features

  • review-pr: opt-in rule-check mode, one verifier per rule (#4573) (b2d4acd)

10.0.0-beta.139 (2026-10-01)

Features

  • skills: brainstorm diverges at effort low, implement effort ladder (#4572) (d90788d)

10.0.0-beta.138 (2026-10-01)

Features

  • hooks: warn before a model switch re-caches a warm large context (#4567) (3078407)

Bug Fixes

  • agents: raise security and code review agents to effort high (#4566) (8287954)
  • skills: teach auto tool_choice on Claude 5.5, dedup antipatterns (#4568) (f7f0856)

10.0.0-beta.137 (2026-09-30)

Miscellaneous

  • evals: tag run-evals claude launch as headless-ok (#4564) (aaf1e22)

10.0.0-beta.136 (2026-09-30)

Miscellaneous

10.0.0-beta.135 (2026-09-29)

Miscellaneous

  • ci: make the G1 standards gate a required main check (#4559) (f6e5fa0)

10.0.0-beta.134 (2026-09-29)

Miscellaneous

  • frontmatter: normalize allowed-tools values (#4556) (105a980)

10.0.0-beta.133 (2026-09-29)

Bug Fixes

  • models: mark Opus 4 / Sonnet 4 dated IDs retired (2026-06-15) (#4550) (62a32ad)

Miscellaneous

  • deps: bump ip-address from 10.4.0 to 10.7.2 in /src/mcp-server (#4548) (62d2efa)
  • release: pin stable channel to v9.8.1 (#4552) (5533e53)

Documentation

  • doctor: treat policyHelper sources as not observable in fallback (#4551) (59a1df7)

CI/CD

  • triage: read the triage model from a repo variable (#4555) (2b98f2c)

10.0.0-beta.132 (2026-09-28)

Features

  • cc: adopt CC 2.1.281 to 2.1.284 (#4545) (91aa9ee)
  • models: adopt Claude Sonnet 5.5 as the sonnet alias target (#4541) (7784b90)

Bug Fixes

  • ci: fetch the CC CHANGELOG raw, not base64 through execSync (#4546) (2cafd20)

10.0.0-beta.131 (2026-09-28)

Bug Fixes

  • agents: m3 script keeps quoted commas and validates first (#4533) (#4535) (18ea67b)

10.0.0-beta.130 (2026-09-28)

Code Refactoring

  • agents: move house frontmatter keys under metadata, drop context (m3) (#4531) (bd09735)

10.0.0-beta.129 (2026-09-28)

Features

  • mods: memory-lens recalls related memories on each prompt (#4530) (8371ff5)

10.0.0-beta.128 (2026-09-28)

Bug Fixes

  • ci: Docs Site Build always reports so docs-data PRs can merge (#4529) (0325ff1)
  • ci: traffic snapshot app token; release-please guard skips release/* bases (#4525) (73f86d5)

Miscellaneous

Code Refactoring

  • skills: move house frontmatter keys under metadata (m1) (#4521) (220adb4)

10.0.0-beta.127 (2026-09-28)

Bug Fixes

  • standards-gate: skip head steps when the head predates the gate (#4518) (89e8530)

10.0.0-beta.126 (2026-09-28)

Bug Fixes

  • standards: list agent as a Claude Code skill key in G1 registry (#4522) (6d0ef80)

10.0.0-beta.125 (2026-09-28)

Features

  • assess: Phases 2 and 2.5 run as a Workflow script (#4513) (69809f8)

10.0.0-beta.124 (2026-09-28)

CI/CD

  • port G1 advisory frontmatter gate from hq-ext-plugin (#4514) (10326db)

10.0.0-beta.123 (2026-09-28)

Bug Fixes

  • hooks: stash secret-bearing MCP results to a 0600 session file (#4511) (d160302)

10.0.0-beta.122 (2026-09-28)

Miscellaneous

  • deps-dev: bump @types/node in the npm-minor-patch group (#4502) (901aeac)
  • deps-dev: bump the npm-minor-patch group (#4504) (fb931c0)
  • deps: bump the npm-minor-patch group (#4503) (d162455)
  • deps: bump the npm-minor-patch group (#4505) (a3ec5f6)
  • deps: bump the npm-minor-patch group across 1 directory with 11 updates (#4507) (09b786f)
  • deps: bump the remotion group (#4506) (83d04f0)

CI/CD

  • bump the github-actions group with 6 updates (#4508) (d0624f5)

10.0.0-beta.121 (2026-09-28)

Bug Fixes

  • skills: pair defer_loading with tool search in llm-integration (#4498) (baa07a5)

10.0.0-beta.120 (2026-09-27)

Features

  • review-pr: Phases 3 and 4.5 run as a Workflow script (#4494) (9be3e7f)
  • site: name developer resources, expose WebMCP registration (#4497) (abdf05f)

Bug Fixes

  • skills: finish Wave 1 leftovers from the 2026-09-25 Opus 5.5 audit (#4495) (a89a746)

10.0.0-beta.119 (2026-09-27)

Features

  • evals: lane suite for triage, review, promote, ci-debug (#4488) (0d0d438)

10.0.0-beta.118 (2026-09-27)

Bug Fixes

  • ci: report docs build on every pull request (#4486) (b30450a)

10.0.0-beta.117 (2026-09-26)

Bug Fixes

  • analytics: spell the curl flag plainly in the gateway source (#4484) (0e814a0)

10.0.0-beta.116 (2026-09-26)

Features

  • analytics: add an OTEL gateway source (Loki + Prometheus) (#4479) (e50f3e0)

10.0.0-beta.115 (2026-09-26)

Bug Fixes

  • ci: reject untracked generated lab copies (#4480) (a80ff12)
  • docs: validate npm hosts and advertise markdown twins (#4481) (6f21d5c)
  • scripts: require schema evidence for hook exemptions (#4478) (ad3085d)

10.0.0-beta.114 (2026-09-26)

Bug Fixes

  • hooks: preserve consumer file content in write-headers (#4475) (bb0a7ae)
  • hooks: stop antipattern-warning overwriting tracked project rules (#4476) (bf0e999)

10.0.0-beta.113 (2026-09-26)

Bug Fixes

  • tests: isolate pre-push Test 12 from the real tree's hooks install (#4473) (a919e97)

10.0.0-beta.112 (2026-09-26)

Bug Fixes

  • lesson-cards: 1.0.7 Cancelled: by you; Waiting text (#4470) (add98ae)

10.0.0-beta.111 (2026-09-26)

Bug Fixes

  • evals: judge via SDK, price cache tokens, require ORK_EVALS_API_KEY (#4466) (fe3148b)
  • lesson-cards: plain cancel deny, no Error or Fix (#4468) (26ae33e)

10.0.0-beta.110 (2026-09-26)

Bug Fixes

  • promote-lights: a rejected status clear keeps the lights (#4464) (665bcf4)

10.0.0-beta.109 (2026-09-26)

Bug Fixes

  • lesson-cards: read the real Escape throw as cancelled by you (#4462) (fd5961d)

10.0.0-beta.108 (2026-09-26)

Miscellaneous

10.0.0-beta.107 (2026-09-26)

Bug Fixes

  • lesson-cards: draw the lesson once, in the card (#4453) (4ca3515)

10.0.0-beta.106 (2026-09-26)

Bug Fixes

10.0.0-beta.105 (2026-09-26)

Bug Fixes

  • promote-lights: draw the lights once, in the band (#4454) (30bc6e0)

10.0.0-beta.104 (2026-09-26)

Features

  • secrets-veil: opt-in copy of a masked value to the clipboard (#4437) (8ae5d2f)

10.0.0-beta.103 (2026-09-26)

Bug Fixes

  • promote-lights: find promote PR past first 30 open PRs (#4435) (ab35c75)
  • site: put the home hero art in the grid and redraw the nav mark (#4414) (522646b)

10.0.0-beta.102 (2026-09-26)

Bug Fixes

  • promote-lights: never draw another session's stored lights (#4447) (335754e)

10.0.0-beta.101 (2026-09-25)

Features

  • promote-lights: cap the band at five problem lines (#4445) (fa6fbdc)

Bug Fixes

  • lesson-cards: keep a stopped call to one short line (#4442) (9a43923)
  • secrets-veil: keep masking linear on long lines (#4441) (67a8b7d)

10.0.0-beta.100 (2026-09-25)

Features

  • promote-lights: compact band, one line per failing check (#4440) (d5faf0e)

Bug Fixes

  • lab: commit missing coderabbit-runnable-examples page (#4439) (f35a756)

Miscellaneous

  • evals: move eval judges and holdout grader to claude-opus-5-5 (#4443) (8fe5fb8)
  • skills: retire testing-patterns stub, disambiguate memory cluster (#4428) (01fd47b)

10.0.0-beta.99 (2026-09-25)

Bug Fixes

  • secrets-veil: fail closed when session.start never ran (#4436) (c319189)

10.0.0-beta.98 (2026-09-25)

Bug Fixes

  • lesson-cards: draw cards with $.ui.resolve and ask on block lessons (#4429) (2489dfc)
  • promote-lights: draw the band with $.ui.resolve, add /lights watch (#4430) (08a4377)
  • skills: runnable examples in verify refs, cover, review-pr (#4424 CodeRabbit follow-up) (#4433) (c42e850)

10.0.0-beta.97 (2026-09-25)

Bug Fixes

  • secrets-veil: show a toast with masked value and byte counts (#4431) (974e18b)

10.0.0-beta.96 (2026-09-25)

Features

  • verify: Phase 2 dispatch runs as a Workflow script (#4424) (65ae19d)

Documentation

  • readme: skills.sh badge and measured starter 12 (#4426) (3d9df6d)

10.0.0-beta.95 (2026-09-25)

Bug Fixes

  • promote-lights: match only real promote PRs into main (#4420) (7162dd1)

10.0.0-beta.94 (2026-09-25)

Miscellaneous

  • skills: Opus 5.5 wording, model ids, Claude API tool rules (#4415) (720af09)

10.0.0-beta.93 (2026-09-25)

Miscellaneous

  • skills: Opus 5.5 budgets, split create-pr and dream (#4416) (e6388d9)

10.0.0-beta.92 (2026-09-25)

Bug Fixes

  • mods: port the 3 mods to the CC 2.1.282 mods API (#4413) (8ad54b3)

10.0.0-beta.91 (2026-09-25)

Bug Fixes

  • cover: heal-loop never rewrites an expected value (#4407) (7d8a533)

Documentation

  • memory: drop LIVE claims for removed memory-writer hook (#4410) (05982df)
  • memory: remove stale memory-writer.ts hook reference (#4409) (ed3e259)

10.0.0-beta.90 (2026-09-25)

Features

  • cover: gate generated tests on behaviour and target by risk (#4408) (28111e9)

Bug Fixes

  • hooks: word-based wrapper option parsing in the egress check (#4406) (9e163a0)

10.0.0-beta.89 (2026-09-24)

Features

  • skills: error-analysis, evals-first failure taxonomy from traces (#4402) (47c9489)

Bug Fixes

  • hooks: exact option grammar for command wrappers in the egress check (#4404) (8db0cde)

10.0.0-beta.88 (2026-09-24)

Bug Fixes

  • docs-site: align three stale landing.spec cases with hero A (#4398) (eb5bc52)
  • hooks: resolve relative interpreter program paths against the effective working directory (#4400) (5b42ec9)

10.0.0-beta.87 (2026-09-24)

Bug Fixes

  • hooks: fail-closed on shell-expanded interpreter arguments (#4397) (4f47f3e)

10.0.0-beta.86 (2026-09-24)

Bug Fixes

  • hooks: skip project rules write when global copy is identical (#4393) (a8ab7c2)
  • hooks: stdin path handling in egress interpreter argv (#4396) (5941f90)

10.0.0-beta.85 (2026-09-24)

Bug Fixes

  • hooks: fail-closed interpreter argument classification in egress check (#4394) (6b57b3a)
  • hooks: per-interpreter flag handling in egress pipe check (#4390) (4366980)

10.0.0-beta.84 (2026-09-23)

Features

  • git-hooks: opt-in targeted pre-push mode with a 2-slot governor (#4238) (#4389) (4d1dfe7)

10.0.0-beta.83 (2026-09-23)

Bug Fixes

  • hooks: emit PermissionRequest decision.behavior for auto-approve (#4374) (7725095)
  • hooks: read tool_response and lock attribution state (sweep T1) (#4386) (39d67e5)
  • security: XS hardening batch from the 2026-09-17 audit (#4220) (#4383) (5fbfa83)
  • skills: grant Stitch tools the skills call, and guard skill MCP refs (#4376) (3cbe64d)
  • skills: keep rc capture and cleanup working under set -e (sweep T2) (#4387) (df86f66)

10.0.0-beta.82 (2026-09-23)

Features

  • docs-site: hero option A conductor bleed (#4345) (d6abd66)

Bug Fixes

  • hooks: load telemetry sinks only from user scope over https (#4218) (#4380) (f5785fe)
  • hooks: secret-handler AUDIT default and redacted security-audit log (#4379) (ee487e4)
  • skills: Gemini model IDs and provider wording missed by #4381 (#4382) (b22d76a)

10.0.0-beta.81 (2026-09-23)

Bug Fixes

  • docs: skip pycache and .pyc in generated skill folder lists (#4378) (15853a8)
  • hooks: age-cap session events and bound Stop git status (#4367) (a1ed4e0)
  • skills: correct provider API parameters in skill examples (#4381) (922e2bc)
  • skills: drop dead agent: bindings under context inherit (F26) (#4358) (de2b771)

10.0.0-beta.80 (2026-09-23)

Features

  • agents: adopt Anthropic's Opus 5.5 working guidance (#4363) (af501e2)

Bug Fixes

  • agents: grant stitch and storybook-mcp tools (F12) (#4362) (5c3662d)
  • hooks: ConfigChange --no-verify false positives (SC47 F3) (#4368) (ee96f03)
  • hooks: fall back to raw --no-verify scan on settings parse failure (#4370) (2d5b9cc)
  • hooks: require ORK_TEST_MODE for ORK_HOOKS_DIST_DIR (#4371) (060195e)
  • llm-integration: make CoT examples thinking-model aware (#4363) (#4372) (94d1d04)
  • prompts: apply high-confidence Opus 5.5 prompt-audit findings (#4366) (bb6f995)
  • site: guard view transitions against aborted transitions (PostHog issue 01a0a900) (#4353) (d05adb9)
  • skills: replace phantom Stitch tool names in prose (#4373) (15f5451)
  • testing-e2e: reject backtick refs that escape skill root (#4364) (17bf40f)

Miscellaneous

  • docker: bump node from 50c8e8c to ebfe2f9 (#4369) (ef53a00)

10.0.0-beta.79 (2026-09-22)

Features

  • cc: adopt CC 2.1.278 and 2.1.280, Opus 5.5 as the default Opus (#4357) (606a574)

Bug Fixes

  • design: reserve.py exits 3 on generation-cap exhaustion (#4359) (0c9e678)
  • release: stop marketplace count stamp from rewriting stable ork (#4355) (32e93a8)
  • site: close orank OG, llms.txt, and /docs landing gaps (#4354) (2488c3f)
  • testing-e2e: repoint dead planner-agent reference (F27) (#4356) (e79f6ce)

10.0.0-beta.78 (2026-09-22)

Bug Fixes

Miscellaneous

  • playground: drop the homeos-arieh exemplar (#4348) (104752c)

10.0.0-beta.77 (2026-09-22)

Bug Fixes

  • site: align title and social meta with the OG card (#4343) (7cba8c9)

10.0.0-beta.76 (2026-09-22)

Bug Fixes

  • dream: classify index budget moves by filename prefix (#4342) (c6189a5)
  • skills: stacked-PR rebase uses the onto form and real trunk (#4337) (94edc1b)

Documentation

10.0.0-beta.75 (2026-09-22)

Features

  • site: new link card with live skills, agents and hooks counts (#4335) (8908e53)

Bug Fixes

  • hooks: stacked-PR guards read the trunk from origin/HEAD (#4336) (5e822b0)

10.0.0-beta.74 (2026-09-21)

Bug Fixes

  • hooks: redact-secrets scans object-shaped Bash tool_response (#4330) (fc249a1)
  • scripts: derive-cc-output-keys --check arbitrates EVENTS_WITH_HOOK_EVENT_NAME (#4331) (f8340e9)

10.0.0-beta.73 (2026-09-21)

Features

  • site: make Jev search typeahead rerank non-blocking behind its own flag (#4325) (0c62edb)

Miscellaneous

  • deps: bump dotenv from 17.4.2 to 18.0.0 in /orchestkit-demos (#4322) (892efaf)

CI/CD

  • bump the github-actions group with 7 updates (#4323) (f10dbf7)

10.0.0-beta.72 (2026-09-21)

Bug Fixes

  • test: separate a Chrome launch failure from an empty render (#4324) (02cfe98), closes #4296

10.0.0-beta.71 (2026-09-21)

Bug Fixes

  • hooks: allow tip-equal dist restores without MERGE_HEAD (#4308) (4aaaf17)

Miscellaneous

  • deps-dev: bump @types/node (#4312) (06f1792)
  • deps-dev: bump the npm-minor-patch group (#4317) (2d8dff8)
  • deps-dev: bump the npm-minor-patch group across 1 directory with 2 updates (#4315) (871dfa2)
  • deps-dev: bump the vitest group (#4320) (8b3ce70)
  • deps-dev: bump the vitest group across 1 directory with 2 updates (#4314) (737aa81)
  • deps-dev: bump vitest from 5.0.0 to 5.0.1 in the vitest group (#4313) (3b354d1)
  • deps-dev: bump vitest in /docs/site in the vitest group (#4316) (36c7b06)
  • deps: bump the npm-minor-patch group (#4318) (e132c60)
  • deps: bump the npm-minor-patch group across 1 directory with 6 updates (#4321) (2999add)
  • deps: bump the remotion group (#4319) (5fa63e5)

10.0.0-beta.70 (2026-09-20)

Miscellaneous

  • cc: raise the CC support floor 2.1.251 to 2.1.277 (#4305) (02cb935)

10.0.0-beta.69 (2026-09-20)

Bug Fixes

  • ci: skip the playground gate on test-only diffs (#4303) (fea779f)

10.0.0-beta.68 (2026-09-20)

Features

  • jev: add shared shadow foundation and passive Codex adapter (#4286) (a70f076)

10.0.0-beta.67 (2026-09-20)

Features

  • expect: jev live per-field decisions in form steps (#4284) (145521b)

10.0.0-beta.66 (2026-09-20)

Features

  • glyph: add a triage page template with a parsable DECIDE pick (#4267) (8cd716c)

Bug Fixes

  • jev: persist paired shadow decisions across local seams (#4278) (549f38d)

10.0.0-beta.65 (2026-09-20)

Features

  • jev: add conservative weak outcome sidecars (#4281) (b8cb918)

Bug Fixes

  • docs-site: stop the picker repeating the hero install command (#4293) (f3f9c97)

10.0.0-beta.64 (2026-09-20)

Features

  • site: search autocomplete suggestions in docs search (#4258) (a9cc587)

Bug Fixes

  • hooks: never emit hookSpecificOutput without hookEventName (#4285) (8d568b6)

10.0.0-beta.63 (2026-09-19)

Bug Fixes

  • eval: document Langfuse Python SDK v4 and fail zero-item runs (#4272) (5239ca1)

10.0.0-beta.62 (2026-09-19)

Bug Fixes

  • hooks: gate session-identity generator to one spawn per session (#4270) (3622749)

Code Refactoring

10.0.0-beta.61 (2026-09-19)

Features

  • manifest: wire webhook URL/token through userConfig (#4273) (788fdcd)

Bug Fixes

  • build: drop object-shaped triggers from built SKILL.md (#4276) (526873c)
  • ci: compare hook-bundle guard against current base tip (#4283) (bc35750)
  • git: reject release bundle helper failures (#4287) (7ac8bc9)
  • jev: validate confidence and expose shadow disagreements (#4271) (66e2ccc)

10.0.0-beta.60 (2026-09-19)

Features

  • browser: adopt agent-browser 0.38.1 (delta snapshots, if-changed shots, persistent refs, 30 fps, WebMCP) (#4260) (6e23e38)
  • site: generated Related block on reference pages (#4253) (9b6b576)

10.0.0-beta.59 (2026-09-19)

Features

  • codex: add glyph skill adapter and export coverage (#4277) (ab8d756)

Bug Fixes

  • hooks: key credential-path reject on the path, not the reader (#4279) (d8f0742)

10.0.0-beta.58 (2026-09-19)

Features

  • doctor: check per-agent memory dirs for orphans and stale notes (#4263) (e01f577)

Bug Fixes

  • ci: restore py311 compatibility for shipped skill scripts (#4275) (76f6923)
  • glyph: render once instead of handing off to ork:glyph (#4261) (7e63c45)
  • hooks: close auto-approve bypass via substitution and redirects (#4264) (4a246c6)

Documentation

  • contributing: add a first pull request example (#4269) (a58c4b0)
  • seo: internal link-graph audit for the docs site (#4251) (3dd24c1)

CI/CD

  • compile-check and ruff-lint the 51 skill scripts (#4265) (9390c8d), closes #3500
  • pre-commit: block plugins/skills edits missing the src counterpart (#4266) (0e93be5), closes #3499

10.0.0-beta.57 (2026-09-19)

Bug Fixes

  • ci: name both causes when hook bundles differ (#4256) (9425d85)
  • release: retry the GHCR manifest preflight (#4257) (aa959ff)

10.0.0-beta.56 (2026-09-19)

Bug Fixes

  • playground: render mermaid when a tab opens (#4252) (6b1a5a0)

Miscellaneous

10.0.0-beta.55 (2026-09-18)

Documentation

  • doctor: extend the CC version matrix through 2.1.277 (#4247) (68ed58e)

10.0.0-beta.54 (2026-09-18)

Features

Bug Fixes

10.0.0-beta.53 (2026-09-18)

Bug Fixes

  • hooks: pre-push applies one load back-off to every stage and keeps the security log (#4238) (#4241) (961349b)

10.0.0-beta.52 (2026-09-18)

Features

  • hooks: Jev routing seam for ork:auto behind ORK_ROUTE_JEV (#4233) (#4237) (ef91b52)

10.0.0-beta.51 (2026-09-18)

Bug Fixes

  • expect: pass agent result and summary to report.py on stdin, never as source (#4234) (6466d0e)
  • hooks: read last_assistant_message at SubagentStop, agent_output is never sent (#4232) (355d167)

10.0.0-beta.50 (2026-09-18)

Features

  • expect: Jev act mode for the step pick (ORK_EXPECT_JEV) (#4230) (c5668f7)

10.0.0-beta.49 (2026-09-18)

Features

  • expect: opt-in Jev shadow pick per step (ORK_EXPECT_JEV_SHADOW) (#4228) (c2f52bf)

10.0.0-beta.48 (2026-09-17)

Bug Fixes

  • security: stop logging token reference, escape lab attributes fully (#4226) (fd42932)

10.0.0-beta.47 (2026-09-17)

Miscellaneous

  • security: genericize op:// paths and home dirs in public files (#4223) (a827671)

Documentation

  • replace U+2014 in sixteen rule titles with colon (#4221) (8effd3d)

10.0.0-beta.46 (2026-09-17)

Features

  • glyph: v3 render shape, 50-line budget, domain icons, inventory template (#4212) (7b973fc)

10.0.0-beta.45 (2026-09-17)

Features

  • hooks: criteria-B for both category paths, Jev cascade at 0.8 (#4211) (ff9e99e)

10.0.0-beta.44 (2026-09-17)

Features

  • hooks: opt-in shadow session category as a typed Choice (#4208) (b1769c7)

10.0.0-beta.43 (2026-09-16)

Bug Fixes

  • ci: shrink the release rebuild window to the hooks-only set (#4205) (2bd4462)

10.0.0-beta.42 (2026-09-16)

Bug Fixes

  • hooks: output-validator pass warnings are log-only (#4202) (aaee1b1)

10.0.0-beta.41 (2026-09-16)

Bug Fixes

  • hooks: read subagent result via shared reader in output-validator (#4199) (40181a0)
  • tests: exec-bit preflight enumerates NUL-delimited, quoting off (#4200) (5ff4038)

10.0.0-beta.40 (2026-09-16)

Bug Fixes

CI/CD

10.0.0-beta.39 (2026-09-16)

Features

  • mods: add secrets-veil mod with pattern and entropy masking (#4189) (20bdf71)

10.0.0-beta.38 (2026-09-16)

Documentation

  • create-pr: drop stale lab-data.ts conflict advice (#4191) (bdcd608)

10.0.0-beta.37 (2026-09-16)

Bug Fixes

  • site: generate lab-data.ts at build time instead of committing it (#4188) (0732f0d)

10.0.0-beta.36 (2026-09-16)

Features

  • mods: add promote-lights mod for CI status monitoring (#4165) (9974271)

Bug Fixes

  • site: union merge driver for generated lab-data.ts (#4184) (733dc7f)

10.0.0-beta.35 (2026-09-16)

Bug Fixes

  • glyph: cap non-TTY renders at 72 columns (#4179) (988cbeb)

10.0.0-beta.34 (2026-09-16)

Features

  • mods: add lesson-cards mod for hint cards on matched patterns (#4169) (c06d532)

Bug Fixes

  • ci: allow .devin-plugin/plugin.json in release rebuild guard (#4178) (dbe7064), closes #4173
  • hooks: anchor stdin-timeout test write to child readiness (#4163) (8204546)
  • hooks: read subagent result from last_assistant_message (GH-4158) (#4168) (9616be0)
  • lab: exact URL matches in yonyon crosslink checker (refs #4144) (#4161) (1de3f6a)
  • site: report reduced motion in host picker tests (#4170) (9d65921)
  • stop the pre-commit runner from chmod-mutating tracked scripts (#4171) (8494ef0)

CI/CD

  • playground: treat generated Lab artifacts as inert (#4164) (52d4793)

10.0.0-beta.33 (2026-09-15)

Bug Fixes

  • docs: name the PostHog ph_ cookie on the privacy page (#4152) (10bab21)
  • site: crosslink apex Yonyon brand page from /yonyon (#4154) (f59a74f)

10.0.0-beta.32 (2026-09-15)

Features

  • site: add Devin host card and rebuild picker as animated grid (#4150) (a910b39)

Bug Fixes

  • site: render host cards visibly during SSR (#4153) (dbc05e2)

10.0.0-beta.31 (2026-09-15)

Features

  • devin: expose OrchestKit skills to Devin plugin install (#4148) (6004e79)

10.0.0-beta.30 (2026-09-15)

Documentation

  • cc: join the 2.1.269 to 2.1.272 rows to the matrix table (#4142) (e025081)

10.0.0-beta.29 (2026-09-15)

Features

  • site: close honest ora.ai score holes for discovery and SDKs (#4137) (22298e8)

10.0.0-beta.28 (2026-09-15)

Bug Fixes

  • release: retry the mcp-publisher and GHCR token downloads (#4139) (ce96210)

Documentation

  • cc: adopt CC 2.1.269 to 2.1.272, latest_known 2.1.263 to 2.1.272 (#4138) (5312c59)

10.0.0-beta.27 (2026-09-15)

Features

  • product MCP sandbox for orchestkit.yonyon.ai (#4131) (205cea7)
  • site: publish the auth chain with optional agent identity (#4129) (e51000b)

Bug Fixes

  • tests: sample the paste guard latency delta three times (GH-4130) (#4132) (9817ed7)

Documentation

10.0.0-beta.26 (2026-09-14)

Bug Fixes

  • build: generate docs data after stamping counts (#4126) (335a4d4)

10.0.0-beta.25 (2026-09-14)

Bug Fixes

  • site: withhold the negotiated Markdown variant from shared caches (#4125) (5f8997b)

CI/CD

  • name the src/ path when build drift is under plugins/ork/skills (GH-3507) (#4121) (83300dc)

10.0.0-beta.24 (2026-09-14)

Bug Fixes

Miscellaneous

10.0.0-beta.23 (2026-09-14)

Features

  • create-pr: run CodeRabbit CLI review before opening the PR (#4117) (34a0456)

Bug Fixes

  • ci: track classic.PreToolUse rename in function hooks canary (#4102) (b75194e)
  • create-pr: clarify CodeRabbit credit spend and fix root tools key (#4118) (46c58a7)
  • tests: report hook timeouts in test-image-paste-guard instead of swallowing them (GH-4087) (#4111) (504040c)

Miscellaneous

  • deps: bump orchestkit-demos npm-minor-patch group with the peer resolution (supersedes GH-4098) (#4110) (b46c17f)
  • deps: bump the npm-minor-patch group in /docs/site with 2 updates (#4120) (d9143e6)

Documentation

  • site: generate a reference page for every user-invocable skill (GH-3902) (#4114) (8dc4863)

10.0.0-beta.22 (2026-09-14)

Features

  • site: enable PostHog session replay on /docs/reference/* only (#4105) (6f04509)
  • site: register WebMCP page tools (search_docs, get_page) with server-rendered evidence (#4107) (697f304)
  • site: static homepage, hero install command, host_selected, /mcp passthrough (#4103) (26150a8)

Bug Fixes

  • docs: give Lab pages a way back to the site, share cards and page views (#4101) (4e0462b)
  • hooks: stop worktree pre-push fixtures flipping core.bare (#4070) (#4090) (4c7ff5f)
  • security: close CodeQL XSS highs, scope previews, clear zizmor (#4100) (a273aa6)
  • site: correct plugin install command and split install analytics (#4104) (1617f79)
  • site: match the hero install test to the corrected install command (#4112) (e622751)

Miscellaneous

Documentation

  • site: split configure reference and make API/MCP pages findable by name (ora.ai) (#4106) (496649f)

CI/CD

  • bump the github-actions group across 1 directory with 5 updates (#4099) (4b538f7)

10.0.0-beta.21 (2026-09-13)

Bug Fixes

10.0.0-beta.20 (2026-09-13)

Documentation

  • load diagnosis, zsh word splitting, rsync dialect guidance (#4081) (b449653)

10.0.0-beta.19 (2026-09-13)

Miscellaneous

  • plugins: drop test configs from the hooks mirror (#4079) (94a030d)

10.0.0-beta.18 (2026-09-12)

Bug Fixes

10.0.0-beta.17 (2026-09-12)

Features

10.0.0-beta.16 (2026-09-12)

Features

  • evals: claude plugin eval suite, calibration, and the fixes it found (#4072) (2d39e0a)

10.0.0-beta.15 (2026-09-11)

Bug Fixes

10.0.0-beta.14 (2026-09-11)

Bug Fixes

  • build: strip devDependencies from the hooks manifest mirror (#3971) (#4029) (7a04718)
  • hooks: display-lint no longer nudges the shape it recommends (#3936) (#4037) (378c6c1)
  • hooks: remove matcher-group if keys CC never supported (#4062) (30e1be6)
  • release: stamp the codex and cursor plugin.json mirrors (#4038) (#4046) (23d4ce3)
  • skills: clear remaining allowed-tools exceptions (#4043) (929fd91)
  • skills: gate allowed-tools against a skill's own prose (#4040) (7508648), closes #4008
  • tests: pipe the remaining here-string in dead-generation guard (#4061) (dc28a58)
  • tests: replace here-strings in dead-generation guard for bash 3.2 (#4059) (4340d1e)
  • tests: stop two tests rewriting the working tree (#4025) (#4047) (b21b4f6)

Miscellaneous

Documentation

  • doctor: triage CC 2.1.264 to 2.1.268 into the compatibility matrix (#4058) (192dbf8)
  • sync explore and review-pr to post-2.1.232 fork model (#4063) (e13a9a3)

CI/CD

  • bump anthropics/claude-code-action in the github-actions group (#4035) (00c45f7)

10.0.0-beta.13 (2026-09-10)

Bug Fixes

  • hooks: migrate the src/hooks suite to vitest 5 (#3981) (#4028) (1c15851)
  • security: npm audit gate could not run without GNU timeout, misreported as no output (#4015) (#4023) (b37f3db)

Documentation

  • site: host picker, library tabs, and readable hook lists (#4022) (7e2d8ce)

10.0.0-beta.12 (2026-09-09)

Miscellaneous

  • deps: bump hono from 4.13.0 to 4.13.7 in /src/mcp-server (#4014) (cb6ccc4)

Documentation

  • engines: generic floor Lab and portable skill invoke (#4018) (4f38cbc)
  • site: host-first install wizard and multi-host getting started (#4016) (efd9640)

10.0.0-beta.11 (2026-09-08)

Features

  • codex: ship the ork-mech profile and the plugin cache-lag note (#4012) (bcc122d)

10.0.0-beta.10 (2026-09-08)

Features

  • cursor: export rules to .cursor-plugin and rewrite wrapper paths at generation time (#4011) (7476d54)
  • pi: ship a pi manifest and a .pi/mcp.json allowlist template (#4009) (7b6f278)

10.0.0-beta.9 (2026-09-08)

Documentation

  • engines: pi, Codex and Cursor guide, matrix and audit (#4006) (101922a)

10.0.0-beta.8 (2026-09-08)

CI/CD

  • deps: ignore vitest major under /src/hooks (#3996) (34c940d)

10.0.0-beta.7 (2026-09-08)

Bug Fixes

  • ci: report an unreadable version as SKIP, not DRIFT (#3989) (bea3149)

Documentation

  • rules: scope the modules-key ban to shipped hooks.json (#3993) (31cfe79)

CI/CD

  • playground: treat .claude/rules/ as inert (#3994) (75a3b26)

10.0.0-beta.6 (2026-09-08)

Miscellaneous

  • docs: drop private-plugin specifics from public source (#3990) (0ac908c)

10.0.0-beta.5 (2026-09-07)

Features

  • skills: page-serve, hand a human a page at a port-free URL (#3899) (#3987) (1e12502)

Bug Fixes

Documentation

10.0.0-beta.4 (2026-09-07)

Miscellaneous

  • deps-dev: bump the vitest group across 1 directory with 2 updates (#3978) (5bc26d9)
  • deps-dev: bump vitest from 4.1.11 to 5.0.0 (#3960) (104afcf)
  • deps: bump the npm-minor-patch group across 1 directory with 7 updates (#3965) (002dd8f)
  • deps: bump the remotion group across 1 directory with 21 updates (#3964) (4bbaeda)

CI/CD

10.0.0-beta.3 (2026-09-07)

Miscellaneous

  • deps-dev: bump @types/node (#3958) (56ff3c7)
  • deps-dev: bump the npm-minor-patch group in /src/hooks with 2 updates (#3957) (7ffa630)

10.0.0-beta.2 (2026-09-07)

Features

Miscellaneous

10.0.0-beta.1 (2026-09-06)

Miscellaneous

  • release: flip the prerelease train to beta (#3952) (045b7f6)

10.0.0-alpha.86 (2026-09-06)

Documentation

10.0.0-alpha.85 (2026-09-06)

Miscellaneous

10.0.0-alpha.84 (2026-09-06)

Code Refactoring

10.0.0-alpha.83 (2026-09-06)

Miscellaneous

Code Refactoring

10.0.0-alpha.82 (2026-09-06)

Bug Fixes

  • ci: fail 37 gates that passed on absent input (#3938) (016dbe5)

10.0.0-alpha.81 (2026-09-06)

Documentation

  • cc: triage CC 2.1.262 and 2.1.263, latest_known 2.1.261 to 2.1.263 (#3933) (550e5c8)

10.0.0-alpha.80 (2026-09-05)

Documentation

  • cc: triage CC 2.1.259 to 2.1.261, latest_known 2.1.258 to 2.1.261 (#3929) (8c225ca)

10.0.0-alpha.79 (2026-09-05)

Bug Fixes

  • skills: make activation audits portable and coverage-aware (#3923) (a3f64a7)

10.0.0-alpha.78 (2026-09-05)

Bug Fixes

10.0.0-alpha.77 (2026-09-05)

Documentation

10.0.0-alpha.76 (2026-09-04)

Features

10.0.0-alpha.75 (2026-09-04)

Bug Fixes

  • announce: preflight release API credential (#3916) (908c1bc)

10.0.0-alpha.74 (2026-09-03)

Features

  • create-pr: CodeRabbit harvest stage, refute then fix or dismiss, resolve every thread (#3910) (cb2e99f)

Documentation

  • create-pr: a zero from the harvest can mean never reviewed (#3912) (56029de)

10.0.0-alpha.73 (2026-09-02)

Features

  • pricing: adopt gemini-3.8-flash as its own pricing row (#3905) (725471b)

Bug Fixes

  • pricing: a hyphenated suffix never inherits a shorter model's row (#3909) (59af1fb)

Documentation

  • doctor: repoint the model-routing reference at the live sources (#3908) (6c48343)

10.0.0-alpha.72 (2026-09-02)

Features

  • doctor: warn when a project scope turns the sandbox off (#3877) (#3887) (16221f1)
  • hooks: honour CC managed modelPricing in the cost estimator (#3889) (cc619bd), closes #3878
  • site: real analytics pin + install_copied/star_clicked (A1-A3) (#3882) (2868be7)

Bug Fixes

  • skills: accept relative and SKILL_DIR paths in the lint (#3822 step 1) (#3890) (bc042f2)

Miscellaneous

  • deps: weekly upstream pin sync (#3886) (22e7b14)
  • hooks: delete deprecated generate-http-hooks CLI (channel 1) (#3888) (cd56e69)

10.0.0-alpha.71 (2026-09-02)

Miscellaneous

  • cc: adopt CC 2.1.258, latest_known to 2.1.258 (#3883) (6752906)

Documentation

CI/CD

  • probe: pin the model price table to the binary's pricing tiers (#3885) (5d53ea9), closes #3878

10.0.0-alpha.70 (2026-09-02)

Miscellaneous

10.0.0-alpha.69 (2026-09-01)

Miscellaneous

  • agents: pin 15 agent model tiers (2026-09-01 routing) (#3872) (f8b94b1)

10.0.0-alpha.68 (2026-09-01)

Features

  • cc: CC 2.1.257 follow-ups on top of the adoption PR (#3871) (18e60a0)

Bug Fixes

Miscellaneous

  • cc-adoption: advance latest_known to 2.1.252, featureless for ork (#3863) (e1394d6)
  • cc: adopt CC 2.1.252 and 2.1.257, latest_known to 2.1.257 (#3870) (3967924)
  • hooks: delete legacy debris the purge never covered (#3868) (102887c)

10.0.0-alpha.67 (2026-09-01)

Bug Fixes

  • setup: drop 11 deny rules CC cannot enforce, payload v4 (#3855) (4ca49aa)

Miscellaneous

  • deps-dev: bump the npm-minor-patch group in /src/hooks with 2 updates (#3824) (4982e01)
  • deps-dev: bump the npm-minor-patch group with 2 updates (#3825) (896c78c)
  • deps: bump the remotion group (#3826) (3437122)

CI/CD

  • playground: exempt docs/audits from the playground gate (#3853) (41fe011), closes #3848

10.0.0-alpha.66 (2026-08-31)

Features

  • dream: budget pass and rule-based demotion for MEMORY.md (#3833) (4687957), closes #3741
  • hooks: name untracked playground pages at SessionStart (#3831) (283a84a), closes #3727
  • setup: operator-scope permission vehicle for the purge (#3836) (cec5bdf)

Bug Fixes

  • hooks: runner names a dist bundle that exists but cannot load (#3819) (cd2bfdc), closes #3817
  • hooks: task nudge gives up after N unanswered prompts (#3832) (5fa000a), closes #3733
  • verify: evidence gate becomes a script, not prose (#3830) (f6136cc)

Miscellaneous

  • cc-watch: drop the stale 2.1.219 parse_failed row (#3847) (612ae31)
  • deps-dev: bump @types/node (#3823) (d943481)
  • hooks: purge wave 1, six diverging sub-hooks deleted (#3838) (11d3b7d)
  • hooks: purge wave 2, the ask storm ends (#3839) (10c7d2c)
  • hooks: purge wave 3, file-guard and credential-read-guard deleted (#3840) (c88cb78)
  • hooks: purge wave 4, the blocker and git-validator deleted (#3841) (60c2725)

Documentation

  • audits: platform adoption triage, 08-01 to 08-27 (#3845) (96c4ed8)
  • rules: divergence-purge verdicts land in the drift register (#3837) (235335e)

CI/CD

  • bump the github-actions group across 1 directory with 5 updates (#3829) (54c4185)

10.0.0-alpha.65 (2026-08-30)

Bug Fixes

  • hooks: redact-secrets reads tool_response, the field CC sends (#3815) (9e01296), closes #3725

Documentation

10.0.0-alpha.64 (2026-08-30)

Bug Fixes

  • hooks: cross-instance-test-validator checks the session's files at Stop (#3812) (c7608b5)

Documentation

  • hooks: CLAUDE_ENV_FILE exports reach Bash only, not later hooks (#3814) (70af11c), closes #3806
  • playgrounds: land three 2026-08-29 session explainers (#3811) (a53b164)

10.0.0-alpha.63 (2026-08-30)

Features

Bug Fixes

10.0.0-alpha.62 (2026-08-29)

Features

  • agents: experimental.cacheTtl 1h on five long-running agents (#3797) (c741b3b)
  • ci: --restricted smoke lane against a stub Messages API (#3774) (#3799) (65d9812)
  • glyph: v3 two-dial front door (audience x surface) (#3793) (8bbc63d)
  • hooks: a maxTurns stop is a partial result, continue the agent (CC 2.1.246) (#3796) (b2421d8)
  • hooks: asyncRewake on the three background verifiers (#3800) (6026443)
  • hooks: PreModelSwitch consent gate + PostModelSwitch telemetry (CC 2.1.251) (#3792) (f9bf02f)
  • hooks: prompt-type hooks measured against the Bash guards (#3802) (0e50084)
  • hooks: verdict telemetry + probes; fix two guards they exposed (#3805) (f55ee5f)

Bug Fixes

  • assess: resolve a conversational reference instead of refusing (#3781) (435a864)
  • cc-watch: two gates reported verdicts they had not established (#3779) (11663ad)
  • ci: make npm run typecheck actually see hook tests (#3761) (a130ac5)
  • hooks: cc-version-check never ran, and never said so (#3759) (b7eacd6)
  • hooks: stale-import-detector never ran its grep (#3801) (#3803) (c50828b)
  • hooks: stdin data handler ran the hook a second time (#3778) (190a2d7)
  • skills: bump upstream pins and correct what the bodies claim (#3757) (520bf59)
  • watch: three sensors reported states they never observed (#3755) (66499fb)

Miscellaneous

  • cc-support: latest_known 2.1.241 -> 2.1.251 + hooks README drift (#3791) (cbfa694)
  • cc-support: support floor 2.1.220 -> 2.1.251 (#3794) (2aed0b0)
  • cc-watch: snapshot upstream CHANGELOG (2.1.250) (#3776) (e371fca)
  • cc-watch: snapshot upstream CHANGELOG (2.1.251) (#3790) (c2650a7)

Documentation

  • messaging: cross-session reply and platform notes (CC 2.1.248) (#3798) (b1a2d7b)

10.0.0-alpha.61 (2026-08-25)

Features

  • ork-codex: ship ork-implement and one install matrix (#3753) (dc3bfe8)

10.0.0-alpha.60 (2026-08-25)

Features

Bug Fixes

  • ork: Cursor import rejects ./plugins/ork as invalid_argument (#3750) (303b1f3)

10.0.0-alpha.59 (2026-08-25)

Features

  • ci: playground gate must verify the page is viewable (#3745) (053f6e3)

Bug Fixes

  • hooks: make a subagent empty result loud instead of silent (#3744) (5641a58)

Miscellaneous

  • ci: add CodeRabbit as an advisory review lane (#3746) (13d15ae)

10.0.0-alpha.58 (2026-08-25)

Bug Fixes

  • evals: run the train/test split through jq, mikefarah yq has no -y (#3730) (57cd01d)
  • hooks: guards must parse shell structure, not raw text (#3742) (d20cae0)
  • hooks: read git status without taking the index lock (#3743) (17d4285)
  • hooks: restrict-bash must not read quoted pipes as shell operators (#3740) (b7f425b)

10.0.0-alpha.57 (2026-08-25)

Bug Fixes

  • cc-watch: fallback stub ignores below-floor sentinels (#3724) (3f5359f)

CI/CD

  • register the repo-root plugin.json with the drift roster (#3728) (2c7d758)

10.0.0-alpha.56 (2026-08-24)

Bug Fixes

  • hooks: debt-surfacer reads a HEAD-keyed ledger (#3710) (f075001), closes #3708
  • hooks: redact GitLab tokens in all three redaction layers (#3722) (18f3aa6), closes #3589

Miscellaneous

  • cc-watch: advance latest_known to 2.1.241 with triage doc (#3723) (104b538)
  • cc-watch: snapshot upstream CHANGELOG (2.1.241) (#3719) (f2c8662)
  • orank: repo-root plugin.json mirror and fuller pricing.md (#3712) (849b055), closes #3675

Documentation

  • cc-adoption: fix four skill docs describing pre-change CC behaviour (#3721) (6b85ca8)

10.0.0-alpha.55 (2026-08-24)

Bug Fixes

  • hooks: rewrite inert security-scan-aggregator (#3709) (592cc6b)

Miscellaneous

10.0.0-alpha.54 (2026-08-24)

Miscellaneous

  • deps: bump the npm-minor-patch group across 1 directory with 4 updates (#3701) (ae43fbd)

10.0.0-alpha.53 (2026-08-24)

Bug Fixes

  • docs-site: auth.md carries no frontmatter, on spec grounds (#3694) (a9cf3f1)

Miscellaneous

  • deps-dev: bump the npm-minor-patch group in /src/hooks with 3 updates (#3698) (c9373dd)
  • deps-dev: bump vitest in the npm-minor-patch group (#3697) (74085a0)
  • deps: bump the remotion group (#3700) (f0f4e3a)

CI/CD

  • bump the github-actions group with 5 updates (#3702) (5d27ff2)

10.0.0-alpha.52 (2026-08-23)

Miscellaneous

  • git: untrack committed session state, close the anchoring gap (#3692) (5e31893)

10.0.0-alpha.51 (2026-08-23)

Features

  • build: ship a conforming Agent Plugins 1.0.0 manifest (#3685) (c334a8b)

10.0.0-alpha.50 (2026-08-23)

Features

  • agents: multimodal-specialist documented every modality but one (#3680) (46ec55b)

Bug Fixes

  • hooks: a quoted heredoc body is payload, not an operator (#3676) (a9bbdeb)
  • hooks: the analytics liveness guard was watching 3 of 11 writers (#3678) (f6180dd), closes #3665
  • scripts: gh-api took its token through the transport it replaces (#3679) (8708b96), closes #3653

Documentation

10.0.0-alpha.49 (2026-08-22)

Features

  • skills: add paired-probe, a gate on checks that cannot disagree (#3664) (a2bfb67)

Bug Fixes

  • docs: order arrowed flow lanes by step number, not file position (#3671) (e892212)
  • hooks: git-validator reads the branch from the dir CC says the session is in (#3673) (43e9457)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.238) (#3644) (dd02441)

10.0.0-alpha.48 (2026-08-22)

Bug Fixes

  • cc-watch: the adoption board can no longer lose history (#3662) (787ad7a)

Documentation

  • sandbox: pattern excludedCommands + allowUnixSockets (#3659) (deda673)

10.0.0-alpha.47 (2026-08-22)

Features

  • docs-site: claim the CLI now that it is actually published (#3658) (f0713f4)

Bug Fixes

  • ci: playground gate blocked the playground prune (#3661) (ed99c87)
  • cli: npm strips a bin value that starts with ./ (#3656) (028ea3d)

10.0.0-alpha.46 (2026-08-22)

Features

10.0.0-alpha.45 (2026-08-21)

Features

Bug Fixes

  • pricing: Sonnet 5 is $2/$10, and retire retired model IDs (#3647) (d16f33a)

Miscellaneous

  • cc: snapshot 2.1.238, hand-triage, advance latest_known (#3636) (5a21e79)

10.0.0-alpha.44 (2026-08-21)

Features

  • cc-watch: watch the Claude Platform release notes (#3627) (#3635) (d273c6f)
  • doctor: read the macOS sandbox violation log, fail closed (#3631) (3a6dd7b)
  • scripts: add gh-api.mjs, a GitHub API client for gh-broken shells (#3633) (1f3baa9)
  • worktree: adopt .worktreeinclude and pin baseRef plus spawn depth (#3582) (db65336)

Bug Fixes

  • cc-triage: tolerate fenced JSON and resolve the umbrella with state=all (#3580) (0a3d19a)
  • docs-site: regenerate the CC adoption board data in the build (#3581) (ac27a86)
  • hooks: resolve verifier base against origin (#3570) (6d3640a), closes #3569
  • hooks: stamp PostToolUseFailure on failure-handler context exits (#3625) (e05d703)
  • security: make the mutation gate fail closed (#3626) (803808c), closes #3583

10.0.0-alpha.43 (2026-08-19)

Bug Fixes

  • doctor: report the credential guard that actually exists (#3577) (5bf1a0b)
  • hooks: a file operand means stdin is data, not the script (#3574) (a4da29f), closes #3559
  • hooks: size the projected file, not the tool that wrote it (#3560) (c429c02)
  • tests: give every mktemp a template so it honours TMPDIR (#3579) (4fb82d5)

Documentation

  • configure: caveat when excludedCommands is inert (#3573) (4d77e58)
  • glyph: replace the UNLIMITED chat licence with a measured budget (#3576) (2a4206b)

10.0.0-alpha.42 (2026-08-19)

Bug Fixes

  • build: stop a denied scratch file from zeroing the hook subcount (#3567) (21b11c5), closes #3564
  • ci: bound the second apt-get, the one outside .github/ (#3561) (f2c7866), closes #3557
  • hooks: scope the context7 cap to the session it is named after (#3568) (dd96b2c), closes #3542

Documentation

  • configure: add good first issues link to issue template (#3555) (37fd4d4), closes #3506

10.0.0-alpha.41 (2026-08-19)

Features

  • ork-codex: ship context7 as a plugin-native MCP server (#3549) (cf231c6)

10.0.0-alpha.40 (2026-08-18)

Features

  • agents: grant context7 deliberately, per agent (#3547) (e065b28)

Bug Fixes

  • ci: bound and skip system dep installs (#3550) (4c4c25c), closes #3545
  • tests: read the MCP gate's server list from a tracked manifest (#3546) (36bf3c1)

Documentation

  • configure: add staged operator-scope settings (#3553) (11cde42)
  • mcp: document context7 as a prerequisite, Pro tier, hosted transport (#3548) (f7a1b1a)

10.0.0-alpha.39 (2026-08-18)

Bug Fixes

  • plugin: move shared fragments out of component trees (#3543) (87bc499)

Miscellaneous

10.0.0-alpha.38 (2026-08-18)

Features

  • workflows: default fan-out stages to specialist agentType (#3535) (22e35c4)

Bug Fixes

  • agents: grant the MCP tools 27 agents declare (#3532) (16330df)
  • hooks: close agent-browser URL-coverage hole in safety hook (#3531) (f69f3ea)
  • hooks: reclaim the plugin cache the hook never touched (#3523) (129c286)
  • hooks: stop test fixtures contaminating live spawn telemetry (#3533) (e6b847a)

Miscellaneous

  • dependabot: hold docker majors, keep digest bumps flowing (#3528) (5b59b2b)
  • deps-dev: bump @sanity/client in /orchestkit-demos (#3520) (a710f0a)
  • deps-dev: bump the npm-minor-patch group across 1 directory with 3 updates (#3516) (e0140dd)
  • deps: bump the npm-minor-patch group across 1 directory with 4 updates (#3519) (5a87e53)
  • deps: bump the remotion group across 1 directory with 21 updates (#3540) (b143785)

Documentation

  • skills: correct agent-browser command forms against live 0.34.0 (#3534) (4bf6449)

10.0.0-alpha.37 (2026-08-17)

Miscellaneous

  • deps: bump the remotion group across 1 directory with 21 updates (#3518) (08c815c)

CI/CD

  • probe: route drift to one long-lived issue instead of a summary (#3529) (39efdb4)

10.0.0-alpha.36 (2026-08-17)

Bug Fixes

  • 21st-dev: grant the MCP tools four surfaces already claimed to use (#3509) (c198ee6)
  • docker: pin node 24 to match .nvmrc (#3510) (93e84bf)

Miscellaneous

  • deps-dev: bump @types/node in the npm-minor-patch group (#3514) (79c03ec)
  • deps-dev: bump the npm-minor-patch group across 1 directory with 2 updates (#3515) (52f9908)

Documentation

  • 21st-dev: argue the picker from choice, not from the 2/day cap (#3513) (57150dc)
  • skills: re-sync upstream refs; portless defaults to 443 (#3525) (45895c7)

CI/CD

  • bump the github-actions group across 1 directory with 5 updates (#3521) (2878a31)
  • probe: a failed probe is UNKNOWN, not "every pin is current" (#3527) (a335d3f)

10.0.0-alpha.35 (2026-08-16)

Bug Fixes

Miscellaneous

  • agents: delete the inert hooks: block from 33 agent frontmatters (#3494) (95700b2)
  • git: land the reap-lane branch recovery evidence (#3497) (1a7123f)

Documentation

  • contributing: plugins/ is generated AND tracked (#3498) (c9cf392)

CI/CD

  • playground: exempt fork PRs from the playground gate (#3508) (831ad2f), closes #3502

10.0.0-alpha.34 (2026-08-14)

Bug Fixes

  • docs: publish release-rail to the Lab, drop the render proxy (#3491) (04ea99f)
  • release: real component counts, and a named announce identity (#3489) (7e9e1ed)

10.0.0-alpha.33 (2026-08-14)

Documentation

  • mcp: drop the last ork-elicit leftover from the esbuild header (#3485) (1189c60)

10.0.0-alpha.32 (2026-08-14)

Bug Fixes

  • deps: bump nanoid to 3.3.18 for advisory 1139427 (#3481) (1fac660)
  • hooks: squash-aware worktree verifier, and close a shell injection (#3480) (3a9f776)

10.0.0-alpha.31 (2026-08-13)

Bug Fixes

Miscellaneous

  • cc: adopt 2.1.231, renew override to 2026-11-20 (#3474) (a8bf560)

10.0.0-alpha.30 (2026-08-13)

Bug Fixes

Miscellaneous

  • cc: adopt 2.1.227 through 2.1.229, fix the dead detect stage (#3469) (13214c9)

10.0.0-alpha.29 (2026-08-13)

Code Refactoring

10.0.0-alpha.28 (2026-08-12)

Bug Fixes

  • cc-adoption: adopt MAX_SUBAGENT_SPAWN_DEPTH, fix depth 5/3 mismatch (#3463) (84549cc), closes #3324

Miscellaneous

10.0.0-alpha.27 (2026-08-12)

Bug Fixes

  • hooks: bidirectional contract gate, real PostToolUseFailure field (#3457) (3b701ab), closes #3418
  • hooks: measure context, stop reading fields CC never sends (#3459) (e968377)
  • hooks: settle PostToolUseFailure/PostCompact exceptions (#3458) (c1ec1aa)

10.0.0-alpha.26 (2026-08-12)

Bug Fixes

  • hooks: allow the safe IFS= read idiom (#3454) (92d66af)
  • hooks: stop pre-compact-guard blocking auto-compaction (#3452) (7648c82)

10.0.0-alpha.25 (2026-08-12)

Bug Fixes

10.0.0-alpha.24 (2026-08-11)

Bug Fixes

  • security-tests: fail loudly on a broken payload (#3446) (0ca7660)
  • workflows: rename plugin workflows to .js so they actually register (#3445) (537e778)

10.0.0-alpha.23 (2026-08-11)

Bug Fixes

  • chain-patterns: correct Pattern 8 /loop facts against official docs (#3442) (c884823)
  • hooks: stop git fetch reading as a network source (#3441) (268524b)

10.0.0-alpha.22 (2026-08-11)

Features

  • workflows: adopt CC 2.1.224 capabilities and evidence-replay gate (#3439) (106a03e)

10.0.0-alpha.21 (2026-08-11)

Bug Fixes

  • hooks: stop run-hook.mjs failing silently and crashing (#3436) (e5cc860)

10.0.0-alpha.20 (2026-08-11)

Bug Fixes

  • hooks: stop the pipe-to-shell deny from inventing a fetch (#3434) (4ea34b4)

10.0.0-alpha.19 (2026-08-11)

Bug Fixes

  • agents: drop two frontmatter keys CC provably ignores (#3431) (9842537)

10.0.0-alpha.18 (2026-08-11)

Bug Fixes

  • hooks: match the agent names CC actually sends, drop unread writes (#3428) (61d0661)

10.0.0-alpha.17 (2026-08-11)

Features

  • hooks: adopt autoCompactWindow, delete what could not fire (#3427) (448794e)

Bug Fixes

  • ci: contract gate now names its own blind half on every run (#3421) (5e933e6)
  • hooks: give guard denials a rewrite and a repetition signal (#3423) (7b7f2fa)
  • hooks: read the payload keys CC actually sends (#3425) (f51ad8e)
  • skills: teach both reachability gates the 4th activation path (#3426) (f8c256d)

Miscellaneous

  • cc: advance latest_known to 2.1.227, floor stays frozen (#3419) (44e9d53)

10.0.0-alpha.16 (2026-08-10)

Bug Fixes

  • ci: contract probe looked in the one layout runners lack (#3417) (b4ac670)
  • hooks: branch guard's third path never ran, in 1763 sessions (#3414) (6cca3eb)

10.0.0-alpha.15 (2026-08-10)

Bug Fixes

  • security: scanners knew one worktree layout, not two (#3412) (b187c3a)

10.0.0-alpha.14 (2026-08-10)

Bug Fixes

  • agents: correct 8 false "slash-only" comments (#3405) (b7f8bca)

CI/CD

  • arm the CC contract probe nothing was running (#3407) (e0ac334)
  • tee probe diagnostics to stdout, not just the step summary (#3408) (01d9636)

10.0.0-alpha.13 (2026-08-10)

Bug Fixes

  • hooks: stop logging an invented cache-break token cost (#3403) (4d6c8fa)

10.0.0-alpha.12 (2026-08-10)

Code Refactoring

  • mcp: retire the ork-elicit server for AskUserQuestion (#3400) (e4e08de)

10.0.0-alpha.11 (2026-08-10)

Bug Fixes

  • docs: gate the README prose counts instead of correcting them (#3398) (ae432ba)

10.0.0-alpha.10 (2026-08-10)

Bug Fixes

  • docs: retire the keybindings that never fired (#3395) (cd8d890)
  • hooks: admit bare ctx.log() through the default level gate (#3386) (#3390) (e68b2de)
  • hooks: scope gh enforcers to this repo (#3392) (249abcc)

10.0.0-alpha.9 (2026-08-10)

Bug Fixes

  • hooks: two guards that treated "cannot observe" as "not there" (#3387) (cf6ae25)

Miscellaneous

  • deps-dev: bump @biomejs/biome from 2.5.6 to 2.5.7 in /src/hooks in the npm-minor-patch group (#3379) (386e8c8)
  • deps-dev: bump js-yaml in the npm-minor-patch group (#3378) (a4ba323)
  • deps: bump the npm-minor-patch group (#3382) (2769e88)
  • deps: bump the remotion group in /orchestkit-demos with 21 updates (#3381) (3176706)

CI/CD

  • bump the github-actions group with 6 updates (#3383) (61802f2)

10.0.0-alpha.8 (2026-08-10)

Bug Fixes

10.0.0-alpha.7 (2026-08-09)

Bug Fixes

  • hooks: two hooks read the legacy alias, never the field CC sends (#3372) (909997d)

10.0.0-alpha.6 (2026-08-09)

Bug Fixes

  • hooks: --allow-file-access widened one rule and switched off eight (#3370) (68b8f41), closes #3320

10.0.0-alpha.5 (2026-08-09)

Bug Fixes

  • docs: re-stamp the hook counts in the source, not the build output (#3369) (2d12630)
  • hooks: stop owning worktree provisioning, CC does it better (#3366) (6a26db0)
  • security: close the two actionable code-scanning alerts (#3367) (4110c92)

10.0.0-alpha.4 (2026-08-09)

Bug Fixes

  • hooks: stop teaching the sibling-worktree idiom (#3363) (08fb023)

10.0.0-alpha.3 (2026-08-09)

Bug Fixes

  • cc-triage: make the feature cap announce itself (#3361) (46f2fbc)
  • pre-push: resolve the branch from the pushed refspec, not HEAD (#3299) (1809519), closes #3290

10.0.0-alpha.2 (2026-08-09)

Features

  • hooks: deny credential-file reads at the Read tool (#3360) (eed7bb0)

Bug Fixes

  • agents: scope peer messaging by lifecycle (#3355) (24d5fb6)
  • security: let the DYLD probe observe instead of aborting (#3358) (57a6cfa)
  • settings: retire the plugin settings.json keys CC never reads (#3357) (a93ccb7)

10.0.0-alpha.1 (2026-08-08)

Bug Fixes

  • hooks: advisory context-gate on SubagentStart, arm sandbox (#3345) (7658118)

Miscellaneous

Code Refactoring

  • hooks: trim the bus halves nothing ever called (#3351) (7bf5153)

10.0.0-alpha (2026-08-08)

⚠ BREAKING CHANGES

  • glyph: the skill formerly invoked as /ork:quickviz is now /ork:glyph. There is no alias, because a stub named quickviz would reintroduce the /quit prefix collision this change exists to remove.

Features

Bug Fixes

  • build: keep the version sync off pinned channel entries (#3344) (6c80f65), closes #164 #3333
  • counts: read the release version off the tracking entry by name (#3341) (18fdb71), closes #164 #3333
  • gate: the README command check reads history as advertisement (#3343) (ebd8e3d), closes #164 #3333
  • hooks: stop stale-team-cleanup deleting a live session's task list (#3305) (a0e56f7)
  • release: make version comparison semver-correct across 8 sites (#3339) (d9d9387), closes #164 #3333
  • release: stop the build stamping the pinned channel entry (#3342) (263840f), closes #164
  • security: clear the nanoid and DOMPurify advisories blocking main (#3303) (d4ce3c4)

Miscellaneous

  • cc: triage 2.1.223 through 2.1.226, advance latest_known to 2.1.226 (#3338) (c80783c), closes #3334

Code Refactoring

  • glyph: rename quickviz, its prefix collided with /quit (#3301) (bdce685)

9.8.0 (2026-08-07)

Features

  • dream: STEP 9 cross-repo promotion candidates (#3296) (b6e6524)

Bug Fixes

  • hooks: read the permission mode key CC actually sends (#3294) (6b14358)
  • pre-push: replace mapfile so the hook runs on macOS bash 3.2 (#3287) (5c98332)
  • release: let release-please own the lockfile version (#3293) (e9c217d)
  • repair two reported bugs and make 26 blind security tests able to fail (#3288) (1f4cf61)
  • security: close the interpreter-pipe path bypass (#3292) (a64613c)
  • security: close the line-continuation bypass of the command blocker (#3289) (9839bf8)
  • verify: stop waiting forever on a background task that never ran (#3297) (6fd73ff)

Miscellaneous

  • cc: adopt 2.1.221 and 2.1.222, bump cc-support latest_known (#3284) (a28e2b3)
  • community: add code of conduct and user-facing issue templates (#3300) (e0edf17)

9.7.0 (2026-08-05)

Features

  • visualize-plan: make DASHBOARD a first-class archetype (#3275) (2081957)

Bug Fixes

  • hooks: validate agent spawn targets against the real registry (#3280) (de5ed6b)
  • security: audit every tracked lockfile, not four of six (#3274) (607c88c)

Miscellaneous

  • worktree: fix a silent no-op and remove the dead worktree CLI (#3282) (c5b2d28)
  • worktree: share one node_modules across worktrees (#3281) (e2c788a)

9.6.1 (2026-08-04)

Bug Fixes

  • deps: clear the npm-audit wave via in-range transitive updates (#3272) (4f76215)

9.6.0 (2026-08-03)

Features

9.5.4 (2026-08-03)

Bug Fixes

  • ci: run the demos' 284 tests, stop chmod churn on .mjs (#3262) (c0adbee)

Miscellaneous

  • deps-dev: bump jsdom from 29.1.1 to 30.0.1 in /orchestkit-demos (#3259) (9204825)
  • deps-dev: bump the npm-minor-patch group across 1 directory with 3 updates (#3258) (bf71ee5)
  • deps-dev: bump the npm-minor-patch group in /src/hooks with 2 updates (#3254) (8c3d018)
  • deps: bump the npm-minor-patch group in /src/mcp-server with 2 updates (#3255) (0c73383)
  • deps: bump the remotion group (#3257) (2782ff3)

CI/CD

  • bump the github-actions group with 5 updates (#3260) (6b216b8)

9.5.3 (2026-08-03)

Miscellaneous

  • deps-dev: bump @types/node in the npm-minor-patch group (#3253) (bc61e37)

Documentation

  • contributing: add entry-point section and fork-CI expectation note (#3250) (747321e)
  • fix nonexistent scripts/hooks path in SECURITY.md (#3248) (c7855ba)

9.5.2 (2026-08-02)

Bug Fixes

  • hooks: branch-switch awareness in git-validator protection (#3246) (9ef10e9)

CI/CD

  • glob the five orphan test dirs into blocking coverage (#3245) (79f9479)

9.5.1 (2026-08-02)

Bug Fixes

  • skills: never block a user-typed invocation (#3243) (a9f7ea3)

9.5.0 (2026-08-01)

Features

  • hooks: nudge TaskCreate on 3+ step work and measure compliance (#3232) (ec1ca84)

Bug Fixes

  • docs-site: fix count and command drift at source, add a drift gate (#3231) (0ed071d)
  • skills: keep user-typed fork skills interactive on CC 2.1.218+ (#3239) (a47d763)
  • tests: dead E2E wiring, gitignore-aware gates, warn ratchet (#3236) (a4535bc)

Miscellaneous

  • evals: retire the prose-grading eval, wire the nudge reader (#3237) (80b0e26)

Documentation

9.4.0 (2026-08-01)

Features

  • quality: add the python half of the phantom-symbol gate (#3230) (63bf412)
  • quality: skill-delta ratchet gate (campaign phase 0) (#3211) (0e6a3a8)

Bug Fixes

  • agents: clear all 94 dead skill preload declarations (#3221) (8e6213b)
  • ci: accept the app/ author form and stop swallowing gh failures (#3227) (e08ee88)
  • ci: regenerate CHANGELOG-derived files on release PRs (#3226) (113b67a)
  • drift: schedule the currency probe and kill two phantom APIs (#3222) (e267c5c)
  • runner: honour the interpreter and reclaim 56 orphaned tests (#3228) (95836b4)
  • skills: bound mcp below v2 and teach the gate about ceilings (#3229) (33db0ef)
  • skills: drop impossible floor-CC path from dataviz gate (#3213) (11e1dad)
  • skills: repair 32 dead skill preload edges (#3216) (e2172dd)
  • skills: retire three stale JsonRenderComposition references (#3225) (240a242)
  • tests: ratchet orphaned tests and repair the one that failed (#3223) (e9e74e3)
  • tests: reachability linter was blind to block-list YAML (#3219) (4a88597)

Miscellaneous

  • ci: bump internal CI Node pins from EOL Node 20 to 22 (#3214) (bc0eced)
  • regen v9.3.0 build output and widen the drift gate (#3217) (acfbc4f)

Documentation

  • add SECURITY.md and a commercial surface to the README (#3209) (75eb781)

9.3.0 (2026-07-31)

Features

  • agents: key Tavily guidance on the tvly CLI rail (#3204) (7fd6ca2)
  • skills: fix 5 shipped bugs and add 4 missing CI gates (#3181) (d4422c3)

Bug Fixes

  • ci: decide the playground gate from the diff, not the branch (#3196) (33969f5)
  • hooks: stop output-validator killing five sibling SubagentStop hooks (#3201) (c5bea1e)

Documentation

  • marketplace: prep official plugin directory submission (#3205) (d1297fb)

CI/CD

  • demos: report Validate Renderer to the merge queue (#3199) (356e9be)

9.2.1 (2026-07-28)

Bug Fixes

  • evals: propagate INCONCLUSIVE through the --changed fan-out (#3188) (5d91f2c)
  • quickviz: ship visual-style.md inside the skill bundle (#3185) (db038a1)

CI/CD

  • bump the github-actions group across 1 directory with 2 updates (#3174) (ea01f8e)

9.2.0 (2026-07-28)

Features

  • docs: derive a flow graph for every skill reference page (#3184) (060ab5e)

Bug Fixes

  • hooks: skip ASK tiers in bypassPermissions mode (#3177) (f2c3d38)
  • visual-style: stop the emoji vocabulary rejecting the ASCII palette (#3169) (a5dd7fb)

Miscellaneous

  • deps: bump the npm-minor-patch group across 1 directory with 4 updates (#3179) (5c8169a)
  • deps: bump the remotion group across 1 directory with 21 updates (#3175) (728a7e7)

Documentation

  • readme: rebuild the community table and drop two dead sections (#3183) (1b06de9)

CI/CD

  • gate README command references, sweep dead links weekly (#3187) (cf961b1)

9.1.1 (2026-07-26)

Documentation

  • sync README and changelog data to v9.1.0 (#3167) (5694f4b)

9.1.0 (2026-07-26)

Features

  • retro-hardening mechanisms 5+6 (invocability check, byte-budget guard) (#3161) (c5ac1aa)
  • viz: re-anchor visual-style rule + quickviz on-ramp (#3164) (9ff788e)

Bug Fixes

  • viz: stop throttling explicit visual asks (#3166) (dde0e87)

9.0.2 (2026-07-26)

Miscellaneous

  • ci: remove the Claude PR Review workflow and its labeler (#3159) (6223023)

9.0.1 (2026-07-26)

Bug Fixes

  • bound rotated analytics archives, unstale the CC floor (#3156) (bf2bd9d)
  • build: default-deny frontmatter gate, pass all 10 fields (#3158) (4e02e1c)
  • readme: refresh What's New for 9.0.0 and isolate the RTL table cell (#3153) (6dbcb9a)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.220) (#3155) (b56a1fd)

9.0.0 (2026-07-25)

⚠ BREAKING CHANGES

  • cc: the Claude Code support floor is now 2.1.220. Claude Code 2.1.206 through 2.1.219 are no longer supported; hooks and skills may no-op or error on them, and that is not treated as a bug. Upgrade Claude Code before updating OrchestKit.

Features

  • cc: adopt Opus 5, strict-renew the support floor to 2.1.220 (#3141) (47b78fb)
  • verify: add the Reachability Proof axis (REACHED vs UNREACHED) (#3149) (5fc5031)

Bug Fixes

  • docs: pin playground links to the commit SHA, not the branch (#3148) (30d575f)
  • hooks: bound hook-timing.jsonl, the one analytics file with no cap (#3151) (d389c37)

Miscellaneous

8.85.0 (2026-07-25)

Features

Bug Fixes

  • 3 defects found by /ork:assess on the M170 diff (#3136) (b8be78c)
  • build: stop dropping argument-hint from generated commands (#3146) (be63597)
  • hooks: make network-egress-guard DENY tier quote-aware (#3124) (357e404)
  • hooks: redirect pipe-to-interpreter deny instead of dead-ending (#3121) (c72bf74)
  • LangChain ecosystem currency + dual-registry version watchdog (#3142) (de60c1b)
  • narrow skill-eval pathspec + close ASK-tier quote gap (#3138) (8355412)
  • skills: remove 24 phantom imports found by the import gate (#3144) (cba9633)
  • skills: remove phantom-API leftovers the import gate cannot see (#3147) (774ac7b)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.218) (#3120) (2414aaf)
  • cc-watch: snapshot upstream CHANGELOG (2.1.220) (#3140) (5dc0175)
  • deps-dev: bump @biomejs/biome from 2.5.4 to 2.5.5 in /src/hooks in the npm-minor-patch group across 1 directory (#3107) (dc50099)
  • deps: bump postcss from 8.5.16 to 8.5.23 in /orchestkit-demos (#3143) (7edc78e)

8.84.10 (2026-07-23)

Bug Fixes

  • ci: drop the What's New freshness gate blocking the PR queue (#3118) (c9308d0)

Miscellaneous

  • deps: bump the npm-minor-patch group across 1 directory with 4 updates (#3111) (5efd7e6)
  • deps: bump the remotion group across 1 directory with 21 updates (#3110) (7278d19)

CI/CD

  • bump the github-actions group across 1 directory with 6 updates (#3112) (6e83f3a)

8.84.9 (2026-07-23)

Bug Fixes

  • ci: scope What's New --check to PRs, refresh for 8.84.8 (#3113) (b14c62d)

CI/CD

  • review: allow the release bot to trigger a review (#3114) (bbff822)
  • review: skip auto-label on Dependabot PRs (no secret access) (#3116) (8707f66)

8.84.8 (2026-07-23)

Bug Fixes

  • hooks: stop denying local data piped to an interpreter (#3097) (4c9b973)
  • security: remove privesc in release regen, tighten allowlist (#3102) (7a11882)

CI/CD

  • demos: make renderer check gate-able, split remotion bumps (#3105) (aba09d4)
  • review: auto-run Claude review on sensitive PRs, ban dead gates (#3104) (5900e29)

8.84.7 (2026-07-23)

Bug Fixes

  • demos: repair the Remotion renderer, validate it pre-merge (#3099) (1447afa)

8.84.6 (2026-07-23)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.218) (#3094) (bbd760e)

CI/CD

  • readme: auto-generate What's New from CHANGELOG (#3096) (667aebe)

8.84.5 (2026-07-22)

Bug Fixes

  • ci: scope Claude workflow tools, gate triage on author (#3085) (2835d07)
  • security: audit all four trees in CI, pin sharp (#3090) (b23e208)
  • security: correct audit-gate id reporting, allowlist hono (#3087) (cd19fee)

8.84.4 (2026-07-22)

Miscellaneous

  • skills: burn model-recency ratchet to zero (#3056) (f0f9557)

8.84.3 (2026-07-22)

Miscellaneous

  • hook-contract: bump to 0.1.2, domain-anchored SDK provenance (#3080) (dd17002)

CI/CD

  • pin npm to 11.x in the hook-contract publish job (#3082) (0aad2e7)

8.84.2 (2026-07-22)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.217) (#3070) (565e5e7)
  • deps: bump body-parser from 2.2.2 to 2.3.0 in /src/mcp-server (#3065) (c7dba92)
  • deps: bump fast-uri from 3.1.2 to 3.1.4 in /orchestkit-demos (#3076) (82b63e3)
  • deps: bump fast-uri from 3.1.2 to 3.1.4 in /src/mcp-server (#3074) (c4d79ee)
  • deps: bump hono from 4.12.25 to 4.12.31 in /src/mcp-server (#3075) (2f57247)

CI/CD

  • make nightly health failures diagnosable (#3072) (d3b2901)
  • validate pinned Claude model IDs at commit time (#3077) (2e850e3)

8.84.1 (2026-07-21)

CI/CD

  • fix playground gate messaging and release-PR stranding (#3063) (95d8e8d)

8.84.0 (2026-07-21)

Features

  • evals: honest verdicts, three new gates, and a budget governor (a1490b0)

Bug Fixes

  • dream: stop auto-pruning live memories with out-of-repo refs (1df5cd4)
  • evals: score effective recall over the full positive population (#3049) (e821cbf)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.216) (#3053) (4a671c9)

Documentation

  • add AGENTS.md for AI coding agent discoverability (#3059) (df25757)

8.83.1 (2026-07-20)

Bug Fixes

  • evals: report an eval outage as inconclusive, not as a skill regression (#3047) (105cd62)

8.83.0 (2026-07-20)

Features

  • cmux: add declarative team layout for cmux (#3045) (1eb838b)
  • cover,hooks,auto: close four AI-developer-workflow gaps (#3044) (f644290)

8.82.1 (2026-07-20)

Bug Fixes

  • evals: refuse to grade dead generations (#3043) (6c40c80)
  • hooks: mark phantom stops, recover model, drop dead fields (#3041) (7fcdce8)

Miscellaneous

Documentation

  • reconcile every surface this session made stale (#3040) (5b6239a)

8.82.0 (2026-07-20)

Features

  • playgrounds: living-plan archetype (LPP) (#3019) (cb03e49)

Bug Fixes

  • analytics: stop advertising data the file cannot provide (#3037) (96bc00c)

Miscellaneous

  • deps: bump the npm-minor-patch group across 1 directory with 21 updates (#3030) (927229c)

CI/CD

  • bump the github-actions group across 1 directory with 2 updates (#3031) (e062c96)

8.81.0 (2026-07-20)

Features

  • hooks: analytics dead-writer liveness guard + routing_edge view (#3026) (ac3d7ba)

8.80.3 (2026-07-20)

Bug Fixes

  • hooks: latch the session-title /rename guard so it can't leak (#3025) (b2f075f)
  • security: resolve 5 real CodeQL findings, dismiss 4 as noise (#3022) (ddc4759)

8.80.2 (2026-07-19)

Bug Fixes

  • hooks: here-string guard escalates to ASK instead of DENY (#3021) (69fbfcb)
  • hooks: stop flagging quoted-heredoc bodies as shell attacks (#3015) (6747007)

CI/CD

  • make the gates real, and stop the pipe guard blocking quoted text (#3018) (eb3e4e9)
  • pin actions in skill-claims-audit to SHA (#3020) (f7349ec)

8.80.1 (2026-07-19)

Miscellaneous

  • deps-dev: bump @biomejs/biome from 2.5.3 to 2.5.4 in /src/hooks in the npm-minor-patch group (#3006) (89057c4)
  • deps: bump the npm-minor-patch group (#3009) (deebb70)

CI/CD

  • bump the github-actions group with 7 updates (#3010) (def4f72)
  • skip autoheal on dependabot PRs instead of failing (#3013) (ade336b)

8.80.0 (2026-07-19)

Features

  • docs-site: scroll-world ride homepage + yonyon rebrand (d348761)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.215) (#3004) (6f0747a)
  • remove agentation and the never-published release channels (b23da59)

8.79.0 (2026-07-18)

Features

  • cc-watch: evidence-gated filing lanes - precision gate + sub-floor recall (#2992, #2993) (#2995) (efee8f1)
  • cc: adopt CC 2.1.209-214, advance latest_known to 2.1.214 (#2994) (04b6906)
  • evals: eval coverage walker + starter-spec fill (#2192) (#2998) (523bd02)
  • evals: quality index aggregator + real regression gate (#2194) (#2999) (fb98c98)
  • hooks: regenerate session title on direction change (#2974) (a1dee8b)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.214) (#2989) (a951b80)
  • hygiene: drift fixes - claims gate, islands gate, hooks.json (#2996) (01c5dbd)

8.78.0 (2026-07-17)

Features

  • docs-site: reference root tab + sidebar IA repair (#2963) (a001df0)

Bug Fixes

  • audit: correct silently-wrong line counts in audit scripts (#2966) (017841a)
  • counts: correct the CI line gate and the complexity LOC scan (#2973) (740dd9c)
  • hooks: consolidate branch resolution, drop the last two git spawns (122175e), closes #2970
  • hooks: kill the sync git spawn behind webhook-forwarder timeouts (2e941e3)
  • hooks: stop display-lint and pipe guard misreading shell grammar (2b95f3d)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.212) (#2946) (0210a37)

8.77.1 (2026-07-16)

Bug Fixes

  • hooks: close two display-lint exemption bypasses (#2932) (fe8b8c2)

8.77.0 (2026-07-16)

Features

  • hooks: display-lint gate, de-clutter Bash transcript at the source (#2930) (86f1b56)

8.76.0 (2026-07-16)

Features

  • hooks: fired-census + fix a 4.5-month telemetry blind spot (#2913) (958b897)
  • telemetry-inspect: consume the fired-census (#2915) (7e40ffd)

Bug Fixes

  • cc-triage: raise opus extraction timeout to 180s (was 60s) (#2922) (8885c8e)
  • hooks: session-scope shared state, kill cross-session bleed (#2920) (e21b720), closes #2919

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.211) (#2918) (38f5155)
  • cc-watch: snapshot upstream CHANGELOG (2.1.211) (#2921) (56306f9)
  • cc-watch: snapshot upstream CHANGELOG (2.1.211) (#2929) (77d9fb1)

8.75.0 (2026-07-15)

Features

Bug Fixes

  • ci: unbreak main tails - skill counts, eval content, file guard (#2911) (dd75fc1)
  • hooks: unbreak main — 3 red checks from the P3-5 merge (#2910) (a20eac1)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.210) (#2907) (6190c5f)
  • ci: bump remaining CC pins from 2.1.148 to floor 2.1.206 (#2912) (2d22a3a)

8.74.0 (2026-07-14)

Features

  • hooks: bury all 51 dead hooks — registry closure now exact (#2889) (51c4d21)
  • hooks: CI gate for hook-registry reachability closure (#2887) (0859f22)

Bug Fixes

  • hooks: team-size-gate never fired — matcher is Agent, not Task (#2886) (546613d)

8.73.0 (2026-07-14)

Features

8.72.0 (2026-07-14)

Features

Bug Fixes

8.71.1 (2026-07-14)

Bug Fixes

  • vocab: un-suspend Fable 5 — GA, consent-gated (not suspended) (#2879) (ee91a33)

8.71.0 (2026-07-14)

Features

  • loop-gates: harden streak + cross-model gates, review holdout labels (M161) (#2873) (a470f19)

Miscellaneous

  • hooks: ORK_NO_NOTIFY opt-out + playground-warner opt-in-by-detection (#2876) (f5ef746)

8.70.3 (2026-07-14)

Miscellaneous

  • cc: reconcile 7 auto-filed 2.1.207-208 adoption issues (#2871) (ab3282d)

8.70.2 (2026-07-14)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.208) (#2867) (9d8942a)
  • cc: advance latest_known 2.1.206 → 2.1.208 (soft, floor frozen) (#2870) (90996b5)
  • deps-dev: adopt TypeScript 7.0.2 across remaining 4 trees (#2869) (9266f8a)
  • deps-dev: bump @types/node in the npm-minor-patch group (#2853) (df87ef1)
  • deps-dev: bump the npm-minor-patch group in /src/hooks with 5 updates (#2855) (795deb6)
  • deps-dev: bump typescript from 6.0.3 to 7.0.2 in /src/hooks (#2856) (a57d266)
  • deps: bump the npm-minor-patch group (#2860) (cba8842)

8.70.1 (2026-07-13)

Bug Fixes

  • quote argument-hint YAML values so Copilot CLI ≥1.0.65 loads all skills (closes #2863) (#2864) (4d7d2d0)

Miscellaneous

CI/CD

  • add argument-hint quoting lint, skip fork PR report comment (#2865) (fc8c8ed)

8.70.0 (2026-07-12)

Features

  • auto: reframe /ork:auto as the default front door (#2849) (4d4c0e8)

Bug Fixes

  • hooks: restore skill-usage analytics + agent-type attribution at SubagentStop (#2850) (f2e0159)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.207) (#2848) (a7d9678)
  • untrack 7.5MB build artifacts + local state, remove dead playgrounds/ (#2845) (d2c724c)

Documentation

  • add READMEs for the 4 convention-anchor dirs (#2847) (154f9fb)

8.69.1 (2026-07-11)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.207) (#2839) (1031ff3)
  • de-bloat logo + add playground-drift recurrence guard (#2844) (8abd7d0)
  • prune 418 stale PR-playground dirs from main, add recurrence fix (#2841) (a17c50b)
  • repo hygiene sweep — declutter root, untrack generated coverage (#2840) (5c3acc7)

8.69.0 (2026-07-10)

Features

  • cc: bump support floor 2.1.183 → 2.1.206 (renew strict pin) (#2830) (10a2d30)

Bug Fixes

  • hooks: re-architect InstructionsLoaded to real single-file payload (#2832) (7061fcb), closes #2475

8.68.3 (2026-07-10)

Bug Fixes

  • bug-triage sweep — eval-runner grounding, telemetry dark-check, orphaned stager, autobuild re-trigger (#2828) (8426e45)

8.68.2 (2026-07-10)

Bug Fixes

Miscellaneous

  • 1p: genericize committed op:// refs in public repo (#2814) (d70a23e)
  • cc-watch: snapshot upstream CHANGELOG (2.1.205) (#2812) (2b062f3)
  • cc-watch: snapshot upstream CHANGELOG (2.1.206) (#2825) (9a7f501)
  • cc: adopt 2.1.205-206 (latest_known 2.1.204→2.1.206) (#2826) (8cb6884)
  • ci: move private-index URL into one secret (recon hardening) (#2815) (11770e9)

8.68.1 (2026-07-08)

Bug Fixes

  • agent-readiness: orank gap triage - block scrape bots, fix PyPI homepage (#2797) (b0526a6)
  • hooks: telemetry event field falls back to '' instead of 'unknown' (#2800) (81f8563), closes #2799

8.68.0 (2026-07-08)

Features

  • skills: adopt MengTo/Skills quality-bar + stylecard patterns (#2794) (f1104d4)

8.67.0 (2026-07-08)

Features

Documentation

  • skills: document fable symbolic model name in --model plumbing (#2791) (8dba6fb)

8.66.0 (2026-07-08)

Features

  • cc-adoption: adopt CC 2.1.202 (2669c62)

Bug Fixes

  • emulate-seed: mention emulate 0.9.0 (Nuxt adapter) in skill body (#2778) (347b2a5)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.202) (#2777) (79761bb)
  • cc-watch: snapshot upstream CHANGELOG (2.1.204) (#2789) (37cb450)
  • deps-dev: bump typescript in /orchestkit-demos (#2768) (89b5072)

CI/CD

  • auth: move 4 Claude workflows to OAuth, throttle sentinel to daily (#2790) (81f449a)
  • release-announce: growth channels, no LinkedIn (#2781) (2f6a1d1)
  • release-announce: handle 202 queued response (#2782) (59d67e6)

8.65.2 (2026-07-06)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.201) (#2771) (95e468e)
  • deps-dev: bump the npm-minor-patch group across 1 directory with 3 updates (#2774) (0b9b330)
  • deps-dev: bump tsx (#2766) (ee8f29e)
  • deps: bump dotenv from 16.6.1 to 17.4.2 in /orchestkit-demos (#2767) (32d5355)
  • labs: bump upstream-version-tested pins for drifted skills (#2772) (2cdf179)

CI/CD

  • bump the github-actions group with 5 updates (#2770) (1324ad6)

8.65.1 (2026-07-06)

Miscellaneous

8.65.0 (2026-07-05)

Features

  • cc-adoption: adopt CC 2.1.198-2.1.201 (#2756) (6416db9)
  • fix-issue: add Phase 4-5 blast-radius clarification gate (#2758) (2347e64)
  • implement: add Step 0b blast-radius clarification interview (#2754) (5227e4f)
  • verify: add Verification Manifest (VERIFIED vs CLAIMED) gate (#2753) (70ad7f5)

Bug Fixes

  • cc-triage: graduate thin product announcements token-free (#2760) (0bd5c1d)
  • skills: heal labs skill version drift + close CI gap (#2759) (541d5cc)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.199) (#2743) (9adf6dc)
  • cc-watch: snapshot upstream CHANGELOG (2.1.201) (#2750) (d2677cc)
  • cc-watch: snapshot upstream CHANGELOG (2.1.201) (#2752) (3eadc90)
  • gitignore session-state artifacts (#2762) (5e155f1)

8.64.1 (2026-07-02)

Bug Fixes

  • docs: heal stale CC floor + fact rot, add docs-facts freshness gate (#2736) (1379b8b)

8.64.0 (2026-07-02)

Features

  • auto: tune routing rules from measured LLM misses + eval harness (#2731) (878c7a5)
  • hooks: advisor reads configured default model for inherit (#2710) (716ffd1)
  • skills: audit-driven quality + eval hardening (#2715) (dfb4c53)
  • skills: visualize-plan rents CC /dataviz for validated chart marks (#2732) (dd2299c)

Bug Fixes

  • ci: release-please guard fetches full base history (#2711) (356fb66)
  • docs-site: prebuilt search index cuts NLWeb /ask cold-start (#2719) (7059bc8)
  • docs: stamp README CC support floor (≥2.1.148 → ≥2.1.183) (#2717) (42b372c)
  • git-hooks: pre-push mirrors dependabot/release-please skip (#2733) (9e70423)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.195) (#2688) (d6dbadc)
  • cc-watch: snapshot upstream CHANGELOG (2.1.196) (#2708) (2dafe24)
  • cc-watch: snapshot upstream CHANGELOG (2.1.197) (#2713) (77d81ed)
  • cc-watch: snapshot upstream CHANGELOG (2.1.198) (#2730) (5e821f2)
  • cc: adopt Claude Sonnet 5 + CC 2.1.197 (#2709) (4c9f6ef)
  • deps-dev: bump @sanity/client in /orchestkit-demos (#2696) (9de8861)
  • deps-dev: bump @types/node (#2689) (4aa5fad)
  • deps-dev: bump @types/node in /orchestkit-demos (#2698) (55c1e42)
  • deps-dev: bump eslint from 9.39.4 to 10.6.0 in /orchestkit-demos (#2697) (7426122)
  • deps-dev: bump jsdom from 27.4.0 to 29.1.1 in /orchestkit-demos (#2695) (9d2314f)
  • deps-dev: bump the npm-minor-patch group across 1 directory with 2 updates (#2691) (d2c6985)
  • deps-dev: bump the npm-minor-patch group with 2 updates (#2690) (5f5b977)
  • deps: bump the npm-minor-patch group across 1 directory with 33 updates (#2734) (d96f41c)
  • labs: weekly upstream pin sync (#2699) (2c45312)
  • repo: stamp GitHub About + untrack overcut/ orphan (#2718) (6fe0b2c)

CI/CD

  • bump the github-actions group across 1 directory with 3 updates (#2693) (bd0ea33)

8.63.2 (2026-06-27)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.195) (#2686) (754c986)
  • release: exclude docs/ from plugin version bumps (#2687) (b4d1f74)

8.63.1 (2026-06-26)

Documentation

  • chain-patterns: add Workflow parse-safety section (the 3 killers) (#2679) (1ae5d75)

8.63.0 (2026-06-26)

Features

  • ci: draft release announcements via HQ platform marketing (#2677) (d3ba41e)

Bug Fixes

  • site: make OpenAPI/NLWeb discoverable at probed paths (orank) (#2676) (4bb40d8)

8.62.1 (2026-06-26)

Bug Fixes

  • mcp: resolve memory KG path at runtime so the graph persists (#2674) (d7322be), closes #2631

8.62.0 (2026-06-26)

Features

  • adoption: 3 CC security/hygiene wins + triage sweep of 17 tickets (#2661) (f102bd6)
  • cc-support: bump latest_known 2.1.183 to 2.1.191 (#2663) (36284ee)
  • cc-support: bump latest_known 2.1.183 to 2.1.193 (#2673) (147726a)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.193) (#2671) (6fd3b63)

8.61.0 (2026-06-25)

Features

  • agents: wire 7 under-activated agents/skills (P2) (#2644) (f2bf2d2)

Bug Fixes

  • hooks: pre-commit runner treats timeouts as skips, not failures (#2658) (456f582)
  • skills: salvage standalone fixes from superseded #2643 (#2657) (e169b2e)

8.60.1 (2026-06-25)

Bug Fixes

  • untrack node_modules symlink + rewire activation telemetry (#2653) (ac3a86b)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.191) (#2652) (623b726)

8.60.0 (2026-06-24)

Features

  • skills: add /ork:auto intent router (+ telemetry breaker fix) (#2645) (252ce9d)

8.59.0 (2026-06-24)

Features

  • agents: ground 14 specialists + wire monitoring-engineer (#2641) (67cad0c)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.187) (#2640) (7efb950)

8.58.0 (2026-06-23)

Features

  • cc-watch: add precision+recall verify gate to cc-triage (#2626) (2b20c3f)
  • doctor: complete MEMORY.md budget check (line cap + path fix) (#2623) (bd01b2f)
  • hooks: advisor asks to redirect generic spawns to specialists (#2632) (3998bac)
  • implement: wire event-driven-architect into Phase 4 spawn block (#2630) (a4bee99)
  • skills: add audit-activation (agent-side activation audit) (#2628) (701609e)

Documentation

  • cc-watch: release-integration SOP + cc-support SoT schema (#2627) (592533c)

8.57.0 (2026-06-23)

Features

  • doctor: add auto-memory MEMORY.md index budget check (#2610) (bcb5820)

Bug Fixes

  • counts: stamp pyproject description so counts can't drift (#2604) (28cbbd4)
  • labs: refresh stale Vercel Labs docs + extend watcher TARGETS (#2605) (d5c47a3)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.186) (#2622) (79ca903)
  • counts: retire update-counts.sh + enforce stamped counts in CI (#2606) (3e85b90)
  • labs: bump upstream-version-tested pins for drifted skills (#2594) (24a6208)
  • rules: remove dead recent-decisions.md from auto-loaded context (#2609) (4e3a4ca)

Documentation

  • ci-debug: add pattern #11 (vercel status orphaned on path-skip) (#2595) (544a022)

Code Refactoring

  • agents: defer Labs versions to the skill, add guard (#2608) (4617e5f), closes #2600

8.56.0 (2026-06-22)

Features

  • hooks: nudge general-purpose test-run spawns toward Bash (#2576) (ed73464)
  • hooks: reliable session auto-rename + content-emoji color (#2593) (2e4da25)

Bug Fixes

  • skill-evolution: store per-entry decisions outside .claude/rules/ (#2592) (9facc9a)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.185) (#2587) (5087b3f)
  • deps-dev: bump @types/node from 25.9.3 to 26.0.0 (#2579) (7586104)
  • deps-dev: bump @types/node from 25.9.3 to 26.0.0 in /docs/site (#2582) (3a8a399)
  • deps-dev: bump @types/node from 25.9.3 to 26.0.0 in /src/hooks (#2580) (604454c)
  • deps-dev: bump @types/node in /src/mcp-server (#2578) (fdb84a8)
  • deps-dev: bump @vitejs/plugin-react in /docs/site (#2584) (43565fd)
  • deps-dev: bump js-yaml from 4.2.0 to 5.0.0 (#2577) (99398c9)
  • deps: bump @dagrejs/dagre from 2.0.4 to 3.0.0 in /docs/site (#2585) (f00d65a)
  • deps: bump @vercel/speed-insights in /docs/site (#2583) (a453ccb)
  • deps: bump the npm-minor-patch group in /docs/site with 4 updates (#2581) (36d5d4d)

CI/CD

  • bump the github-actions group across 1 directory with 2 updates (#2586) (2db8005)

8.55.2 (2026-06-21)

CI/CD

  • cc-triage: exclude presentational rewords from the now-<verb>s hint (#2570) (745f9b9)

8.55.1 (2026-06-21)

CI/CD

  • cc-watch: move release-watch cron to 06:00 IL (03:00 UTC) (#2566) (512dd62)

8.55.0 (2026-06-20)

Features

  • skill-evolution: make holdout gate executable for assess (#2562) (7031857)

Bug Fixes

  • hooks: stop mcp-health-check false-positive on op-read keys (#2563) (e345c79)

8.54.0 (2026-06-20)

Features

  • cc: modernize to 2.1.183 (drop TeamCreate/Task, bump floor) (#2559) (5700399)

Bug Fixes

  • hooks: CCR reads tool_response, the field CC actually sends (#2552) (#2553) (e3ae43d)
  • skills: soften overclaims and fix streak goal race (#2554) (d05782f)

CI/CD

  • security: resolve all 109 zizmor code-scanning alerts (#2560) (0f1694b)

8.53.0 (2026-06-20)

Features

Bug Fixes

  • hooks: route CLAUDE_PLUGIN_ROOT off the user-facing banner (#2543) (848da6b)

CI/CD

8.52.0 (2026-06-20)

Features

8.51.0 (2026-06-19)

Features

  • adopt CC 2.1.183 (git/IaC guard parity + capability docs) (#2529) (f02bb72)

Bug Fixes

  • agents: add worktree isolation to 2 frontend Bash+Write agents (#2530) (5905f9b)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.179) (#2526) (995f309)

CI/CD

8.50.0 (2026-06-18)

Features

  • playground: release-notes player + execution-router decision board (visualize-plan) (#2518) (435dfd9)

Miscellaneous

  • triage + close adopted/N-A CC 2.1.178-179 backlog issues (#2525) (3514449)

8.49.0 (2026-06-18)

Features

  • adopt CC 2.1.181 (config, fg depth-cap, AskUserQuestion) (#2523) (6a672af)
  • playground: Arieh-grade visual standard + decision boards (#2516) (e386dd2)
  • playground: CC release-notes generator + adoption playgrounds (#2522) (5639c01)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.179) (#2515) (2e5b227)

8.48.2 (2026-06-17)

Bug Fixes

Miscellaneous

  • cc: adopt CC 2.1.179 (latest_known 2.1.176 → 2.1.179) (#2510) (8fe436d)
  • deps-dev: bump @types/node from 24.12.4 to 25.9.3 (#2497) (95709a6)
  • deps-dev: bump @types/node from 24.12.4 to 25.9.3 in /docs/site (#2507) (65c007a)
  • deps-dev: bump @types/node from 24.12.4 to 25.9.3 in /src/hooks (#2500) (b807ba9)
  • deps-dev: bump @types/node in /src/mcp-server (#2496) (83b79ad)
  • deps-dev: bump agentation from 2.3.3 to 3.0.2 in /docs/site (#2504) (c8972cd)
  • deps-dev: bump the npm-minor-patch group across 1 directory with 4 updates (#2499) (6ca0035)
  • deps-dev: bump typescript from 5.9.3 to 6.0.3 (#2498) (d2e39ef)
  • deps-dev: bump typescript from 5.9.3 to 6.0.3 in /docs/site (#2506) (348cfb1)
  • deps-dev: bump typescript from 5.9.3 to 6.0.3 in /src/hooks (#2501) (1f7a170)
  • deps-dev: bump typescript from 5.9.3 to 6.0.3 in /src/mcp-server (#2494) (ef868f4)
  • deps-dev: bump vitest (#2495) (25069fb)
  • deps: bump @vercel/analytics from 1.6.1 to 2.0.1 in /docs/site (#2503) (d49242d)
  • deps: bump the npm-minor-patch group (#2502) (a0b05b0)
  • infra: migrate stale yonatangross.dev refs to yonyon.ai (#2513) (4bccc6a)

Documentation

  • readme: add Ask DeepWiki documentation badge (#2509) (6c8599d)

CI/CD

  • bump anthropics/claude-code-action (#2505) (ed4cf33)
  • deps: grouped Dependabot + auto-merge across all npm trees (#2492) (1490ea6), closes #2490
  • hold Dependabot majors from auto-merge (semver gate) (#2508) (02f2442)

8.48.1 (2026-06-16)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.178) (#2484) (36af4bb)

CI/CD

8.48.0 (2026-06-16)

Features

  • memory: close the VERIFY loop — consult log + dream gate (#2351) (#2486) (a16c4a5)

Bug Fixes

8.47.3 (2026-06-16)

Bug Fixes

  • deps: bump hono to 4.12.25 (5 high CVEs) (#2487) (5cf5a48)
  • hooks: Wave 1 — revive 3 born-dead CC event hooks + #1264 drift-check (#2476) (71a72b8)

8.47.2 (2026-06-16)

Bug Fixes

  • security: unblock security-tests gate (audit + stat fix) (#2470) (32fd3cb)

Miscellaneous

  • deps: Bump the npm_and_yarn group across 2 directories with 5 updates (#2472) (8df1cd1)
  • labs: bump upstream-version-tested pins for drifted skills (#2461) (5f86cff)
  • triage: close 29 verified done/obsolete issues (wave 0) (#2468) (cae3158)

Documentation

  • add README hero banner + preview playground (#2467) (fb80d1a)

CI/CD

  • bump anthropics/claude-code-action from 1.0.140 to 1.0.148 (#2456) (951af4b)

8.47.1 (2026-06-15)

Bug Fixes

  • hooks: unify debt-scan extensions to stop capture/surface drift (#2463) (9dd1dbb)

8.47.0 (2026-06-15)

Features

  • hooks: auto-track deferred-debt markers (ork-debt ledger) (#2457) (1513a7f)

8.46.0 (2026-06-14)

Features

  • docs-site: add /yonyon page + Yonyon entity for brand-keying (#2453) (f795dd0)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.176) (#2454) (1b036ec)

8.45.0 (2026-06-14)

Features

  • hooks: haiku-powered session auto-rename + auto-color (#2449) (df86117)

Bug Fixes

  • hooks: respect CC language setting in session-title prompt (#2452) (f901335)

Documentation

  • configure: catalog CC 2.1.172-176 settings (#2451) (e0d8e81)

8.44.1 (2026-06-14)

Miscellaneous

8.44.0 (2026-06-14)

Features

  • cc-watch: note when CC binary ships ahead of its changelog notes (#2444) (1dc5d4c), closes #2442

Bug Fixes

  • cc-watch: key staleness alarm on latest_known, not upstream head (#2427) (a0d2489)

Miscellaneous

  • cc-adoption: advance latest_known to 2.1.176 (#2445) (5b1d6be)
  • cc-adoption: groom 6 no-surface 2.1.176 issues (#2446) (c7b07be)
  • cc-adoption: zero the milestone backlog with close verdicts (#2425) (7ce8cae), closes #2199
  • cc-watch: snapshot upstream CHANGELOG (2.1.175) (#2428) (8d99b2d)
  • cc-watch: snapshot upstream CHANGELOG (2.1.176) (#2441) (d4bbc05)

Documentation

  • models: mark claude-fable-5 unavailable, access suspended (#2436) (48f1c53)

8.43.0 (2026-06-12)

Features

  • cc-adoption: staleness alarm, depth-3 example, /usage attribution (#2423) (35f7a9b), closes #2418

8.42.2 (2026-06-12)

Bug Fixes

  • cc-adoption: model resilience under availableModels (WS-2) (#2420) (0a10d37)

Documentation

  • multimodal-llm: clarify 2,576px is 3x the Opus 4.6 vision budget (#2422) (f791df3)

8.42.1 (2026-06-12)

Bug Fixes

  • cc-watch: catch up snapshots to 2.1.175 + unstick featureless gate (#2415) (03d1336)

8.42.0 (2026-06-12)

Features

  • docs-site: orank Discovery — crawlable content + entity graph + flywheel prep (#2407) (318d627)

8.41.1 (2026-06-11)

Bug Fixes

  • hooks: correct hook count drift in hooks.json description (#2405) (690b721)

8.41.0 (2026-06-11)

Features

  • docs-site: category share-of-voice content batch (#2403) (14e3757)

8.40.0 (2026-06-11)

Features

  • docs-site: discovery on-site gaps — density, brand tie, dev hub (#2401) (e57afb2)

8.39.0 (2026-06-11)

Features

  • docs-site: anonymous-only RFC 9728 PRM + async-job pattern (#2392) (6656be7)

8.38.0 (2026-06-11)

Features

  • docs-site: surface MCP stdio package, policy, SDKs to checkers (#2390) (80985df)

8.37.1 (2026-06-11)

Bug Fixes

8.37.0 (2026-06-11)

Features

Bug Fixes

  • release: canonical OCI format + publish hardening (#2387) (48a9e42)

8.36.2 (2026-06-11)

CI/CD

  • release: auto-publish MCP registry on every release (#2383) (65db4a9)

8.36.1 (2026-06-11)

Bug Fixes

8.36.0 (2026-06-11)

Features

  • docs-site: skills.sh bi-directional verification + registry version sync (#2379) (f4d15fe)

8.35.0 (2026-06-11)

Features

  • delegation: specialist routing + Agent() grant repairs (#2374) (27aff02)
  • docs-site: close orank agent-readiness gaps (#2378) (bba0e84)

8.34.0 (2026-06-11)

Features

  • docs: add Glama ownership claim files (server + connector) (#2372) (7cc33f8)

8.33.0 (2026-06-11)

Features

  • adoption: adopt CC nested sub-agents + /release-notes (#2368) (6f1219d)
  • agents: nested-delegation sweep + depth injection (#2370) (59bc53b)

8.32.2 (2026-06-10)

Bug Fixes

8.32.1 (2026-06-10)

Bug Fixes

  • build: ship hook dist bundles + hard-fail stale-bundle paths (#2362) (a680f3c), closes #2360

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.170) (#2361) (f3bdc89)

8.32.0 (2026-06-10)

Features

  • hooks: fable spend-consent gate + Claude Fable 5 docs guide (#2358) (59110a2)

8.31.0 (2026-06-10)

Features

  • rubric: adopt ork-rubric/1.0 stop-gating contract (#2356) (0ce1c7f)

Miscellaneous

  • cc: bump supported floor to 2.1.170 + adopt 2.1.169 features (#2355) (44f5d29)

8.30.0 (2026-06-10)

Features

  • models: adopt Claude Fable 5 via single model vocabulary (#2339) (7b47e77), closes #2338

Bug Fixes

  • hooks: own WorktreeCreate provisioning per current CC contract (#2336) (64b7c76), closes #2335

8.29.1 (2026-06-09)

Bug Fixes

  • docs-site: move buildLlmsTxt out of the route module (#2337) (7aaf9eb), closes #2329

Miscellaneous

  • mcp: link mcp.so listing in sameAs (#2334) (d112226)
  • mcp: link Smithery listing in sameAs (#2333) (e3f3103)
  • mcp: publish to official MCP registry + link in sameAs (#2331) (0d481df)

8.29.0 (2026-06-08)

Features

  • docs-site: orank layers 2/3/4 honest wins (#2329) (615a356)

Documentation

  • security-patterns: document CC ≥2.1.166 permission-rule semantics (#2328) (ff0a52a)

8.28.1 (2026-06-08)

CI/CD

  • cc-stale: never auto-stale breaking/new_perm adoption issues (#2326) (3097b9d), closes #2325

8.28.0 (2026-06-08)

Features

  • docs-site: GEO-enhance /compare + /alternatives for AI citation (#2323) (2b20fcd)

8.27.2 (2026-06-08)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.167) (#2319) (16e3c1d)

Documentation

  • hooks-ref: clarify 143 global vs 211 total hook count (#2321) (e92df60), closes #2279

8.27.1 (2026-06-08)

Bug Fixes

Miscellaneous

CI/CD

  • labs: rebuild plugins/ in version-watch bot to fix drift gate (#2317) (dc9c299)

8.27.0 (2026-06-08)

Features

  • docs-search: show stable facet counts on browser pills (#2282) (#2311) (7b99d77)
  • docs-site: content-negotiated JSON 404 + MCP App-view CSP (#2316) (128aae9)
  • docs-site: MCP Apps (ui:// resources) + JSON /api index (#2312) (4f28250)
  • docs-site: orank B→A — API standards, schema, llms, MCP (#2308) (4144c08)

Miscellaneous

  • mcp: add server.json for MCP registry publish (#2309) (cef2957)

Documentation

CI/CD

  • bump actions/checkout from 6.0.2 to 6.0.3 (#2300) (7a5899c)

8.26.0 (2026-06-08)

Features

  • docs-site: link homepage sameAs to Wikidata (Q140128295) (#2305) (db60407)

Miscellaneous

  • hooks: label headroom inert + fix test env isolation (#2304) (51fd608)

CI/CD

  • bump anthropics/claude-code-action from 1.0.133 to 1.0.140 (#2299) (fd12189)
  • bump github/codeql-action from 4.36.0 to 4.36.2 (#2298) (fcab076)

8.25.2 (2026-06-07)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.167) (#2296) (c26db27)
  • cc: bump supported floor to 2.1.168 + fix stale banner counts (#2295) (aae6be4)

8.25.1 (2026-06-07)

Bug Fixes

8.25.0 (2026-06-07)

Features

  • docs-search: unified Orama search (relevance, typo, coverage) (#2278) (129dcc4)
  • docs-site: agent-readiness endpoints for orank gaps (#2279) (ddd5b7d)
  • skills: adopt CC 2.1.163 home-path deny note + verify 2.1.162/163 backlog (#2291) (93bb140)

Bug Fixes

  • cc-triage: run featureless reconciliation token-free + heal the ledger (#2290) (aa9f289)
  • docs-site: open CORS for cross-origin agents + MCP paths (#2292) (c03c8e8)

Miscellaneous

8.24.0 (2026-06-07)

Features

Bug Fixes

  • cc-triage: graduate featureless CC versions + snapshot 2.1.168 (#2271) (9c8c92b)

Documentation

8.23.1 (2026-06-06)

CI/CD

  • exempt ci/ config branches from the PR playground gate (#2262) (797bf61)
  • vercel: production-only ignoreCommand — no preview deploys (#2261) (fb9b04a)

8.23.0 (2026-06-06)

Features

  • audit-full: map-reduce workflow tier for over-budget repos (#2259) (8c74273)
  • skills: adversarial verification for review-pr + audit-full (#2256) (f24926a)

Bug Fixes

  • skills: trim review-pr under length gate + CC-orchestration hygiene (#2260) (48311ca)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.167) (#2255) (375eaa3)

Documentation

  • chain-patterns: label the 6 workflow patterns as ork's taxonomy (#2258) (f68bfdd)

8.22.0 (2026-06-06)

Features

  • docs: agent-ready docs site (well-known + content negotiation) (#2236) (fa749a6)
  • hooks: warn on concurrent same-tree session collision (#2242) (#2245) (7ed8cf2)

8.21.0 (2026-06-06)

Features

  • skills: adversarial verification engine + assess pilot (#2231) (#2235) (f1c90a1)
  • skills: dynamic-workflow patterns reference + quarantine rule (#2240) (a082314)

Bug Fixes

  • cc: stamp LATEST_KNOWN_CC from SoT + repair auto-bump workflow (#2241) (3499996), closes #2238
  • hooks: pin session title to first-seen branch (survives concurrent-session thrash) (#2246) (fec9775)
  • security: correct dependency-confusion package ref + add guard (#2239) (813cf8c)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.165) (#2228) (76a5407)

Documentation

Code Refactoring

8.20.1 (2026-06-05)

Code Refactoring

  • hooks: remove stop/goal-convergence-emitter (drift cleanup) (#2219) (e1236e2), closes #2217

8.20.0 (2026-06-05)

Features

  • eval: ship skill-fitness workflow harness (#2212) (099abd4)

Bug Fixes

  • hooks: give SessionEnd handoff a real budget (CC 2.1.74 env) (#2214) (d3d1205)
  • skills: clear skill-fitness workflow findings (#2215) (5b217b9)
  • skills: portable install paths + C3 static gate (#2218) (c4c507e)

Documentation

  • rules: codify CC-native-first (don't drift from Claude Code) (#2216) (2645b78)

8.19.1 (2026-06-05)

CI/CD

8.19.0 (2026-06-04)

Features

  • evals: CC-idiom conformance grader (advisory) (#2207) (6083a44)

Bug Fixes

  • deps: bump Hono 4.12.18 -> 4.12.23 (4 moderate CVEs) (#2208) (1771c1f)
  • hooks: guard tool_input.file_path against non-string (#2205) (cb8b299)
  • skills: clear the 5 CC-idiom conformance findings (#2209) (6640c13)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.162) (#2204) (cf51913)

8.18.1 (2026-06-04)

Miscellaneous

  • cc: catalogue CC 2.1.162 in version matrix + README ledger (#2196) (8e74901)
  • docs: resync fumadocs reference pages + add drift gate (#2198) (a94b5e0)

8.18.0 (2026-06-03)

Features

  • hooks: memory-driven lint — feedback_*.md rules to linter (#1901) (#2190) (c69ae80)
  • tests: extend tool-call verifier to TaskCreate/TaskUpdate (#2188) (41764f0), closes #2177

8.17.0 (2026-06-03)

Features

  • cc: adopt CC 2.1.142 + 2.1.145 stragglers (3 issues) (#2186) (6897217)
  • tests: schema-conformance verifier for skill tool-call examples (#2185) (f7eb487), closes #2177

8.16.2 (2026-06-03)

Miscellaneous

Documentation

  • cc: close CC 2.1.140-147 verify/doc backlog (8 issues) (#2183) (a41299c)

8.16.1 (2026-06-03)

CI/CD

  • bump actions/upload-artifact from 5.0.0 to 7.0.1 (#2157) (4e714c3)
  • bump gitleaks/gitleaks-action from 2.3.9 to 3.0.0 (#2158) (2fe3106)

8.16.0 (2026-06-03)

Features

  • cc: adopt Claude Code 2.1.157 + refresh README (#2179) (68aadea)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.161) (#2181) (de42100)

CI/CD

  • bump actions/setup-python from 5.6.0 to 6.2.0 (#2159) (9e06d3e)

8.15.0 (2026-06-03)

Features

Bug Fixes

  • skills: correct stale tool-call signatures in skill examples (#2176) (cf07883)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.159) (#2162) (44f38b2)
  • cc-watch: snapshot upstream CHANGELOG (2.1.160) (#2169) (8c4f250)
  • cc-watch: snapshot upstream CHANGELOG (2.1.161) (#2173) (be7513f)

8.14.0 (2026-06-01)

Features

  • hooks: channel-aware skill telemetry (main + subagent) (#2154) (3cb8375)

Bug Fixes

  • ci: split Scorecard publish + SARIF filter into two jobs (#2155) (5f365f3)

8.13.1 (2026-06-01)

CI/CD

  • pin ci-sentinel actions + filter irreducible Scorecard pinned-deps (#2152) (68ddc00)

8.13.0 (2026-06-01)

Features

  • skills: wire 9 island skills + add activation-channel CI gate (#2150) (c2cb0dd)

8.12.8 (2026-06-01)

Bug Fixes

  • close library-audit coverage-misses (PostgresSaver, jose, gpt-5.5) (#2147) (ab523ea)
  • hooks: antipattern hook recommends PyJWT not python-jose (CVE) (#2149) (14f3a30)

8.12.7 (2026-05-31)

Bug Fixes

  • skills: residual library-currency sweep (16 skills) (#2145) (f32cfd7)

8.12.6 (2026-05-31)

Bug Fixes

  • skills: Lane 1 library hard-breaks across 11 skills (#2142) (ac9a7b8)
  • skills: Lane 2+3 library remediation (12 clusters) (#2143) (e4854df)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.158) (#2127) (34f14a1)

8.12.5 (2026-05-31)

Documentation

  • generate reference-skills page from frontmatter (closes #2120) (#2123) (3166e86)
  • refresh stale model sample + CC-floor cites (post-ship sweep) (#2124) (a885c3f)

8.12.4 (2026-05-31)

Bug Fixes

  • cost-estimator: add claude-opus-4-8 pricing row + repoint opus alias; (abcd0ba)
  • post-ship 8.12.3 hardening (doc counts, AUQ guard, CI, pricing) (#2121) (abcd0ba)

Documentation

  • generator: _build-docs-generate.py counted every skill dir (incl (abcd0ba)
  • troubleshooting: bump stale CC floor 2.1.34 -> 2.1.148. (abcd0ba)

CI/CD

  • release-please-guard: trigger on labeled/unlabeled so a (abcd0ba)

8.12.3 (2026-05-30)

Miscellaneous

  • deps: major bumps esbuild 0.28, @types/node 24, lucide 1.x (#2115) (500e238)
  • marketplace: align engine floor to >=2.1.148 (#2116) (260c5da)
  • skills: preview-strip + 4.8 model bump + doctor count refresh (#2118) (4777cee)

Documentation

  • site: uplift — IA, fumadocs 15, perf, a11y + stale-content sweep (#2117) (2c6cd11)

8.12.2 (2026-05-30)

Documentation

  • readme: point doc links at orchestkit.yonyon.ai (#2113) (a5a65c8)

8.12.1 (2026-05-30)

Bug Fixes

  • remediate 2026-05-30 staleness audit (CC 2.1.158 adoption + Opus 4.8 sweep) (#2111) (d9f5ff8)

8.12.0 (2026-05-30)

Features

  • docs-site: serve docs from orchestkit.yonyon.ai (canonical) (#2101) (442a02f)
  • docs: docs-site uplift — IA, nav, fumadocs 15, perf (#2100) (f52b7d3)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.158) (#2110) (981b854)

8.11.1 (2026-05-29)

Bug Fixes

  • ci: npm publish tarball must use an absolute path (#2098) (4b6d22b)

8.11.0 (2026-05-29)

Features

  • ci: lockstep npm publish workflow for hook-contract (#2096) (067d16c)

Bug Fixes

Miscellaneous

  • hook-contract: release 0.1.1 (npm + PyPI lockstep) (#2097) (5475c1c)

8.10.2 (2026-05-29)

Bug Fixes

  • cc-watch: carry triaged gaps + shared visual-style validator (#2090) (b1f3304)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.156) (#2089) (4117ee0)
  • dx: fix Closes-N drift template bug + advisory body lint (#2091) (9cd4f9e)

8.10.1 (2026-05-29)

Bug Fixes

  • adoption: 9 CC 2.1.154 stragglers the sweep missed (#2086) (7ce52b1)
  • cc-watch: restore triaged gaps for 2.1.153/154 (#2084) (d1df958)

8.10.0 (2026-05-29)

Features

  • hooks: finish M168 coordination + subagent budget signal (#2054) (a8313f3)

Documentation

  • skills: adopt CC 2.1.153 tail + close 2.1.153/154 issue sweep (#2083) (112fd32)
  • skills: adopt CC 2.1.154 (Opus 4.8, /simplify, worktree) (#2058) (fc76fa5)

8.9.3 (2026-05-28)

Bug Fixes

8.9.2 (2026-05-28)

Documentation

8.9.1 (2026-05-28)

Bug Fixes

  • cc-watch: dedup on changelog-ref hash, not LLM slug (#2041) (#2044) (d0765f4)
  • cc-watch: resilient cc-stale label-ensure (don't die on a 403) (#2040) (1b8f56e)
  • ci: build hooks from source in Manifests job (#2042) (295589a)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.152) (#2037) (d99fc5f)

8.9.0 (2026-05-27)

Features

  • cc-watch: loud auth-failure signal + below-floor cc-stale sweep (#2035) (65fba9b)

Bug Fixes

  • cc-watch: preserve parse_failed gaps + umbrella the fallback issue (#2034) (3a381b5)
  • skills: conform AskUserQuestion calls to CC 2.1.152 schema (#2023) (19d42c8)

Miscellaneous

  • cc-watch: collapse per-version milestones into rolling umbrella (#2019) (e24aa5d)
  • cc-watch: snapshot upstream CHANGELOG (2.1.152) (#2022) (388b725)
  • cc-watch: snapshot upstream CHANGELOG (2.1.152) (#2024) (8f48559)
  • cc-watch: snapshot upstream CHANGELOG (2.1.152) (#2033) (2fe2a08)

8.8.0 (2026-05-26)

Features

  • hooks: M168 skill_invocation table for usage analytics (#2015) (bc1e1bb), closes #2010

Bug Fixes

Miscellaneous

  • hooks: M168 SQLite test/build cleanup + ci-report Node 22 (#2018) (73cefe3), closes #2017

8.7.1 (2026-05-26)

Bug Fixes

  • hooks: issue-ref-checker scope, worktree branch, non-blocking (#2004) (99d6d00)

Miscellaneous

  • labs: bump upstream-version-tested pins for drifted skills (#2001) (dd32737)

8.7.0 (2026-05-25)

Features

  • hooks: migrate session registry to node:sqlite (#2005) (#2008) (d8f88b5)
  • visualize-plan: format front-door + before/after arch (#2002) (42f9ede)

Documentation

  • cc-2147: /code-review rename + multi-Agent tools guard (#1999) (19230b9)

8.6.6 (2026-05-25)

Bug Fixes

  • hooks: hook bundles load without node_modules (lazy better-sqlite3) (#2003) (294471e)

CI/CD

  • bump actions/download-artifact from 4.3.0 to 8.0.1 (#1993) (bd855b3)
  • bump actions/setup-node from 4 to 6 (#1995) (74f819e)
  • bump anthropics/claude-code-action from 1.0.123 to 1.0.133 (#1992) (7fb1619)
  • bump github/codeql-action from 4.35.5 to 4.36.0 (#1994) (53abe17)
  • bump pypa/gh-action-pypi-publish from 1.12.4 to 1.14.0 (#1991) (de2400e)

8.6.5 (2026-05-25)

Bug Fixes

  • hooks: WorktreeCreate emits empty stdout, not the envelope (#1996) (4d4400b)

8.6.4 (2026-05-24)

Bug Fixes

8.6.3 (2026-05-24)

Bug Fixes

8.6.2 (2026-05-24)

Documentation

  • skills: adopt CC 2.1.140-2.1.148 long/bg-session features (#1981) (7d94722)

8.6.1 (2026-05-24)

Bug Fixes

Miscellaneous

8.6.0 (2026-05-23)

Features

8.5.0 (2026-05-23)

Features

  • hooks: M168 Phase 4 — Layer 4 worktree + settings overrides (#1969) (094f595)

Bug Fixes

  • hooks: emit security-hook override warning before bundle load (#1974) (d846f99)
  • hooks: tighten watchdog session-id filter (#1971) (a434421)

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.150) (#1973) (cb7113f)

8.4.0 (2026-05-23)

Features

  • hooks: adopt CC 2.1.145 background_tasks in agent-watchdog (#1967) (800feb0)
  • tests: scan SKILL.md compat floors (closes #1963) (#1968) (c476f9f)

Bug Fixes

Miscellaneous

  • skills: bump 108 SKILL.md floors to 2.1.139 + flip STRICT default (#1970) (82d4830)

Documentation

  • milestones: mark M170 as superseded after prior-art assessment (#1962) (cbf5920)

8.3.0 (2026-05-23)

Features

  • skills: pilot visual-style on explore, assess, brainstorm (#1955) (fd3095d)
  • visual-style: lint PR titles + bodies + code comments (#1960) (9830e2e)

Documentation

  • milestones: draft M170 spec — skill auto-eval vs latest CC (#1961) (e332e1b)

8.2.1 (2026-05-22)

Bug Fixes

  • deps: bump transitive qs 6.15.0 → 6.15.2 (GHSA-q8mj-m7cp-5q26) (#1957) (201229d)
  • hooks: cross-session watchdog filter + auto-rebuild bundle (#1952) (a625044)

Miscellaneous

  • deps: remove stale pnpm-lock.yaml (closes 3 dependabot alerts) (#1953) (218dfe0)

Documentation

  • audits: postmortem for release-please version drift incident (#1950) (4e30fd5)
  • skills: brainstorm CC 2.1.141 + 2.1.143 version-floor notes (#1951) (f6d360c)

CI/CD

  • bump the npm_and_yarn group across 2 directories with 1 update (#1958) (5159133)

8.2.0 (2026-05-22)

Features

  • hooks: events.jsonl + chain-staleness + goal-emitter (M168 Phase 3) (#1948) (f1fcb70)

8.1.0 (2026-05-22)

Features

  • hooks: SQLite Layer 1 — session registry + locks (M168 Phase 2) (#1920) (96267da)

Bug Fixes

  • release: revert version drift + guard against recurrence (#1947) (eb4059d)

Documentation

  • rules: add visual-style rule for ASCII art and emoji usage (#1939) (f831757)

8.0.0 (2026-05-22)

⚠ BREAKING CHANGES

  • removes the entire monitors mechanism from ork. The 3 monitors (goal-convergence, chain-state-watcher, uncommitted-work-reminder) are deleted with no replacement in this PR. Subsequent M168 phases (#1912-#1916) add SQLite-backed event-driven replacements.

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • #117: close M117 — doctor MCP pinning + hook property tests (#1496) (6adc493)
  • #1489: M121 observability consolidation (#1492) (d5ea635)
  • #1525: /ork:dev gains scripts, rules, test-cases (#1536) (cb30501)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • ASCII Design System + status glyph vocab + preview audit + drift fixes (#1619) (1770365)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • cc: adopt CC 2.1.122 + 2.1.126 features (M128) (#1585) (372b9d0)
  • cc: adopt CC 2.1.128 + bump floor to 2.1.125 (#1616) (4686952)
  • cc: bolder floor bump 2.1.125 → 2.1.132 (#1623) (2292524)
  • cc: bump CC floor 2.1.132 → 2.1.138 (#1682) (#1734) (70c2e4a)
  • cc: M130 auto-adoption pipeline (#1486 #1487 #1488) (#1589) (ec14387)
  • ci: cross-language parity gate (M141-6, #1807) (903d870)
  • ci: de-noise PR rollup signals + auto-sync version files (Closes #1403–#1407) (#1409) (5ed87b3)
  • ci: pre-push gate + commit-msg linter + token-budget warn tier (#1411) (77dc14c)
  • ci: PyPI publish workflow for hook-contract-py (#1864) (204efe9)
  • ci: Road to 10 Wave 3 — agent registry + npm audit + backlog admin (#1418) (f902098)
  • claude-design: Bet A — closed-loop pipeline (Milestone #113) (#1390) (f93fa11)
  • claude-design: Bet A polish — schema rewrite + mdx-compile guard (Closes #1400, #1401) (#1402) (e97e18c)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.110 full adoption — matrix, skills, doctor (#1349) (0d166e9)
  • compat: CC 2.1.111 + Opus 4.7 full adoption (#1376) (425a7f7)
  • compat: CC 2.1.117 adoption + milestone #117 Tier A closeout (#1454) (b7c5898)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • docs: redesign landing page from Claude Design handoff (Closes #1247) (#1399) (1005354)
  • doctor: wire agent-browser doctor --json (#1559) (#1573) (399b030)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hard-delete monitors.json (M168 Phase 1) (#1919) (49591db)
  • hook-contract-py: scaffold PyPI sibling (M141-3, #1804) (2960185)
  • hook-contract: codegen scaffold + drift gate (M141-2 step 1, #1864) (b17fb03)
  • hook-contract: HMAC signing protocol + impl (M141-4, #1805) (#1878) (762ab0f)
  • hook-contract: per-event payload schemas (M141-2 step 3, #1864) (38eeb28)
  • hook-contract: scaffold @orchestkit/hook-contract + setup token enforcement (M141-1) (2cbf5e2)
  • hooks,#1478: content-aware token estimator — replace char/4 (#1484) (782677c)
  • hooks,#1478: coverage hardening — close all P1 + P2 audit gaps (#1485) (b1e7757)
  • hooks,evals: CC 2.1.116 enforcement + eval coverage (#1437) (7f73d05)
  • hooks,M119: dogfood cluster — zones, cache-aware, outcome (#1515) (8641214)
  • hooks,skills: MS#117 Bundles 1+2 — integration harness + wiring (#1456) (082b1da)
  • hooks/lib: dual-write analytics to yonatan-hq platform (#1799) (7070412)
  • hooks: CC 2.1.113 adoption — native binary, find/rm hardening, deniedDomains (#1388) (104226e)
  • hooks: cross-session state bus for #1885 (#1906) (9bab036)
  • hooks: deferred worktree advisory consumer — PR-2 #1794 follow-up (#1801) (f017d9e)
  • hooks: local-dev hygiene bundle — 3 advisory hooks (#1574) (dc446a1)
  • hooks: M119 usage hygiene — pre-compact prompt + 150k context warn (#1474) (39507eb)
  • hooks: rules-size-check + storage-patterns (#1815 #1816) (#1817) (a6ca836)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • hooks: sweep stale empty worktree paths (closes #1884) (#1905) (834467d)
  • hooks: tool-invocation-linter framework (#1883 seed) (#1900) (3800941)
  • hooks: Usage-Driven Hardening bundle — 5 analytics hooks + batch rule (#1378) (7c5a3d1)
  • hooks: Usage-Driven Hardening completion — milestone close-out (3 issues) (#1380) (e23be85)
  • implement: codify manual-worktree for parallel agents (#1824) (a0b80b7)
  • M118: elicitation retrofit (5 issues) (#1510) (b2577df)
  • M122: adopt CC 2.1.118 + 2.1.119 across stack (#1507) (9fad64a)
  • M125 Lane C: json-render dashboards (#1527) + Storybook → genui catalog (#1529) (#1539) (afb19e0)
  • M125: Lane A — labs version watcher + emulate auto-discovery (#1532) (a53f342)
  • M125: Lane B — /ork:dev lab-stack + auto-expect + ARIA snapshots (#1534) (1e626fd)
  • M126: ${CLAUDE_EFFORT} env (#1540) + /skills filter tip (#1546) (#1549) (d5ce5f6)
  • M126: alwaysLoad MCP (#1541) + claude plugin prune (#1544) (#1551) (dedc53e)
  • M126: claude ultrareview CI (#1542) + CLAUDE_CODE_FORK_SUBAGENT eval (#1545) (#1552) (4849d01)
  • M126: secret-handler PostToolUse hook (#1543, Bundle C — last of M126) (#1553) (8cf8448)
  • m127: bundle Vercel Labs 2026-04-29 adoption (5 issues) (#1604) (ac9ec5b)
  • M127: cost-hygiene A1 — CI fork-subagent + slash-only skills (#1572) (1d947a9)
  • M128: monitor quickwins — manifest source, debounce, docs (#1570) (fdc2166)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skill: /ork:ci-sentinel — hourly CI-failure classifier (#1842) (5ce3df1)
  • skill: /ork:swarm-migrate — cross-repo migration coordinator (#1839) (20c54f3)
  • skills,#117: M117 adoption closeout — Monitor + Push + OTEL + cron (#1494) (24cf12d)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: notebooklm v0.5.25 full adoption (#1352) (f49cf3d)
  • skills: round 2 yg-mcp-core consumers — explore + review-pr + fix-issue (#1898) (038ac22)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • skills: wire --permission-mode + --effort dispatch envelope (#1855) (832d4c8)
  • skills: yg-mcp-core consumers across brainstorm/assess/memory (closes #1886, #1887, #1888) (#1889) (1473f05)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • #1525: bump invocable caps + dev argument-hint (#1535) (9709025)
  • #1525: finish real implementations — close 4 vapor gaps (#1538) (f48821f)
  • add the same step-level env block to Step 1 too. (d79dbb5)
  • brainstorm: restore 81 lines of headroom under cap (#1578) (75ba909)
  • brainstorm: trim SKILL.md to 500-line cap (unblocks 4 PRs) (#1576) (247943a)
  • build: preflight check for missing node_modules (closes #1899) (#1904) (431c66e)
  • cc-watch: normalize double-dash bullets + --retry-failed flag (#1614) (4a9f52c)
  • cc-watch: silent-skip + reissue flag + LLM retry (#1730) (78aa054)
  • cc-watch: skip cc-triage for versions below supported floor (#1739) (#1740) (5e551a6)
  • ci-sentinel: drop --bare (auth broken), bump budget 500k→1M (#1857) (42ee3ab)
  • ci-sentinel: file-backed apiKeyHelper + safer smoke test (#1859) (25cfbfc)
  • ci-sentinel: install Claude Code on the runner (#1856) (b723901)
  • ci-sentinel: wire apiKeyHelper auth + shared --bare gotcha rule (#1858) (1bd3d19)
  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add GH_TOKEN to Snapshot step (#1598) (d79dbb5)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: bump token-overhead + skill-length caps + add missing frontmatter (#1865) (9951ffd)
  • ci: cc-triage drops --bare so CLAUDE_CODE_OAUTH_TOKEN is honored (#1629) (fbb3199)
  • ci: cc-triage uses opus alias + logs stdout on failure (#1633) (311a514)
  • ci: cc-watch PR uses PAT + version-stamped title (#1685 #1697) (#1698) (fbd1728)
  • ci: cc-watch uses GH App token (#1735) (#1736) (e726b8d)
  • ci: clean up claude-release-watch on: block + Title Case name (#1594) (d686e91)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • ci: enable auto-merge on cron-generated snapshot PRs (#1603) (cfea028)
  • ci: heredoc breaks YAML + secrets context not in if: (#1595) (649d889)
  • ci: install CC native binary explicitly when --ignore-scripts (#1622) (c6fb9c3)
  • ci: install claude CLI + PR-based snapshot commit (#1596) (dc19147)
  • ci: persist-credentials=false on checkout (#1599) (8d3b728)
  • ci: re-register cc-release-watch workflow_dispatch trigger (#1592) (2ed4e7f)
  • ci: regenerate docs/site lockfile on-the-fly, delete committed stub (#1631) (7f6f53b)
  • ci: release-video composition paths — strip fictional folder prefix (#1432) (4e639f2)
  • ci: rename cc-release-watch.yml → claude-release-watch.yml (#1593) (b89bd7e)
  • ci: scope GH_TOKEN to per-step (peter-evans needs clean git env) (#1597) (6eb594f)
  • ci: Scorecard alerts + full GitHub App migration (retire PAT) (#1426) (8400096)
  • ci: skip pre-push version check on feat/fix/perf/refactor branches (#1458) (065454d)
  • demos: bump zod to ^4.3.6 (unblock Release Video pipeline) (#1430) (8f26d0a)
  • deps: override fast-uri to ^3.1.2 for CVE-2026-6321/6322 (#1687) (#1688) (6ea510e)
  • deps: override hono / ip-address / express-rate-limit for CVEs (#1624) (95acb78)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync stale skill/agent/hook counts in fumadocs site (#1516) (f4fb5ad)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: #1795 — ORK_ASK_FALLBACK=text picker stall mitigation (#1798) (19fbc6e)
  • hooks: allow SessionStart + PostCompact in output-guard (#1234) (#1822) (921e113)
  • hooks: cap agent-watchdog at 24h to reap zombies (#1882) (#1896) (a03e5cb)
  • hooks: close JSON-leak holes that mkdir literal {...} dirs (#1606) (f93e88d)
  • hooks: envelope input guard at WorktreeCreate.name (#1826) (#1837) (9ac4e75)
  • hooks: isProtectedBranch treats empty string as not-protected (#1425) (b4f9b40)
  • hooks: kill #1794 — WorktreeCreate output shape + dispatcher guard (#1797) (3e6d164)
  • hooks: preserve /rename via sessionTitle delta-detect (#1779) (2cb8ec6)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: titler double-prefix — strip leading ork: before wrap (#1789) (2abf7e0)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • m130: repair cron pipeline + harden stamper + fill triage test gap (#1591) (99ab20d)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • release: SBOM/attestation skip + changelog-to-props parser (#1428) (1159475)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: override ws to ^8.20.1 — patches GHSA-58qx-3vcg-4xpx (#1879) (ab159cd)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skill: port /ci-debug into ork (cross-skill-ref alignment) (#1843) (7068d71)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)
  • tests: isolate stop-uncommitted-check fixtures (closes #1917) (#1918) (6eac2a9)
  • ultrareview: label-gate + fix broken TARGET arg + suppress empty comment (#1569) (c8097f5)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • agents: recategorize all 37 agents — 7 readable categories (#1523) (ae9c650)
  • claude: replace static counts with plugin-details pointer (D2) (#1777) (e04ba7d)
  • release: add v7.86 release-window slideshow + infographic (#1758) (c5b00e8)
  • release: add v7.86 release-window slideshow + infographic (#1776) (6d673c1)
  • skills: codify manual-worktree pattern in /ork:cover (closes platform#3224) (#1835) (523b39d)
  • skills: deepen @json-render/directives coverage in skill body (#1755) (5de3d5b)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)
  • tests: drop fragile counts from cc-version-matrix (#1761) (f2a29bd)

Performance

  • hooks: migrate agentation-context → SessionStart (M104 PR-B) (#1827) (cfa0984)
  • hooks: migrate ask-fallback-injector → SessionStart (M104 PR-A) (#1825) (34d80de)

CI/CD

  • 50 SUCCESS / 3 FAILURE — all 3 failures pre-existing on origin/main, tracked in #1865. (2cbf5e2)
  • actionlint enforcement (pre-commit + CI workflow) (#1601) (cf3bcaf)
  • bump actions/cache from 5.0.4 to 5.0.5 (#1421) (b548f86)
  • bump actions/checkout from 4 to 6 (#1851) (9e6dd22)
  • bump actions/checkout from 5 to 6 (#1420) (4c18de7)
  • bump actions/create-github-app-token from 3.1.1 to 3.2.0 (#1850) (a861187)
  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 4.4.0 to 6.4.0 (#1521) (08140ff)
  • bump actions/setup-node from 4.4.0 to 6.4.0 (#1612) (ce35dd9)
  • bump actions/setup-node from 5 to 6 (#1423) (77105cc)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 4.6.2 to 7.0.1 (#1854) (bb76cbe)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.101 to 1.0.107 (#1519) (23d2a7c)
  • bump anthropics/claude-code-action from 1.0.107 to 1.0.111 (#1611) (6cfd862)
  • bump anthropics/claude-code-action from 1.0.111 to 1.0.119 (#1743) (c672ae8)
  • bump anthropics/claude-code-action from 1.0.119 to 1.0.123 (#1852) (88b4134)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump anthropics/claude-code-action from 1.0.93 to 1.0.101 (#1422) (2b0bfec)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • bump github/codeql-action from 4.35.1 to 4.35.2 (#1522) (f5fdc20)
  • bump github/codeql-action from 4.35.2 to 4.35.3 (#1609) (fe98da4)
  • bump github/codeql-action from 4.35.3 to 4.35.4 (#1744) (f99744e)
  • bump github/codeql-action from 4.35.4 to 4.35.5 (#1853) (d889da6)
  • bump googleapis/release-please-action from 4.4.0 to 4.4.1 (#1419) (ec12455)
  • bump googleapis/release-please-action from 4.4.1 to 5.0.0 (#1520) (7005750)
  • bump peter-evans/create-pull-request from 7.0.5 to 8.1.1 (#1610) (d172010)
  • bump the npm_and_yarn group across 2 directories with 3 updates (#1630) (2e6010c)
  • playground-gate: allow-list cron snapshot + orchestkit-release-bot (#1820) (7069d57)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.95.0] - 2026-05-22

⚠ BREAKING CHANGES

  • monitors: the entire src/monitors/ mechanism is deleted with no backwards compatibility. The 3 polling shells (goal-convergence, chain-state-watcher, uncommitted-work-reminder) are removed in this release as Phase 1 of M168 Coordination Redesign (#1908). Subsequent M168 phases (#1912-#1916) ship SQLite-backed event-driven replacements. commit-nudge.ts (PostToolUse hook) continues to fire and covers the only useful monitor signal (uncommitted-work-reminder).

Removed

  • src/monitors/ directory (entire)
  • "monitors" registration from manifests/ork.json
  • ## Monitors section from CLAUDE.md
  • Monitors copy step + has_monitors flag from scripts/build-plugins.sh

Fixed

  • Closes #1877 — goal-convergence notification spam (monitor deleted entirely)

Related

  • Parent umbrella: #1908 (M168 Coordination Redesign)
  • This PR: #1911 (Phase 1)

7.94.0 (2026-05-22)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • #117: close M117 — doctor MCP pinning + hook property tests (#1496) (6adc493)
  • #1489: M121 observability consolidation (#1492) (d5ea635)
  • #1525: /ork:dev gains scripts, rules, test-cases (#1536) (cb30501)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • ASCII Design System + status glyph vocab + preview audit + drift fixes (#1619) (1770365)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • cc: adopt CC 2.1.122 + 2.1.126 features (M128) (#1585) (372b9d0)
  • cc: adopt CC 2.1.128 + bump floor to 2.1.125 (#1616) (4686952)
  • cc: bolder floor bump 2.1.125 → 2.1.132 (#1623) (2292524)
  • cc: bump CC floor 2.1.132 → 2.1.138 (#1682) (#1734) (70c2e4a)
  • cc: M130 auto-adoption pipeline (#1486 #1487 #1488) (#1589) (ec14387)
  • ci: cross-language parity gate (M141-6, #1807) (903d870)
  • ci: de-noise PR rollup signals + auto-sync version files (Closes #1403–#1407) (#1409) (5ed87b3)
  • ci: pre-push gate + commit-msg linter + token-budget warn tier (#1411) (77dc14c)
  • ci: PyPI publish workflow for hook-contract-py (#1864) (204efe9)
  • ci: Road to 10 Wave 3 — agent registry + npm audit + backlog admin (#1418) (f902098)
  • claude-design: Bet A — closed-loop pipeline (Milestone #113) (#1390) (f93fa11)
  • claude-design: Bet A polish — schema rewrite + mdx-compile guard (Closes #1400, #1401) (#1402) (e97e18c)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.110 full adoption — matrix, skills, doctor (#1349) (0d166e9)
  • compat: CC 2.1.111 + Opus 4.7 full adoption (#1376) (425a7f7)
  • compat: CC 2.1.117 adoption + milestone #117 Tier A closeout (#1454) (b7c5898)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • docs: redesign landing page from Claude Design handoff (Closes #1247) (#1399) (1005354)
  • doctor: wire agent-browser doctor --json (#1559) (#1573) (399b030)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hook-contract-py: scaffold PyPI sibling (M141-3, #1804) (2960185)
  • hook-contract: codegen scaffold + drift gate (M141-2 step 1, #1864) (b17fb03)
  • hook-contract: HMAC signing protocol + impl (M141-4, #1805) (#1878) (762ab0f)
  • hook-contract: per-event payload schemas (M141-2 step 3, #1864) (38eeb28)
  • hook-contract: scaffold @orchestkit/hook-contract + setup token enforcement (M141-1) (2cbf5e2)
  • hooks,#1478: content-aware token estimator — replace char/4 (#1484) (782677c)
  • hooks,#1478: coverage hardening — close all P1 + P2 audit gaps (#1485) (b1e7757)
  • hooks,evals: CC 2.1.116 enforcement + eval coverage (#1437) (7f73d05)
  • hooks,M119: dogfood cluster — zones, cache-aware, outcome (#1515) (8641214)
  • hooks,skills: MS#117 Bundles 1+2 — integration harness + wiring (#1456) (082b1da)
  • hooks/lib: dual-write analytics to yonatan-hq platform (#1799) (7070412)
  • hooks: CC 2.1.113 adoption — native binary, find/rm hardening, deniedDomains (#1388) (104226e)
  • hooks: cross-session state bus for #1885 (#1906) (9bab036)
  • hooks: deferred worktree advisory consumer — PR-2 #1794 follow-up (#1801) (f017d9e)
  • hooks: local-dev hygiene bundle — 3 advisory hooks (#1574) (dc446a1)
  • hooks: M119 usage hygiene — pre-compact prompt + 150k context warn (#1474) (39507eb)
  • hooks: rules-size-check + storage-patterns (#1815 #1816) (#1817) (a6ca836)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • hooks: sweep stale empty worktree paths (closes #1884) (#1905) (834467d)
  • hooks: tool-invocation-linter framework (#1883 seed) (#1900) (3800941)
  • hooks: Usage-Driven Hardening bundle — 5 analytics hooks + batch rule (#1378) (7c5a3d1)
  • hooks: Usage-Driven Hardening completion — milestone close-out (3 issues) (#1380) (e23be85)
  • implement: codify manual-worktree for parallel agents (#1824) (a0b80b7)
  • M118: elicitation retrofit (5 issues) (#1510) (b2577df)
  • M122: adopt CC 2.1.118 + 2.1.119 across stack (#1507) (9fad64a)
  • M125 Lane C: json-render dashboards (#1527) + Storybook → genui catalog (#1529) (#1539) (afb19e0)
  • M125: Lane A — labs version watcher + emulate auto-discovery (#1532) (a53f342)
  • M125: Lane B — /ork:dev lab-stack + auto-expect + ARIA snapshots (#1534) (1e626fd)
  • M126: ${CLAUDE_EFFORT} env (#1540) + /skills filter tip (#1546) (#1549) (d5ce5f6)
  • M126: alwaysLoad MCP (#1541) + claude plugin prune (#1544) (#1551) (dedc53e)
  • M126: claude ultrareview CI (#1542) + CLAUDE_CODE_FORK_SUBAGENT eval (#1545) (#1552) (4849d01)
  • M126: secret-handler PostToolUse hook (#1543, Bundle C — last of M126) (#1553) (8cf8448)
  • m127: bundle Vercel Labs 2026-04-29 adoption (5 issues) (#1604) (ac9ec5b)
  • M127: cost-hygiene A1 — CI fork-subagent + slash-only skills (#1572) (1d947a9)
  • M128: monitor quickwins — manifest source, debounce, docs (#1570) (fdc2166)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skill: /ork:ci-sentinel — hourly CI-failure classifier (#1842) (5ce3df1)
  • skill: /ork:swarm-migrate — cross-repo migration coordinator (#1839) (20c54f3)
  • skills,#117: M117 adoption closeout — Monitor + Push + OTEL + cron (#1494) (24cf12d)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: notebooklm v0.5.25 full adoption (#1352) (f49cf3d)
  • skills: round 2 yg-mcp-core consumers — explore + review-pr + fix-issue (#1898) (038ac22)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • skills: wire --permission-mode + --effort dispatch envelope (#1855) (832d4c8)
  • skills: yg-mcp-core consumers across brainstorm/assess/memory (closes #1886, #1887, #1888) (#1889) (1473f05)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • #1525: bump invocable caps + dev argument-hint (#1535) (9709025)
  • #1525: finish real implementations — close 4 vapor gaps (#1538) (f48821f)
  • add the same step-level env block to Step 1 too. (d79dbb5)
  • brainstorm: restore 81 lines of headroom under cap (#1578) (75ba909)
  • brainstorm: trim SKILL.md to 500-line cap (unblocks 4 PRs) (#1576) (247943a)
  • build: preflight check for missing node_modules (closes #1899) (#1904) (431c66e)
  • cc-watch: normalize double-dash bullets + --retry-failed flag (#1614) (4a9f52c)
  • cc-watch: silent-skip + reissue flag + LLM retry (#1730) (78aa054)
  • cc-watch: skip cc-triage for versions below supported floor (#1739) (#1740) (5e551a6)
  • ci-sentinel: drop --bare (auth broken), bump budget 500k→1M (#1857) (42ee3ab)
  • ci-sentinel: file-backed apiKeyHelper + safer smoke test (#1859) (25cfbfc)
  • ci-sentinel: install Claude Code on the runner (#1856) (b723901)
  • ci-sentinel: wire apiKeyHelper auth + shared --bare gotcha rule (#1858) (1bd3d19)
  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add GH_TOKEN to Snapshot step (#1598) (d79dbb5)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: bump token-overhead + skill-length caps + add missing frontmatter (#1865) (9951ffd)
  • ci: cc-triage drops --bare so CLAUDE_CODE_OAUTH_TOKEN is honored (#1629) (fbb3199)
  • ci: cc-triage uses opus alias + logs stdout on failure (#1633) (311a514)
  • ci: cc-watch PR uses PAT + version-stamped title (#1685 #1697) (#1698) (fbd1728)
  • ci: cc-watch uses GH App token (#1735) (#1736) (e726b8d)
  • ci: clean up claude-release-watch on: block + Title Case name (#1594) (d686e91)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • ci: enable auto-merge on cron-generated snapshot PRs (#1603) (cfea028)
  • ci: heredoc breaks YAML + secrets context not in if: (#1595) (649d889)
  • ci: install CC native binary explicitly when --ignore-scripts (#1622) (c6fb9c3)
  • ci: install claude CLI + PR-based snapshot commit (#1596) (dc19147)
  • ci: persist-credentials=false on checkout (#1599) (8d3b728)
  • ci: re-register cc-release-watch workflow_dispatch trigger (#1592) (2ed4e7f)
  • ci: regenerate docs/site lockfile on-the-fly, delete committed stub (#1631) (7f6f53b)
  • ci: release-video composition paths — strip fictional folder prefix (#1432) (4e639f2)
  • ci: rename cc-release-watch.yml → claude-release-watch.yml (#1593) (b89bd7e)
  • ci: scope GH_TOKEN to per-step (peter-evans needs clean git env) (#1597) (6eb594f)
  • ci: Scorecard alerts + full GitHub App migration (retire PAT) (#1426) (8400096)
  • ci: skip pre-push version check on feat/fix/perf/refactor branches (#1458) (065454d)
  • demos: bump zod to ^4.3.6 (unblock Release Video pipeline) (#1430) (8f26d0a)
  • deps: override fast-uri to ^3.1.2 for CVE-2026-6321/6322 (#1687) (#1688) (6ea510e)
  • deps: override hono / ip-address / express-rate-limit for CVEs (#1624) (95acb78)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync stale skill/agent/hook counts in fumadocs site (#1516) (f4fb5ad)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: #1795 — ORK_ASK_FALLBACK=text picker stall mitigation (#1798) (19fbc6e)
  • hooks: allow SessionStart + PostCompact in output-guard (#1234) (#1822) (921e113)
  • hooks: cap agent-watchdog at 24h to reap zombies (#1882) (#1896) (a03e5cb)
  • hooks: close JSON-leak holes that mkdir literal {...} dirs (#1606) (f93e88d)
  • hooks: envelope input guard at WorktreeCreate.name (#1826) (#1837) (9ac4e75)
  • hooks: isProtectedBranch treats empty string as not-protected (#1425) (b4f9b40)
  • hooks: kill #1794 — WorktreeCreate output shape + dispatcher guard (#1797) (3e6d164)
  • hooks: preserve /rename via sessionTitle delta-detect (#1779) (2cb8ec6)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: titler double-prefix — strip leading ork: before wrap (#1789) (2abf7e0)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • m130: repair cron pipeline + harden stamper + fill triage test gap (#1591) (99ab20d)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • release: SBOM/attestation skip + changelog-to-props parser (#1428) (1159475)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: override ws to ^8.20.1 — patches GHSA-58qx-3vcg-4xpx (#1879) (ab159cd)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skill: port /ci-debug into ork (cross-skill-ref alignment) (#1843) (7068d71)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)
  • tests: isolate stop-uncommitted-check fixtures (closes #1917) (#1918) (6eac2a9)
  • ultrareview: label-gate + fix broken TARGET arg + suppress empty comment (#1569) (c8097f5)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • agents: recategorize all 37 agents — 7 readable categories (#1523) (ae9c650)
  • claude: replace static counts with plugin-details pointer (D2) (#1777) (e04ba7d)
  • release: add v7.86 release-window slideshow + infographic (#1758) (c5b00e8)
  • release: add v7.86 release-window slideshow + infographic (#1776) (6d673c1)
  • skills: codify manual-worktree pattern in /ork:cover (closes platform#3224) (#1835) (523b39d)
  • skills: deepen @json-render/directives coverage in skill body (#1755) (5de3d5b)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)
  • tests: drop fragile counts from cc-version-matrix (#1761) (f2a29bd)

Performance

  • hooks: migrate agentation-context → SessionStart (M104 PR-B) (#1827) (cfa0984)
  • hooks: migrate ask-fallback-injector → SessionStart (M104 PR-A) (#1825) (34d80de)

CI/CD

  • 50 SUCCESS / 3 FAILURE — all 3 failures pre-existing on origin/main, tracked in #1865. (2cbf5e2)
  • actionlint enforcement (pre-commit + CI workflow) (#1601) (cf3bcaf)
  • bump actions/cache from 5.0.4 to 5.0.5 (#1421) (b548f86)
  • bump actions/checkout from 4 to 6 (#1851) (9e6dd22)
  • bump actions/checkout from 5 to 6 (#1420) (4c18de7)
  • bump actions/create-github-app-token from 3.1.1 to 3.2.0 (#1850) (a861187)
  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 4.4.0 to 6.4.0 (#1521) (08140ff)
  • bump actions/setup-node from 4.4.0 to 6.4.0 (#1612) (ce35dd9)
  • bump actions/setup-node from 5 to 6 (#1423) (77105cc)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 4.6.2 to 7.0.1 (#1854) (bb76cbe)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.101 to 1.0.107 (#1519) (23d2a7c)
  • bump anthropics/claude-code-action from 1.0.107 to 1.0.111 (#1611) (6cfd862)
  • bump anthropics/claude-code-action from 1.0.111 to 1.0.119 (#1743) (c672ae8)
  • bump anthropics/claude-code-action from 1.0.119 to 1.0.123 (#1852) (88b4134)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump anthropics/claude-code-action from 1.0.93 to 1.0.101 (#1422) (2b0bfec)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • bump github/codeql-action from 4.35.1 to 4.35.2 (#1522) (f5fdc20)
  • bump github/codeql-action from 4.35.2 to 4.35.3 (#1609) (fe98da4)
  • bump github/codeql-action from 4.35.3 to 4.35.4 (#1744) (f99744e)
  • bump github/codeql-action from 4.35.4 to 4.35.5 (#1853) (d889da6)
  • bump googleapis/release-please-action from 4.4.0 to 4.4.1 (#1419) (ec12455)
  • bump googleapis/release-please-action from 4.4.1 to 5.0.0 (#1520) (7005750)
  • bump peter-evans/create-pull-request from 7.0.5 to 8.1.1 (#1610) (d172010)
  • bump the npm_and_yarn group across 2 directories with 3 updates (#1630) (2e6010c)
  • playground-gate: allow-list cron snapshot + orchestkit-release-bot (#1820) (7069d57)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.93.1] - 2026-05-21

Fixed

  • tests: stop-uncommitted-check.test.ts fixtures no longer corrupt the parent worktree's git config or branch ref. Without isolation env (HOME, GIT_CONFIG_GLOBAL, GIT_DIR, GIT_WORK_TREE), the test's git config user.name "Test" walked up from tmpDir and wrote to the parent's .git/config; fixture commits then landed on the active branch ref, losing real work. Closes #1917.

7.93.0 (2026-05-20)

Features

7.92.0 (2026-05-20)

Features

Bug Fixes

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.145) (#1903) (0a4374c)

Documentation

  • skills: codify manual-worktree pattern in /ork:cover (closes platform#3224) (#1835) (523b39d)

7.91.0 (2026-05-20)

Features

  • ci: cross-language parity gate (M141-6, #1807) (903d870)
  • ci: PyPI publish workflow for hook-contract-py (#1864) (204efe9)
  • hook-contract-py: scaffold PyPI sibling (M141-3, #1804) (2960185)
  • hook-contract: codegen scaffold + drift gate (M141-2 step 1, #1864) (b17fb03)
  • hook-contract: HMAC signing protocol + impl (M141-4, #1805) (#1878) (762ab0f)
  • hook-contract: per-event payload schemas (M141-2 step 3, #1864) (38eeb28)
  • hook-contract: scaffold @orchestkit/hook-contract + setup token enforcement (M141-1) (2cbf5e2)
  • skill: /ork:ci-sentinel — hourly CI-failure classifier (#1842) (5ce3df1)
  • skill: /ork:swarm-migrate — cross-repo migration coordinator (#1839) (20c54f3)
  • skills: wire --permission-mode + --effort dispatch envelope (#1855) (832d4c8)
  • skills: yg-mcp-core consumers across brainstorm/assess/memory (closes #1886, #1887, #1888) (#1889) (1473f05)

Bug Fixes

  • ci-sentinel: drop --bare (auth broken), bump budget 500k→1M (#1857) (42ee3ab)
  • ci-sentinel: file-backed apiKeyHelper + safer smoke test (#1859) (25cfbfc)
  • ci-sentinel: install Claude Code on the runner (#1856) (b723901)
  • ci-sentinel: wire apiKeyHelper auth + shared --bare gotcha rule (#1858) (1bd3d19)
  • ci: bump token-overhead + skill-length caps + add missing frontmatter (#1865) (9951ffd)
  • hooks: envelope input guard at WorktreeCreate.name (#1826) (#1837) (9ac4e75)
  • security: override ws to ^8.20.1 — patches GHSA-58qx-3vcg-4xpx (#1879) (ab159cd)
  • skill: port /ci-debug into ork (cross-skill-ref alignment) (#1843) (7068d71)

Miscellaneous

  • 7.91: adopt CC 2.1.142 (M123 P0 bundle) (#1836) (ce6f429)
  • cc-watch: snapshot upstream CHANGELOG (2.1.141) (#1829) (a7f9e87)
  • cc-watch: snapshot upstream CHANGELOG (2.1.142) (#1831) (eb7761c)
  • cc-watch: snapshot upstream CHANGELOG (2.1.143) (#1838) (bddeb69)
  • cc-watch: snapshot upstream CHANGELOG (2.1.143) (#1841) (920f1d2)
  • cc-watch: snapshot upstream CHANGELOG (2.1.144) (#1876) (7a37ae0)
  • skills: kill ork:agents-view, document parallel primitives (#1890) (dbd06cb)
  • workflows: scope GITHUB_TOKEN to least privilege (6 CodeQL highs) (#1880) (fda035c)

Performance

  • hooks: migrate agentation-context → SessionStart (M104 PR-B) (#1827) (cfa0984)

CI/CD

  • 50 SUCCESS / 3 FAILURE — all 3 failures pre-existing on origin/main, tracked in #1865. (2cbf5e2)
  • bump actions/checkout from 4 to 6 (#1851) (9e6dd22)
  • bump actions/create-github-app-token from 3.1.1 to 3.2.0 (#1850) (a861187)
  • bump actions/upload-artifact from 4.6.2 to 7.0.1 (#1854) (bb76cbe)
  • bump anthropics/claude-code-action from 1.0.119 to 1.0.123 (#1852) (88b4134)
  • bump github/codeql-action from 4.35.4 to 4.35.5 (#1853) (d889da6)

7.90.0 (2026-05-14)

Features

  • implement: codify manual-worktree for parallel agents (#1824) (a0b80b7)

Bug Fixes

Performance

  • hooks: migrate ask-fallback-injector → SessionStart (M104 PR-A) (#1825) (34d80de)

7.89.1 (2026-05-13)

CI/CD

  • playground-gate: allow-list cron snapshot + orchestkit-release-bot (#1820) (7069d57)

7.89.0 (2026-05-13)

Features

7.88.1 (2026-05-13)

Miscellaneous

  • deps: npm audit fix in orchestkit-demos + src/mcp-server — final 2 high GHSAs (#1813) (2fb455b)

7.88.0 (2026-05-13)

Features

  • hooks: deferred worktree advisory consumer — PR-2 #1794 follow-up (#1801) (f017d9e)

Miscellaneous

  • deps: npm audit fix in docs/site — 15 GHSAs resolved (#1810) (3428ea1)

7.87.0 (2026-05-12)

Features

  • hooks/lib: dual-write analytics to yonatan-hq platform (#1799) (7070412)

Bug Fixes

7.86.8 (2026-05-12)

Bug Fixes

  • hooks: preserve /rename via sessionTitle delta-detect (#1779) (2cb8ec6)
  • hooks: titler double-prefix — strip leading ork: before wrap (#1789) (2abf7e0)

Miscellaneous

  • adopt: M139 Bundle A — close the M137 loop (#1786) (79623c0)
  • adopt: M140 Bundle B — make /goal real (#1793) (dca28a5)

Documentation

  • claude: replace static counts with plugin-details pointer (D2) (#1777) (e04ba7d)
  • release: add v7.86 release-window slideshow + infographic (#1776) (6d673c1)

7.86.7 (2026-05-12)

Miscellaneous

Documentation

  • release: add v7.86 release-window slideshow + infographic (#1758) (c5b00e8)

Code Refactoring

  • tests: drop fragile counts from cc-version-matrix (#1761) (f2a29bd)

Unreleased

Changed

  • mcp/memory: Memory graph dir moves from user-global to ${CLAUDE_PROJECT_DIR}/.claude/memory/. Resolved by CC 2.1.139 stdio MCP env injection — no manual path expansion needed. Existing entries auto-discovered; manual move possible by relocating the prior memory.json into the new project-scoped path.

Added

  • skill: ork:agents-view — Research Preview wrapper around claude agents and claude plugin details ork for parallel-agent observability and per-session cost projection (CC 2.1.139+).

7.86.6 (2026-05-11)

Miscellaneous

  • labs: thorough pin sync 2026-05 (3 skills) (#1753) (11a6984)

Documentation

  • skills: deepen @json-render/directives coverage in skill body (#1755) (5de3d5b)

7.86.5 (2026-05-11)

Miscellaneous

7.86.4 (2026-05-11)

Miscellaneous

  • adopt: M132 Group F CC 2.1.133 OAuth+session+network (#1747) (ad12153)

7.86.3 (2026-05-11)

CI/CD

  • bump anthropics/claude-code-action from 1.0.111 to 1.0.119 (#1743) (c672ae8)

7.86.2 (2026-05-11)

CI/CD

  • bump github/codeql-action from 4.35.3 to 4.35.4 (#1744) (f99744e)

7.86.1 (2026-05-10)

Bug Fixes

Miscellaneous

  • cc-watch: snapshot upstream CHANGELOG (2.1.123) (#1741) (8f9099e)

7.86.0 (2026-05-10)

Features

Bug Fixes

Miscellaneous

7.85.0 (2026-05-07)

Features

  • ASCII Design System + status glyph vocab + preview audit + drift fixes (#1619) (1770365)

Miscellaneous

7.84.0 (2026-05-07)

Features

  • cc: bolder floor bump 2.1.125 → 2.1.132 (#1623) (2292524)

Bug Fixes

  • ci: cc-triage drops --bare so CLAUDE_CODE_OAUTH_TOKEN is honored (#1629) (fbb3199)
  • ci: cc-triage uses opus alias + logs stdout on failure (#1633) (311a514)
  • ci: install CC native binary explicitly when --ignore-scripts (#1622) (c6fb9c3)
  • ci: regenerate docs/site lockfile on-the-fly, delete committed stub (#1631) (7f6f53b)
  • deps: override hono / ip-address / express-rate-limit for CVEs (#1624) (95acb78)

CI/CD

  • bump the npm_and_yarn group across 2 directories with 3 updates (#1630) (2e6010c)

7.83.2 (2026-05-05)

CI/CD

  • bump actions/setup-node from 4.4.0 to 6.4.0 (#1612) (ce35dd9)
  • bump anthropics/claude-code-action from 1.0.107 to 1.0.111 (#1611) (6cfd862)

7.83.1 (2026-05-05)

CI/CD

  • bump github/codeql-action from 4.35.2 to 4.35.3 (#1609) (fe98da4)
  • bump peter-evans/create-pull-request from 7.0.5 to 8.1.1 (#1610) (d172010)

7.83.0 (2026-05-05)

Features

  • cc: adopt CC 2.1.128 + bump floor to 2.1.125 (#1616) (4686952)

Bug Fixes

  • cc-watch: normalize double-dash bullets + --retry-failed flag (#1614) (4a9f52c)

Miscellaneous

7.82.1 (2026-05-03)

Miscellaneous

7.82.0 (2026-05-03)

Features

  • m127: bundle Vercel Labs 2026-04-29 adoption (5 issues) (#1604) (ac9ec5b)

Bug Fixes

  • hooks: close JSON-leak holes that mkdir literal {...} dirs (#1606) (f93e88d)

7.81.1 (2026-05-03)

Bug Fixes

  • ci: enable auto-merge on cron-generated snapshot PRs (#1603) (cfea028)

Miscellaneous

7.81.0 (2026-05-02)

Features

Bug Fixes

  • add the same step-level env block to Step 1 too. (d79dbb5)
  • ci: add GH_TOKEN to Snapshot step (#1598) (d79dbb5)
  • ci: clean up claude-release-watch on: block + Title Case name (#1594) (d686e91)
  • ci: heredoc breaks YAML + secrets context not in if: (#1595) (649d889)
  • ci: install claude CLI + PR-based snapshot commit (#1596) (dc19147)
  • ci: persist-credentials=false on checkout (#1599) (8d3b728)
  • ci: re-register cc-release-watch workflow_dispatch trigger (#1592) (2ed4e7f)
  • ci: rename cc-release-watch.yml → claude-release-watch.yml (#1593) (b89bd7e)
  • ci: scope GH_TOKEN to per-step (peter-evans needs clean git env) (#1597) (6eb594f)
  • m130: repair cron pipeline + harden stamper + fill triage test gap (#1591) (99ab20d)

CI/CD

  • actionlint enforcement (pre-commit + CI workflow) (#1601) (cf3bcaf)

7.80.0 (2026-05-02)

Features

  • cc: adopt CC 2.1.122 + 2.1.126 features (M128) (#1585) (372b9d0)

Miscellaneous

  • m129: fix lossy path decode + add OTEL panel fixture tests (#1588) (1f30144)

7.79.1 (2026-05-01)

Bug Fixes

  • brainstorm: restore 81 lines of headroom under cap (#1578) (75ba909)

7.79.0 (2026-05-01)

Features

  • hooks: local-dev hygiene bundle — 3 advisory hooks (#1574) (dc446a1)

Miscellaneous

  • deps: bump notebooklm-mcp-cli 0.5.25 → 0.6.1 (2026-04-28) (#1575) (3ba2cc5)

7.78.0 (2026-05-01)

Features

  • doctor: wire agent-browser doctor --json (#1559) (#1573) (399b030)
  • M127: cost-hygiene A1 — CI fork-subagent + slash-only skills (#1572) (1d947a9)
  • M128: monitor quickwins — manifest source, debounce, docs (#1570) (fdc2166)

Bug Fixes

  • brainstorm: trim SKILL.md to 500-line cap (unblocks 4 PRs) (#1576) (247943a)
  • ultrareview: label-gate + fix broken TARGET arg + suppress empty comment (#1569) (c8097f5)

Miscellaneous

  • M126: re-stamp component counts across docs (#1567) (4d9a7fd)

7.77.0 (2026-04-29)

Features

  • M126: secret-handler PostToolUse hook (#1543, Bundle C — last of M126) (#1553) (8cf8448)

Miscellaneous

  • deps: bump Vercel Labs upstreams to 2026-04-29 latest (#1558) (bc582e5)
  • M126 followups — close 5 audit-driven gaps from /ork:assess (#1556) (fb5b563)

7.76.0 (2026-04-28)

Features

7.75.0 (2026-04-28)

Features

7.74.0 (2026-04-27)

Features

  • #1525: /ork:dev gains scripts, rules, test-cases (#1536) (cb30501)

Bug Fixes

  • #1525: finish real implementations — close 4 vapor gaps (#1538) (f48821f)

7.73.0 (2026-04-27)

Features

  • M125: Lane A — labs version watcher + emulate auto-discovery (#1532) (a53f342)
  • M125: Lane B — /ork:dev lab-stack + auto-expect + ARIA snapshots (#1534) (1e626fd)

Bug Fixes

  • #1525: bump invocable caps + dev argument-hint (#1535) (9709025)

7.72.1 (2026-04-27)

Documentation

  • agents: recategorize all 37 agents — 7 readable categories (#1523) (ae9c650)

CI/CD

  • bump actions/setup-node from 4.4.0 to 6.4.0 (#1521) (08140ff)
  • bump anthropics/claude-code-action from 1.0.101 to 1.0.107 (#1519) (23d2a7c)
  • bump github/codeql-action from 4.35.1 to 4.35.2 (#1522) (f5fdc20)
  • bump googleapis/release-please-action from 4.4.1 to 5.0.0 (#1520) (7005750)

7.72.0 (2026-04-27)

Features

  • hooks,M119: dogfood cluster — zones, cache-aware, outcome (#1515) (8641214)

7.71.2 (2026-04-26)

Bug Fixes

  • docs: sync stale skill/agent/hook counts in fumadocs site (#1516) (f4fb5ad)

7.71.1 (2026-04-26)

Miscellaneous

7.71.0 (2026-04-26)

Features

7.70.0 (2026-04-26)

Features

  • M122: adopt CC 2.1.118 + 2.1.119 across stack (#1507) (9fad64a)

Miscellaneous

  • mcp: re-audit MCP version matrix (2026-04-26 refresh) (#1509) (d241758)

7.69.0 (2026-04-25)

Features

  • #117: close M117 — doctor MCP pinning + hook property tests (#1496) (6adc493)

7.68.0 (2026-04-25)

Features

  • skills,#117: M117 adoption closeout — Monitor + Push + OTEL + cron (#1494) (24cf12d)

7.67.0 (2026-04-23)

Features

7.66.0 (2026-04-23)

Features

  • hooks,#1478: content-aware token estimator — replace char/4 (#1484) (782677c)
  • hooks,#1478: coverage hardening — close all P1 + P2 audit gaps (#1485) (b1e7757)
  • hooks: M119 usage hygiene — pre-compact prompt + 150k context warn (#1474) (39507eb)

7.65.1 (2026-04-22)

Bug Fixes

  • ci: skip pre-push version check on feat/fix/perf/refactor branches (#1458) (065454d)

Miscellaneous

7.65.0 (2026-04-22)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • ci: de-noise PR rollup signals + auto-sync version files (Closes #1403–#1407) (#1409) (5ed87b3)
  • ci: pre-push gate + commit-msg linter + token-budget warn tier (#1411) (77dc14c)
  • ci: Road to 10 Wave 3 — agent registry + npm audit + backlog admin (#1418) (f902098)
  • claude-design: Bet A — closed-loop pipeline (Milestone #113) (#1390) (f93fa11)
  • claude-design: Bet A polish — schema rewrite + mdx-compile guard (Closes #1400, #1401) (#1402) (e97e18c)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.110 full adoption — matrix, skills, doctor (#1349) (0d166e9)
  • compat: CC 2.1.111 + Opus 4.7 full adoption (#1376) (425a7f7)
  • compat: CC 2.1.117 adoption + milestone #117 Tier A closeout (#1454) (b7c5898)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • docs: redesign landing page from Claude Design handoff (Closes #1247) (#1399) (1005354)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks,evals: CC 2.1.116 enforcement + eval coverage (#1437) (7f73d05)
  • hooks,skills: MS#117 Bundles 1+2 — integration harness + wiring (#1456) (082b1da)
  • hooks: CC 2.1.113 adoption — native binary, find/rm hardening, deniedDomains (#1388) (104226e)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • hooks: Usage-Driven Hardening bundle — 5 analytics hooks + batch rule (#1378) (7c5a3d1)
  • hooks: Usage-Driven Hardening completion — milestone close-out (3 issues) (#1380) (e23be85)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: notebooklm v0.5.25 full adoption (#1352) (f49cf3d)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • ci: release-video composition paths — strip fictional folder prefix (#1432) (4e639f2)
  • ci: Scorecard alerts + full GitHub App migration (retire PAT) (#1426) (8400096)
  • demos: bump zod to ^4.3.6 (unblock Release Video pipeline) (#1430) (8f26d0a)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: isProtectedBranch treats empty string as not-protected (#1425) (b4f9b40)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • release: SBOM/attestation skip + changelog-to-props parser (#1428) (1159475)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/cache from 5.0.4 to 5.0.5 (#1421) (b548f86)
  • bump actions/checkout from 5 to 6 (#1420) (4c18de7)
  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 5 to 6 (#1423) (77105cc)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump anthropics/claude-code-action from 1.0.93 to 1.0.101 (#1422) (2b0bfec)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • bump googleapis/release-please-action from 4.4.0 to 4.4.1 (#1419) (ec12455)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.64.0] - 2026-04-22

Added

  • CC 2.1.117 matrix entries (14 features, ending at opus_47_context_window_fix). MIN_CC_VERSION bumped 2.1.116 → 2.1.117. Closes #1441.
  • Doc-table backfill in version-compatibility.md for all 14 new entries.

Changed

  • Agent mcpServers: audit for CC 2.1.117's main-thread firing behavior: 37 agents, 30 with non-empty mcpServers:, 8 unique MCP sets — all intentional. No agent changes required (same outcome as 2.1.116 hooks audit).
  • Effort-default messaging updated: on CC ≥ 2.1.117, Pro/Max Opus 4.6 + Sonnet 4.6 now default to high (was medium).

Fixed

  • CHANGELOG [7.62.0] ghost entry removed — no v7.62.0 tag was ever cut; content shipped as v7.63.0 via release-please. Closes #1440.

Verified / No-op

  • Issue #1444 (hook backfill for StopFailure/PostCompact/WorktreeCreate/WorktreeRemove): no action — all 4 hooks already registered in hooks.json. Brainstorm misread the matrix.
  • Issue #1449 (CwdChanged/FileChanged hooks): no action — both already registered.
  • Issue #1453 (Setup/SubagentStart/InstructionsLoaded/TaskCreated hooks): no action — all 4 already registered.
  • Issue #1445 (deprecated-ref sweep): no action — all apparent hits are either legitimate historical documentation or false positives (node/tag in LangGraph, GitHub release URLs).

7.63.1 (2026-04-22)

Miscellaneous

  • cleanup: fix skill count drift + refresh What's New (#1442) (e28e7da)

7.63.0 (2026-04-22)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • ci: de-noise PR rollup signals + auto-sync version files (Closes #1403–#1407) (#1409) (5ed87b3)
  • ci: pre-push gate + commit-msg linter + token-budget warn tier (#1411) (77dc14c)
  • ci: Road to 10 Wave 3 — agent registry + npm audit + backlog admin (#1418) (f902098)
  • claude-design: Bet A — closed-loop pipeline (Milestone #113) (#1390) (f93fa11)
  • claude-design: Bet A polish — schema rewrite + mdx-compile guard (Closes #1400, #1401) (#1402) (e97e18c)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.110 full adoption — matrix, skills, doctor (#1349) (0d166e9)
  • compat: CC 2.1.111 + Opus 4.7 full adoption (#1376) (425a7f7)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • docs: redesign landing page from Claude Design handoff (Closes #1247) (#1399) (1005354)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks,evals: CC 2.1.116 enforcement + eval coverage (#1437) (7f73d05)
  • hooks: CC 2.1.113 adoption — native binary, find/rm hardening, deniedDomains (#1388) (104226e)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • hooks: Usage-Driven Hardening bundle — 5 analytics hooks + batch rule (#1378) (7c5a3d1)
  • hooks: Usage-Driven Hardening completion — milestone close-out (3 issues) (#1380) (e23be85)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: notebooklm v0.5.25 full adoption (#1352) (f49cf3d)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • ci: release-video composition paths — strip fictional folder prefix (#1432) (4e639f2)
  • ci: Scorecard alerts + full GitHub App migration (retire PAT) (#1426) (8400096)
  • demos: bump zod to ^4.3.6 (unblock Release Video pipeline) (#1430) (8f26d0a)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: isProtectedBranch treats empty string as not-protected (#1425) (b4f9b40)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • release: SBOM/attestation skip + changelog-to-props parser (#1428) (1159475)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/cache from 5.0.4 to 5.0.5 (#1421) (b548f86)
  • bump actions/checkout from 5 to 6 (#1420) (4c18de7)
  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 5 to 6 (#1423) (77105cc)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump anthropics/claude-code-action from 1.0.93 to 1.0.101 (#1422) (2b0bfec)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • bump googleapis/release-please-action from 4.4.0 to 4.4.1 (#1419) (ec12455)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

7.61.0 (2026-04-20)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • ci: de-noise PR rollup signals + auto-sync version files (Closes #1403–#1407) (#1409) (5ed87b3)
  • ci: pre-push gate + commit-msg linter + token-budget warn tier (#1411) (77dc14c)
  • ci: Road to 10 Wave 3 — agent registry + npm audit + backlog admin (#1418) (f902098)
  • claude-design: Bet A — closed-loop pipeline (Milestone #113) (#1390) (f93fa11)
  • claude-design: Bet A polish — schema rewrite + mdx-compile guard (Closes #1400, #1401) (#1402) (e97e18c)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.110 full adoption — matrix, skills, doctor (#1349) (0d166e9)
  • compat: CC 2.1.111 + Opus 4.7 full adoption (#1376) (425a7f7)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • docs: redesign landing page from Claude Design handoff (Closes #1247) (#1399) (1005354)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks: CC 2.1.113 adoption — native binary, find/rm hardening, deniedDomains (#1388) (104226e)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • hooks: Usage-Driven Hardening bundle — 5 analytics hooks + batch rule (#1378) (7c5a3d1)
  • hooks: Usage-Driven Hardening completion — milestone close-out (3 issues) (#1380) (e23be85)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: notebooklm v0.5.25 full adoption (#1352) (f49cf3d)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • ci: release-video composition paths — strip fictional folder prefix (#1432) (4e639f2)
  • ci: Scorecard alerts + full GitHub App migration (retire PAT) (#1426) (8400096)
  • demos: bump zod to ^4.3.6 (unblock Release Video pipeline) (#1430) (8f26d0a)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: isProtectedBranch treats empty string as not-protected (#1425) (b4f9b40)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • release: SBOM/attestation skip + changelog-to-props parser (#1428) (1159475)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/cache from 5.0.4 to 5.0.5 (#1421) (b548f86)
  • bump actions/checkout from 5 to 6 (#1420) (4c18de7)
  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 5 to 6 (#1423) (77105cc)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump anthropics/claude-code-action from 1.0.93 to 1.0.101 (#1422) (2b0bfec)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • bump googleapis/release-please-action from 4.4.0 to 4.4.1 (#1419) (ec12455)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.60.1] - 2026-04-21

Fixed

  • Release Video composition paths. release-video.yml constructed composition IDs like Production/Landscape-16x9/ReleaseNotes/RN-v${SAFE_VERSION} with a fictional folder prefix — Remotion Studio's folder tree is cosmetic and the CLI resolves bare id attributes only. Additionally no per-version RN-v* comps get registered (the only one, RN-v602, is from v6.0.2). Both the bespoke path and the Templates/TPL-ReleaseNotes fallback had this bug, so the workflow has silently failed every release since its creation. Simplified to bare TPL-ReleaseNotes (already registered) for landscape and registered a new TPL-ReleaseNotesSquare template in orchestkit-demos/src/Root.tsx for the square render. Verified locally against v7.60.0 CHANGELOG props — both render cleanly (1.7 MB landscape, 1.1 MB square for 30-frame smoke render).

7.60.0 (2026-04-20)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • ci: de-noise PR rollup signals + auto-sync version files (Closes #1403–#1407) (#1409) (5ed87b3)
  • ci: pre-push gate + commit-msg linter + token-budget warn tier (#1411) (77dc14c)
  • ci: Road to 10 Wave 3 — agent registry + npm audit + backlog admin (#1418) (f902098)
  • claude-design: Bet A — closed-loop pipeline (Milestone #113) (#1390) (f93fa11)
  • claude-design: Bet A polish — schema rewrite + mdx-compile guard (Closes #1400, #1401) (#1402) (e97e18c)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.110 full adoption — matrix, skills, doctor (#1349) (0d166e9)
  • compat: CC 2.1.111 + Opus 4.7 full adoption (#1376) (425a7f7)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • docs: redesign landing page from Claude Design handoff (Closes #1247) (#1399) (1005354)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks: CC 2.1.113 adoption — native binary, find/rm hardening, deniedDomains (#1388) (104226e)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • hooks: Usage-Driven Hardening bundle — 5 analytics hooks + batch rule (#1378) (7c5a3d1)
  • hooks: Usage-Driven Hardening completion — milestone close-out (3 issues) (#1380) (e23be85)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: notebooklm v0.5.25 full adoption (#1352) (f49cf3d)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • ci: Scorecard alerts + full GitHub App migration (retire PAT) (#1426) (8400096)
  • demos: bump zod to ^4.3.6 (unblock Release Video pipeline) (#1430) (8f26d0a)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: isProtectedBranch treats empty string as not-protected (#1425) (b4f9b40)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • release: SBOM/attestation skip + changelog-to-props parser (#1428) (1159475)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/cache from 5.0.4 to 5.0.5 (#1421) (b548f86)
  • bump actions/checkout from 5 to 6 (#1420) (4c18de7)
  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 5 to 6 (#1423) (77105cc)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump anthropics/claude-code-action from 1.0.93 to 1.0.101 (#1422) (2b0bfec)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • bump googleapis/release-please-action from 4.4.0 to 4.4.1 (#1419) (ec12455)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.59.1] - 2026-04-21

Fixed

  • orchestkit-demos zod peer-dep conflict. @remotion/zod-types@4.0.447 added a peer zod@"4.3.6" requirement, which npm ERESOLVE'd against the pinned zod@3.22.3. The Release Video workflow had been failing npm install on every release since — including v7.58.0 and v7.59.0. Bumped to zod@^4.3.6; existing usage (z.object, z.string, z.number, z.array, z.union, z.record, z.infer) is API-compatible v3↔v4, tsc --noEmit passes clean.

7.59.0 (2026-04-20)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • ci: de-noise PR rollup signals + auto-sync version files (Closes #1403–#1407) (#1409) (5ed87b3)
  • ci: pre-push gate + commit-msg linter + token-budget warn tier (#1411) (77dc14c)
  • ci: Road to 10 Wave 3 — agent registry + npm audit + backlog admin (#1418) (f902098)
  • claude-design: Bet A — closed-loop pipeline (Milestone #113) (#1390) (f93fa11)
  • claude-design: Bet A polish — schema rewrite + mdx-compile guard (Closes #1400, #1401) (#1402) (e97e18c)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.110 full adoption — matrix, skills, doctor (#1349) (0d166e9)
  • compat: CC 2.1.111 + Opus 4.7 full adoption (#1376) (425a7f7)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • docs: redesign landing page from Claude Design handoff (Closes #1247) (#1399) (1005354)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks: CC 2.1.113 adoption — native binary, find/rm hardening, deniedDomains (#1388) (104226e)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • hooks: Usage-Driven Hardening bundle — 5 analytics hooks + batch rule (#1378) (7c5a3d1)
  • hooks: Usage-Driven Hardening completion — milestone close-out (3 issues) (#1380) (e23be85)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: notebooklm v0.5.25 full adoption (#1352) (f49cf3d)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • ci: Scorecard alerts + full GitHub App migration (retire PAT) (#1426) (8400096)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: isProtectedBranch treats empty string as not-protected (#1425) (b4f9b40)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • release: SBOM/attestation skip + changelog-to-props parser (#1428) (1159475)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/cache from 5.0.4 to 5.0.5 (#1421) (b548f86)
  • bump actions/checkout from 5 to 6 (#1420) (4c18de7)
  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 5 to 6 (#1423) (77105cc)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump anthropics/claude-code-action from 1.0.93 to 1.0.101 (#1422) (2b0bfec)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • bump googleapis/release-please-action from 4.4.0 to 4.4.1 (#1419) (ec12455)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.58.1] - 2026-04-20

Fixed

  • release.yml SBOM + attestation skip. The tag-triggered release workflow was wrapped in if: steps.check.outputs.exists == 'false' on every step, so whenever release-please created the GitHub Release first (which it always does now that both workflows use the App token), SBOM generation, upload, and provenance attestation were silently skipped. v7.57.0 and v7.58.0 both shipped with zero release assets. Only gh release create is now gated on existence; SBOM generation, upload (--clobber for idempotency), and attestation run every time.
  • scripts/changelog-to-props.mjs parses release-please CHANGELOG format. The script only recognized ## [X.Y.Z] - YYYY-MM-DD (bump-version.sh output) and failed on every release-please release with Error: Version X.Y.Z not found in CHANGELOG.md, breaking the Release Video workflow. Now accepts both formats plus the conventional-commit section headings (Features, Bug Fixes, etc.) and strips trailing link artifacts. Added tests/unit/test-changelog-parser.sh to lock both formats in.

7.58.0 (2026-04-20)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • ci: de-noise PR rollup signals + auto-sync version files (Closes #1403–#1407) (#1409) (5ed87b3)
  • ci: pre-push gate + commit-msg linter + token-budget warn tier (#1411) (77dc14c)
  • ci: Road to 10 Wave 3 — agent registry + npm audit + backlog admin (#1418) (f902098)
  • claude-design: Bet A — closed-loop pipeline (Milestone #113) (#1390) (f93fa11)
  • claude-design: Bet A polish — schema rewrite + mdx-compile guard (Closes #1400, #1401) (#1402) (e97e18c)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.110 full adoption — matrix, skills, doctor (#1349) (0d166e9)
  • compat: CC 2.1.111 + Opus 4.7 full adoption (#1376) (425a7f7)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • docs: redesign landing page from Claude Design handoff (Closes #1247) (#1399) (1005354)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks: CC 2.1.113 adoption — native binary, find/rm hardening, deniedDomains (#1388) (104226e)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • hooks: Usage-Driven Hardening bundle — 5 analytics hooks + batch rule (#1378) (7c5a3d1)
  • hooks: Usage-Driven Hardening completion — milestone close-out (3 issues) (#1380) (e23be85)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: notebooklm v0.5.25 full adoption (#1352) (f49cf3d)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • ci: Scorecard alerts + full GitHub App migration (retire PAT) (#1426) (8400096)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: isProtectedBranch treats empty string as not-protected (#1425) (b4f9b40)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/cache from 5.0.4 to 5.0.5 (#1421) (b548f86)
  • bump actions/checkout from 5 to 6 (#1420) (4c18de7)
  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 5 to 6 (#1423) (77105cc)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump anthropics/claude-code-action from 1.0.93 to 1.0.101 (#1422) (2b0bfec)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • bump googleapis/release-please-action from 4.4.0 to 4.4.1 (#1419) (ec12455)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.57.1] - 2026-04-20

Fixed

  • Scorecard supply-chain alerts — closed all 11 open code-scanning findings.
    • Pinned-Dependencies: windows-smoke.yml pins actions/checkout@v6.0.2 and actions/setup-node@v6 to SHA; docs.yml and tests/unit/test-mdx-compile.sh replace bare npm install fallbacks with lockfile regen + npm ci.
    • Token-Permissions: release.yml, prerelease.yml, release-video.yml, channel-sync.yml, and release-please.yml now mint short-lived installation tokens from the OrchestKit Release Bot GitHub App via actions/create-github-app-token (pinned to v3.1.1 SHA). GITHUB_TOKEN stays read-only (or {} where unused). release.yml keeps id-token: write + attestations: write for attest-build-provenance (OIDC-bound, not App-bound).
    • PAT retirement: RELEASE_PLEASE_TOKEN is no longer referenced. Delete the repo secret after merge: gh secret delete RELEASE_PLEASE_TOKEN --repo yonatangross/orchestkit.
    • Commit identity alignment: prerelease.yml and channel-sync.yml configure git user as orchestkit-release-bot[bot] so commits match the token that pushes them (yields "verified" on GitHub).

7.57.0 (2026-04-20)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • ci: de-noise PR rollup signals + auto-sync version files (Closes #1403–#1407) (#1409) (5ed87b3)
  • ci: pre-push gate + commit-msg linter + token-budget warn tier (#1411) (77dc14c)
  • ci: Road to 10 Wave 3 — agent registry + npm audit + backlog admin (#1418) (f902098)
  • claude-design: Bet A — closed-loop pipeline (Milestone #113) (#1390) (f93fa11)
  • claude-design: Bet A polish — schema rewrite + mdx-compile guard (Closes #1400, #1401) (#1402) (e97e18c)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.110 full adoption — matrix, skills, doctor (#1349) (0d166e9)
  • compat: CC 2.1.111 + Opus 4.7 full adoption (#1376) (425a7f7)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • docs: redesign landing page from Claude Design handoff (Closes #1247) (#1399) (1005354)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks: CC 2.1.113 adoption — native binary, find/rm hardening, deniedDomains (#1388) (104226e)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • hooks: Usage-Driven Hardening bundle — 5 analytics hooks + batch rule (#1378) (7c5a3d1)
  • hooks: Usage-Driven Hardening completion — milestone close-out (3 issues) (#1380) (e23be85)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: notebooklm v0.5.25 full adoption (#1352) (f49cf3d)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: isProtectedBranch treats empty string as not-protected (#1425) (b4f9b40)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/cache from 5.0.4 to 5.0.5 (#1421) (b548f86)
  • bump actions/checkout from 5 to 6 (#1420) (4c18de7)
  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 5 to 6 (#1423) (77105cc)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump anthropics/claude-code-action from 1.0.93 to 1.0.101 (#1422) (2b0bfec)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • bump googleapis/release-please-action from 4.4.0 to 4.4.1 (#1419) (ec12455)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.56.3] - 2026-04-20

Fixed

  • TODO: Describe your changes here

7.56.2 (2026-04-19)

Road to 10 — Wave 3. Backlog discipline + architecture clarity + supply chain. Patch release, no user-facing changes.

Added

  • scripts/list-invocable-agents.mjs — scans src/agents/*.md, extracts YAML frontmatter, writes a table to src/agents/README.md between BEGIN/END markers. Answers the "which specific ork agent do I spawn" question that cost cycles during Lane 2.
  • src/agents/README.md — agent registry with decision tree + auto-generated table (37 agents).
  • tests/security/test-npm-audit.sh — npm audit gate across root, docs/site, src/hooks. Fails on moderate+ CVEs not in the allowlist.
  • .claude/audit-allowlist.json — empty-by-default, schema-documented. Entries carry until: dates so exceptions get revisited.

Changed

  • tests/security/run-security-tests.sh — registers the npm audit gate.
  • bin/count-components.sh, bin/validate-counts.sh, scripts/stamp-counts.sh — exclude README.md / INDEX.md / CONTRIBUTING.md when counting agents.

Project admin

  • Milestone #114 Bet B closed. Its 5 issues relabeled research. No drift signal has surfaced from shipped sites; revisit if a real case appears.

7.56.1 (2026-04-19)

Miscellaneous

  • agents: wire up emulate-engineer (close last orphan) (#1381) (42cec16)

7.56.0 (2026-04-19)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • ci: de-noise PR rollup signals + auto-sync version files (Closes #1403–#1407) (#1409) (5ed87b3)
  • ci: pre-push gate + commit-msg linter + token-budget warn tier (#1411) (77dc14c)
  • claude-design: Bet A — closed-loop pipeline (Milestone #113) (#1390) (f93fa11)
  • claude-design: Bet A polish — schema rewrite + mdx-compile guard (Closes #1400, #1401) (#1402) (e97e18c)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.110 full adoption — matrix, skills, doctor (#1349) (0d166e9)
  • compat: CC 2.1.111 + Opus 4.7 full adoption (#1376) (425a7f7)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • docs: redesign landing page from Claude Design handoff (Closes #1247) (#1399) (1005354)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks: CC 2.1.113 adoption — native binary, find/rm hardening, deniedDomains (#1388) (104226e)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • hooks: Usage-Driven Hardening bundle — 5 analytics hooks + batch rule (#1378) (7c5a3d1)
  • hooks: Usage-Driven Hardening completion — milestone close-out (3 issues) (#1380) (e23be85)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: notebooklm v0.5.25 full adoption (#1352) (f49cf3d)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

7.55.2 (2026-04-19)

Road to 10 — Wave 1: session quality gates. Patch release. No user-facing changes.

Added

  • bin/git-hooks/commit-msg: blocks commits with banned celebratory phrases ("shipped cleanly", "all green!", 🎉, etc.) and commit titles > 72 chars. Soft-warns when commit body > 50 lines. See CLAUDE.md § Tone.
  • scripts/prepare (via package.json): sets core.hooksPath bin/git-hooks automatically on npm install, so the repo's git hooks activate for every contributor without a manual step.
  • Warn tier in tests/performance/test-token-overhead.sh: emits a yellow WARN: line when any section crosses 90% of its budget but hasn't exceeded it. Gives notice BEFORE the budget blows.

Changed

  • bin/git-hooks/pre-push: extended to run the mdx-compile guard, test:agents, test:skills, test:manifests, and the token-overhead budget check. Previous gaps let this session push code that CI then rejected — these gates close the loop locally.

Why

During the session that closed milestones #113 and #116 we pushed code that CI caught as broken three times (nested-fence mdx bug, version-sync drift on pyproject.toml, 401 stub-fallback missing from the relocated guard). Each failure traced to a gate that existed but wasn't wired into the pre-push path. Wave 1 closes that gap.

7.55.1 (2026-04-19)

Lane 2 — CI/Tooling de-noise. Patch release. No user-facing changes; pure infra hygiene that pays back on every future PR.

Changed

  • .github/workflows/claude-review.yml: gated on needs-claude-review label so the workflow no longer auto-cancels on every PR and pollutes the rollup state. (#1403)
  • .github/workflows/docs.yml: lowercase build job display name changed to Docs Site Build to disambiguate from the required Build check. (#1404)
  • scripts/build-plugins.sh: invokes stamp-counts.sh before the final summary instead of in postbuild, so npm run build shows accurate counts on a single run. (#1405)
  • scripts/stamp-counts.sh: new sync_versions function reads package.json and propagates the version to pyproject.toml, manifests/ork.json, marketplace.json (top + plugins[0]), .release-please-manifest.json, and CLAUDE.md. (#1407)
  • bin/validate-counts.sh: extended with a version drift check that fails on any mismatch with package.json. (#1407)
  • tests/unit/test-mdx-compile.sh: added the same NPM 401 → analytics-stub fallback as docs.yml so the guard works on PR builds without the private GitHub Packages token. (#1406)

Moved

  • scripts/test-mdx-compile.mjs → docs/site/scripts/test-mdx-compile.mjs
  • tests/fixtures/nested-fence.mdx → docs/site/tests/fixtures/nested-fence.mdx
  • @mdx-js/mdx and remark-gfm removed from root devDependencies; added directly to docs/site/package.json where they belong. Removes 128 transitive packages from root npm install. (#1406)

7.55.0 (2026-04-19)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • claude-design: Bet A — closed-loop pipeline (Milestone #113) (#1390) (f93fa11)
  • claude-design: Bet A polish — schema rewrite + mdx-compile guard (Closes #1400, #1401) (#1402) (e97e18c)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.110 full adoption — matrix, skills, doctor (#1349) (0d166e9)
  • compat: CC 2.1.111 + Opus 4.7 full adoption (#1376) (425a7f7)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • docs: redesign landing page from Claude Design handoff (Closes #1247) (#1399) (1005354)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks: CC 2.1.113 adoption — native binary, find/rm hardening, deniedDomains (#1388) (104226e)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • hooks: Usage-Driven Hardening bundle — 5 analytics hooks + batch rule (#1378) (7c5a3d1)
  • hooks: Usage-Driven Hardening completion — milestone close-out (3 issues) (#1380) (e23be85)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: notebooklm v0.5.25 full adoption (#1352) (f49cf3d)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

7.54.0 (2026-04-19)

Bet A polish — milestone #113 close-out. Two follow-ups from the first real /ork:design-ship dogfood (#1399):

Added

  • tests/unit/test-mdx-compile.sh + scripts/test-mdx-compile.mjs — pre-build mdx-compile guard. Walks every docs/site/content/docs/**/*.mdx, compiles each via @mdx-js/mdx + remark-gfm, fails on parse errors with exact filename + line:col. Runs on npm test. Wired into tests/run-all-tests.sh under UNIT TESTS. (#1401)
  • tests/fixtures/nested-fence.mdx — counter-example fixture intentionally broken with a same-length nested fence. The guard wrapper asserts the checker REJECTS this file (TDD safety net so the guard never silently weakens). (#1401)
  • @mdx-js/mdx@^3 and remark-gfm@^4 as root devDependencies (previously only present transitively in docs/site/node_modules).
  • src/skills/CONTRIBUTING-SKILLS.md → "Nesting code fences" — new authoring rule with ❌/✅ examples. Pairs with the guard.

Changed

  • src/agents/claude-design-orchestrator.md — schema rewritten to match the actual Claude Design handoff format (gzipped tarball with README.md + chats/*.md + optional project/*.html), discovered during the #1399 dogfood. The first-iteration spec assumed a JSON manifest with components[]/design_tokens{}/asset_manifest[] — none of which exist in real bundles. Adds explicit "How to parse" section, fallback for missing project/ (incomplete bundle), and ?open_file= query-param handling. (#1400)
  • src/skills/design-import/SKILL.md — Phase 1 delegation prompt rewritten to match the new schema. Tells the orchestrator to tar -xzf first, read README + chats, then look for project/*.html. Switched outer fence to 4-backticks per the new authoring rule. (#1400)

Why

The #1399 dogfood proved the Bet A pipeline works end-to-end on a real handoff URL, but exposed two regressions: (1) a same-length nested-fence bug in the original design-import.mdx that broke every PR's docs build, and (2) the agent's expected schema diverged from reality. This release closes both gaps so the next /ork:design-ship invocation runs against a guard-protected pipeline with an accurate agent spec.

Claude Design Integration — Bet A (Closed-Loop Pipeline) — Milestone #113 complete. First-class integration with claude.ai/design: handoff bundle URL → reviewable GitHub PR in one command.

Added

  • /ork:design-import — accepts a Claude Design handoff bundle (URL or local file), reconciles tokens against the project's design system, and pipes surviving components through the existing design-to-code pipeline. Does not open a PR. (#1382)
  • /ork:design-ship — end-to-end: import → Storybook stories + Playwright tests (/ork:cover) → diff-aware browser verification (/ork:expect) → PR opened (/ork:create-pr) with the handoff URL, before/after screenshots, component decisions, and coverage delta embedded. Auto-labels claude-design. (#1387)
  • claude-design-orchestrator agent — parses handoff bundles, validates the provisional schema, dedups proposed components via component-search (Storybook MCP → 21st.dev → filesystem), tracks bundle→PR provenance in .claude/design-handoffs/<bundle_id>.json. (#1384)
  • posttool/design-import/auto-verify hook — async PostToolUse hook fires after /ork:design-import completes, emits a non-blocking nudge suggesting /ork:dogfood + /ork:expect. (#1386)
  • Docs cookbook page — docs/site/content/docs/cookbook/claude-design-handoff.mdx with the full walkthrough from prompt-in-Claude-Design to merged PR. (#1383)

Changed

  • Skill count: 103 → 105. Agent count: 36 → 37. Hook count: 178 → 180 (112 global + 46 agent-scoped + 22 skill-scoped). User-invocable commands: 23 → 25.
  • tests/performance/test-token-overhead.sh: MAX_USER_INVOCABLE_SKILLS raised 23 → 25 for the two new Claude Design commands.
  • tests/skills/structure/test-skill-md.sh: EXPECTED_USER_INVOCABLE raised 23 → 25.
  • src/hooks/src/__tests__/e2e/dispatcher-registry-wiring.test.ts: async hook count assertion raised 75 → 76.

Known limitations

  • Claude Design has no public API yet — bundles are one-shot exports. Iteration means re-exporting from claude.ai/design and re-running /ork:design-ship. The programmatic drift-sync loop is Bet B (future milestone).
  • Bundle schema is provisional — the orchestrator agent adapts to deviations and surfaces unknown fields rather than failing silently.
  • No auto-merge — skills open PRs; humans review and merge.

7.51.0 (2026-04-17)

Features

  • hooks: Usage-Driven Hardening bundle — 5 analytics hooks + batch rule (#1378) (7c5a3d1)
  • hooks: Usage-Driven Hardening completion — milestone close-out (3 issues) (#1380) (e23be85)

7.52.0 (2026-04-17)

Usage-Driven Hardening — Milestone close-out — 3 remaining issues (#1278, #1292, #1295) closed by one PR. The 9-issue milestone is now complete.

Added

  • worktree-merge-verifier (PreToolUse/Bash, Phase 6 of sync dispatcher): intercepts git worktree remove, resolves the worktree's branch via git worktree list --porcelain, compares against main/master/develop, and skips squash-merged branches (via git cherry). Advisory with concrete remediation (push or merge first). Opt-out: ORK_DISABLE_WORKTREE_VERIFIER=1. (#1278)
  • metrics-bridge (PostToolUse/catchall, async): accumulates per-session metrics into .claude/state/ork-metrics-{session_id}.json — edits, test_runs, commits, bash_calls, agent_spawns, started_at/updated_at. Flat JSON object for easy jq. Atomic writes, 1 MB cap. Runtime state gitignored. (#1295)
  • outputNotify() + wrapAt() builders (lib/output.ts + re-export from lib/common.ts): consistent hook advisory formatter with prefix + 80-char line wrapping + continuation-line indent. Options bag for maxWidth and event so the signature can grow without breaking callers. 12 unit tests. Existing hook migration deferred — additive change. (#1292)

Changed

  • Total hooks.json count: 177 → 178 (bundle totals 207 → 209). Async hooks: 74 → 75. worktree-merge-verifier runs through the existing sync-bash-dispatcher — bundle export but not a new hooks.json entry.
  • Test count assertions updated across split-bundles.test.ts, async-registry.test.ts, dispatcher-registry-wiring.test.ts.

Milestone status

Usage-Driven Hardening — 9/9 issues closed. 🎉

7.51.0 (2026-04-17)

Usage-Driven Hardening bundle — 6 issues (#1280, #1281, #1282, #1283, #1285, #1286) closed by one PR. Analytics-driven hooks that catch common friction points from the 126-session usage report.

Added

  • stale-import-detector (PostToolUse/Write|Edit): greps src + tests for import sites referencing a just-edited file's module path. Advisory additionalContext listing up to 10 references. Language-aware TS/JS and Python patterns. (#1280)
  • pre-commit-test-gate (PreToolUse/Bash, Phase 5 of sync dispatcher): records test-runner invocations to .claude/state/last-test-run.json; on git commit, compares staged-file mtimes against last recorded test run and injects advisory if stale. Never blocks. (#1281)
  • subagent-scope-auditor (SubagentStop, async): parses agent_transcript_path, extracts Write/Edit targets, compares against scope fragments derived from spawn prompt. Flags up to 8 out-of-scope edits. (#1282)
  • thrash-detector (UserPromptSubmit, async) + edit-history-tracker (PostToolUse/Write|Edit, async): tracks edit events in .claude/state/edit-history.jsonl (capped at 50, self-trimming); flags any file hit ≥3 times in the last 10 events. (#1285)
  • batch-size-governor rule in ork:implement: >10-file refactors enforce max 5 files per agent batch with tests between batches. --batch-size N override. (#1286)
  • Usage-analytics hardening rules (.claude/rules/usage-analytics-hardening.md): 4 codified patterns on sub-agents & worktrees, refactoring checklists, batch changes, and 1Password/GPG interactive-auth handling. (#1283)

Changed

  • Test count assertions updated: split-bundles.test.ts 202 → 207, async-registry.test.ts + dispatcher-registry-wiring.test.ts 70 → 74 async hooks.
  • Hook totals: 173 → 177 (110 global + 45 agent + 22 skill).

7.50.0 (2026-04-17)

Features

  • compat: CC 2.1.111 + Opus 4.7 full adoption (#1376) (425a7f7)

7.50.0 (2026-04-17)

CC 2.1.111 + Opus 4.7 full adoption — 23 issues (#1353–#1375) under milestone #112, closed by a single PR.

Added

  • xhigh effort tier across phase-based skills (brainstorm, implement, cover, verify, fix-issue, assess, audit-full). Semantics: high + one additional deepening pass. Opus 4.7 only; silently falls back to high otherwise. (#1354)
  • 17 new 2.1.111 feature-matrix entries in lib/cc-version-matrix.ts (294 → 311 total). MIN_CC_VERSION 2.1.108 → 2.1.111. (#1353)
  • /ultrareview composition in ork:review-pr — trigger-gated (diff > 500 LOC, sensitive paths, reviewer disagreement, release/hotfix labels) with voice-friendly AskUserQuestion prompt. Never fires by default; quota tracked in .claude/state/ultrareview-usage.json. (#1356)
  • Eval preflight: preflight_check_plugin_errors captures plugin_errors from stream-json init event and fails eval:full loudly on degraded plugin sets. (#1357)
  • W3C Trace Context propagation: eval runners export TRACEPARENT / TRACESTATE so every claude -p call links into Langfuse/OTel. (#1370)
  • Context-window helpers: lib/context-window.ts with getContextWindowTokens, scaleByContextWindow, tokensAsContextPct. Replaces scattered absolute-token caps. (#1355)
  • Doctor Category 14: warns when xhigh effort is configured without Opus 4.7 (silent fallback otherwise). (#1360)
  • Task-budget awareness in ork:implement: advisory budget_remaining_pct in chain state + threshold-driven behavior. (#1371)
  • Assess uncertainty output: confidence + caveats[] per dimension under xhigh effort. (#1375)
  • Directive-density test (npm run test:skills:directive-density) and skill token scorecard (npm run test:skills:tokens) — advisory by default, pair with CC 2.1.111's /skills token sort. (#1362, #1363)
  • Windows PowerShell smoke test workflow (.github/workflows/windows-smoke.yml) — runs under CLAUDE_CODE_USE_POWERSHELL_TOOL=1 on src/hooks/** changes. (#1365)
  • Model pricing: claude-opus-4-7 row added (same rates as 4.6); aliases opus → claude-opus-4-7, sonnet → claude-sonnet-4-6. Historical IDs retained for accurate stats-cache replay. (#1369)

Changed

  • Token budgets converted to %-of-context: hook-priorities.ts and token-budget-tracker.ts now read CLAUDE_MAX_CONTEXT via the shared helper. Absorbs Opus 4.7's 1.0–1.35× tokenizer drift automatically. (#1355)
  • Retry policies audited: retry-manager.ts and http-sink.ts counts retained at 3 with explicit rationale comments. No behavior change. (#1359)
  • Vision ceiling bumped to 2,576 px (~3.75 MP) in design-context-extract, design-to-code, ui-feedback. Dense dashboards, Retina screenshots, full-page desktop mockups now fit natively. (#1358)
  • Filesystem memory lean-in in ork:remember / ork:memory — layered read order (MEMORY.md index → chain handoff → MCP graph). Opus 4.7 reads fs memory more reliably than 4.6. (#1366)
  • Long-chain guidance added to CLAUDE.md: Auto mode without the old flag, scheduled-task resurrection via --resume, task budgets, fs memory strength. (#1368)
  • Permission guidance in ork:configure: since 2.1.111, read-only bash + globs no longer prompt — local Bash(ls:*) / Bash(grep:*) allows are now redundant. Points users at the built-in /less-permission-prompts skill for the prune. (#1361)
  • Plan filename guidance in ork:write-prd: CC 2.1.111 names plan files after the prompt stem. (#1367)
  • Security-skill cyber-capability note added to security-auditor and ai-safety-auditor agents: Opus 4.7 deliberately reduces cyber capabilities + ships detection safeguards; legitimate research should route through the Cyber Verification Program. (#1374)
  • Skill version bumps across 16 modified SKILL.md files — minor for semantic changes, patch for guidance-only edits. (#1372)
  • Debug env vars documented: OTEL_LOG_RAW_API_BODIES, CLAUDE_CODE_USE_POWERSHELL_TOOL, CLAUDE_MAX_CONTEXT, ENABLE_PROMPT_CACHING_1H, ORK_SKIP_PLUGIN_ERROR_CHECK in src/hooks/README.md. (#1364)

7.49.0 (2026-04-16)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.110 full adoption — matrix, skills, doctor (#1349) (0d166e9)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: notebooklm v0.5.25 full adoption (#1352) (f49cf3d)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.48.0

Added

  • TODO: Describe your changes here

7.48.0 (2026-04-16)

Features

  • compat: CC 2.1.110 full adoption — 16 matrix entries, PushNotification in 5 skills, /focus replaces Ctrl+O, session recap default-on, /tui + autoScrollEnabled in setup, doctor MCP duplicate warning, chain-patterns scheduled task recovery, Recommended tier → 2.1.110 (#1349)

7.47.0 (2026-04-15)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)
  • v7.46.0 Vercel Labs refresh + wrapper-skill modernization (#1345) (78c465e)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

7.46.0 (2026-04-15)

Features

  • skills: wrapper-skill modernization sweep (#1345) — 17 skills + expect-agent brought up-to-date with the 2026-04-15 upstream audit. P0 fixes for Stitch MCP tool names (design-to-code, design-context-extract), Playwright 1.58→1.59 + k6 v1.0 API paths (testing-e2e, testing-perf), canonical model IDs (llm-integration, multimodal-llm). P1 adoption of Figma Dev Mode MCP 16-tool surface (figma-design-handoff), Next.js 16.2.3 LTS cache variants + next-browser (react-server-components-framework), v0.app rebrand + Google Stitch + Bolt Cloud (ai-ui-generation), Pact v16 + Testcontainers (testing-integration), gh agent-task + sub-issues + gh-copilot retirement (github-operations). P2 updates across ui-components (shadcn CLI v4), storybook-testing/storybook-mcp-integration (componentsManifest rename), mcp-patterns (SDK floors 1.29/1.27 + --mcp-debug), bare-eval (stream-json), langgraph 1.2 (deferred nodes + node caching), testing-llm (DeepEval 2.3 + RAGAS 1.2), rag-retrieval (voyage-3 tiers + text-embedding-004), component-search (v0.app fallback), design-system-tokens (SD 5.x + DTCG v2025.10), demo-producer (VHS 0.11 + Remotion 4.0.448), and expect-agent (forbids agent-browser chat).

7.45.1 (2026-04-15)

Miscellaneous

  • bump Vercel Labs tool versions (json-render 0.17 / emulate 0.4.1 / agent-browser 0.25.4 / portless 0.10.2) (#1343) (b68168f)

7.45.0 (2026-04-15)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.108 full adoption — matrix, descriptions, doctor, worktree (#1341) (d444dbd)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.44.0] - 2026-04-15

Added

  • compat: CC 2.1.108 full adoption — 25 new matrix entries (253→278), MIN_CC_VERSION bump 2.1.105→2.1.108 across marketplace.json, CLAUDE.md, README badge, docs site
  • skills: Expanded 33 skill descriptions leveraging new 1,536-char cap (was 250) — 11 user-invocable + 22 non-invocable skills now have rich WHAT+WHEN descriptions
  • monitors: First plugin monitors implementation — chain-state-watcher (alerts on stale multi-phase skills) and uncommitted-work-reminder (periodic commit nudge) via CC 2.1.105 monitors manifest key
  • compat: Doctor version-compatibility table updated with 2.1.101-2.1.109 features, new "Recommended" tier at 2.1.108, prompt caching section (ENABLE_PROMPT_CACHING_1H)
  • compat: EnterWorktree path parameter (CC 2.1.105) documented in implement worktree-isolation-mode for reusing existing worktrees
  • compat: /recap session recovery (CC 2.1.108) integrated into brainstorm, implement, cover, verify, fix-issue skills and chain-patterns checkpoint-resume
  • compat: ENABLE_PROMPT_CACHING_1H recommended in setup and help skills for long-running sessions
  • compat: /recap and /undo added to help skill command reference

Fixed

  • docs: Standardized all 15+ fumadocs pages from 5 conflicting hook counts (106/109/115/146) to 173, agents 35→36, skills 98/99/100→103, CC version 2.1.84/2.1.86→2.1.108
  • docs: Installation guide example output updated to v7.43.0 counts, FAQ hook breakdown math corrected
  • monitors: Fixed schema from {"monitors":[...]} to top-level array [...] per CC plugin schema
  • monitors: Removed unrecognized persistent field from monitor entries
  • monitors: Added python3 ISO date parsing fallback for Linux compatibility (macOS date -jf not portable)
  • skills: brainstorm/SKILL.md trimmed 501→499 lines to pass CI skill-length gate
  • docs: CONTRIBUTING.md CC 2.1.59 references clarified with current minimum 2.1.108

7.43.0 (2026-04-14)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: sync stale hook count 169→172 in README.md (#1337) (5cb41fa)
  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105 (#1340) (75b8c90)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • hooks: resolve 8 memory bugs + auto-lint loop + 4 security alerts (#1332) (2787e1c)
  • hooks: Windows CMD spam fix + CC 2.1.105 adoption — PreCompact guard, breaking sweep (#1339) (c20c321)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/github-script from 8.0.0 to 9.0.0 (#1331) (2523518)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump actions/upload-artifact from 7.0.0 to 7.0.1 (#1330) (f40c15e)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump anthropics/claude-code-action from 1.0.90 to 1.0.93 (#1329) (7851ddd)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

7.42.1 (2026-04-14)

Bug Fixes

  • docs: sync website MIN_CC_VERSION 2.1.94 → 2.1.105, fix hardcoded counts on homepage personas (104 hooks → dynamic, 33 agents → dynamic)
  • docs: update alt text on homepage composition thumbnails to be descriptive
  • skills: update stale counts in doctor + setup + demo-producer + validate-counts references (92→103 skills, 33→36 agents, 104→173 hooks)

7.42.0 (2026-04-14)

Features

  • compat: CC 2.1.105 adoption — PreCompact guard hook blocks compaction during active agent work (override via CLAUDE_CODE_ALLOW_COMPACT_DURING_AGENTS=1)
  • hooks: PermissionDenied retry flag (CC 2.1.89) — isolated Bash/Write/Edit denials surface retry UI
  • compat: CC floor bump 2.1.94 → 2.1.105 across marketplace.json, CLAUDE.md, README badge, CI workflows
  • types: HookResult gains decision: 'block' | 'approve' and retry: boolean fields

Bug Fixes

  • hooks: add windowsHide: true to 140+ subprocess calls across 38 source files — fixes Windows CMD window spam (#1338)

Breaking Changes (upstream sweep)

  • Removed TaskOutput from 5 skill allowed-tools — deprecated in CC 2.1.83 (use Read on task output file instead)

7.41.2 (2026-04-14)

Bug Fixes

  • docs: sync stale hook count 169→172 across README, hooks.json description, and GitHub repo about

7.41.1 (2026-04-13)

Bug Fixes

  • hooks: resolve 3 HIGH memory bugs — dedup silent drop, recursion crash, regex truncation (#1137)
  • hooks: resolve 5 MEDIUM memory bugs — mixed workflows, heredoc parsing, ID collision (#1138)
  • hooks: break auto-lint revert loop with hash-guard git-add after format (#1279)
  • hooks: fix extractRationale always returning undefined (g flag + .match() drops capture groups)
  • security: resolve 4 code scanning alerts — stack trace exposure + pinned deps annotation

7.41.0 (2026-04-12)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • hooks: session quality governor — multi-signal self-regulation (#1324) (6c711ac)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve 4 code scanning alerts — info exposure + pinned deps (#1322) (05ea30c)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.40.0] - 2026-04-11

Added

  • Session Quality Governor — multi-signal quality score (0-100, S-F grades) injected into Claude's context on every turn via additionalContext. Claude self-regulates behavior based on grade: S/A normal, B efficiency mode, C conservation, D/F emergency. Computed from 4 signals: context fill %, compaction count, session age, and decision density. No new hooks, skills, or dependencies — enhances existing context-exhaustion-warner.ts from a single-signal threshold alarm into a behavioral governor. ~15ms overhead (3 file reads). Inspired by token-optimizer.

[7.39.1] - 2026-04-11

Fixed

  • Code scanning alert #138, #139 — wrap health_db() endpoint in try/except to prevent stack trace leaking to HTTP response (CodeQL py/stack-trace-exposure)
  • Code scanning alert #143 — pin @anthropic-ai/claude-code@2.1.101 with --ignore-scripts in orchestkit-eval.yml (Scorecard PinnedDependencies)
  • Code scanning alert #144 — bump pinned claude-code 2.1.90 → 2.1.101 in plugin-validation.yml (Scorecard PinnedDependencies)

7.39.0 (2026-04-11)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • compat: CC 2.1.101 full adoption — frontmatter audit, deny-overrides-ask, 18 matrix entries (#1321) (0808db8)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: CC 2.1.98 full adoption — Monitor tool, partial results, subprocess hardening (#1318) (aecedb5)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.38.0] - 2026-04-11

Added

  • CC 2.1.101 full adoption — 18 new entries in cc-version-matrix.ts (253 total): deny-overrides-ask, subagent dynamic MCP inheritance, worktree agent file access, skill context:fork enforcement, /team-onboarding, OS CA cert trust, focus mode summaries, RemoteTrigger run fix, settings resilience, LSP command injection fix
  • /team-onboarding and enterprise TLS documented in setup skill

Fixed

  • 26 skill frontmatter corrections — CC 2.1.101 enforces context: and agent: fields that were previously silently ignored:
    • Removed agent: from commit and create-pr (prevents CC from delegating instead of running skill workflow)
    • Fixed wrong agent: performance→frontend-performance-engineer, devops-deployment→ci-cd-engineer
    • Changed context: fork → inherit on 10 skills that need project state (help, memory, analytics, 7 reference skills)
    • Added missing context: inherit to 9 skills (bare-eval, dream, portless, etc.)
    • Fixed remember from context: none → inherit
  • deny-overrides-ask documented in dangerous-command-blocker.ts — permissions.deny rules now correctly override our outputAsk() decisions

[7.37.0] - 2026-04-10

Added

  • CC 2.1.98 full adoption — 17 new entries in cc-version-matrix.ts covering Monitor tool, SCRIPT_CAPS, subprocess PID sandboxing, Vertex AI wizard, /agents tabbed layout, partial progress from failed bg subagents, hook stderr display, and Bash permission hardening
  • Monitor tool added to implement, cover, verify, review-pr allowed-tools for real-time streaming of background build/test output (CC 2.1.98)
  • Partial result synthesis — all 4 workflow skills now handle partial progress from crashed background agents instead of re-spawning
  • Subprocess hardening section in security-patterns skill covering SCRIPT_CAPS, ENV_SCRUB, PID namespace isolation
  • Hook error diagnostics section in errors skill — CC 2.1.98 shows first stderr line in transcript
  • 4 new env vars documented in setup configure wizard: SCRIPT_CAPS, SUBPROCESS_ENV_SCRUB, PERFORCE_MODE, NO_FLICKER

Changed

  • Recommended CC version bumped from 2.1.97 to 2.1.98
  • implement, cover, verify, review-pr compatibility bumped to CC 2.1.98+ (Monitor tool dependency)

7.36.0 (2026-04-09)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • security: restrict CORS from wildcard to same-origin on docs site (#1316) (65dd828)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.35.1] - 2026-04-09

Fixed

  • CORS hardening — restrict Access-Control-Allow-Origin from wildcard * to https://orchestkit.vercel.app on the docs site. Vercel defaults to * for static assets; this override ensures only same-origin requests are allowed.

7.35.0 (2026-04-09)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • behavioral: superpowers-inspired behavioral hardening (#1313) (20b194d)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • release: add x-release-please-version marker to stop-uncommitted-check (#1315) (acad74e)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.34.1] - 2026-04-09

Fixed

  • release-please build drift — add x-release-please-version marker to stop-uncommitted-check.mjs so release-please updates the baked-in PLUGIN_VERSION const in sync with the manifest version. Previously, release-please bumped manifests/ork.json but left the generated hook file stale, causing CI to detect build drift on every release PR.

[7.34.0] - 2026-04-09

Added

  • Persuasion-type classification — all 103 skills tagged with persuasion-type frontmatter (14 discipline, 32 guidance, 35 reference, 22 collaborative) based on Cialdini's persuasion principles (#1308)
  • Anti-sycophancy protocol — shared rule banning performative agreement ("Great work!", "You're absolutely right!") in review and feedback skills, with pushback protocol (#1309)
  • Verification gate — cross-cutting 5-step evidence rule ("NO COMPLETION CLAIMS WITHOUT FRESH VERIFICATION EVIDENCE") referenced by review-pr, verify, fix-issue, implement, cover, commit, create-pr (#1311)
  • Agent status protocol — standardized DONE/DONE_WITH_CONCERNS/BLOCKED/NEEDS_CONTEXT reporting across all 36 agents (#1312)
  • Pressure-test framework — tests/pressure/ with multi-pressure behavioral scenarios for 5 discipline skills and RED-GREEN runner script (#1310)

Fixed

  • stamp-counts.sh, validate-counts.sh, and lint scripts now exclude shared/ utility dirs from skill/agent enumeration

7.33.0 (2026-04-09)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • compat: CC 2.1.97 full utilization (#1305) (9976a51)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.32.1] - 2026-04-09

Features

  • compat: CC 2.1.97 full utilization — 26 new features tracked (2 for 2.1.95, 24 for 2.1.97)
  • hooks: Worktree-aware session titles — prompt bar shows · wt when inside a linked git worktree
  • hooks: TRACEPARENT propagation to HQ telemetry via JSON metadata and HTTP header (CC 2.1.97)
  • hooks: isWorktree field on HookContext — detects linked worktrees via .git file check
  • docs: TRACEPARENT section in distributed-tracing reference for CC 2.1.97 subprocess tracing
  • docs: Image compression parity note in design-context-extract skill (CC 2.1.97)
  • compat: version-compatibility.md updated — 2.1.97 = Recommended, v7.33.x history row added

7.32.0 (2026-04-08)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: derive CC version from source, not hardcoded (#1303) (3a7cf79)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.31.1] - 2026-04-08

Fixed

  • Docs site CC version — banner showed "Claude Code 2.1.76+" instead of "2.1.94+". Now derived dynamically from MIN_CC_VERSION in cc-version-matrix.ts via the docs data generator, so it stays in sync automatically.

7.31.0 (2026-04-08)

Features

  • /ork:expect — Diff-Aware AI Browser Testing (M99) (#1183) (7b88cb7)
  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • agent attribution system — viral sub-agent tracking for PRs & commits (#1194) (2de086f)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • CC 2.1.83 adoption + Storybook MCP + RTK compat (#1161) (06a9f20)
  • CC 2.1.84 adoption + fuzzy search + RTK docs (#1163) (b6f1adc)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: eval pipeline — eval:skill + description optimizer + CI regression gate (#1136) (bee9be0)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • expect: route-map, init scaffolder, rich agent directive + CC 2.1.86 (#1188) (858360c)
  • fal.ai MCP integration + model pricing refresh (#1164) (c875d3d)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • QA hardening — parser fix, 3 validators, 100% test-cases coverage (#1160) (991c558)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)
  • v7.30.0 — CC 2.1.94/96 integration + skill hook unlock + sessionTitle (#1206) (2c11235)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • ci: coverage 67% → 82% + security fix — Grade C → A (#1200) (f56b9ee)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • security: bump path-to-regexp — ReDoS vulnerabilities (#1193) (c438040)
  • security: resolve 32 security findings across 5 workspaces (#1301) (a4f95bc)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • add agent attribution page to fumadocs site (#1199) (2646561)
  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump anthropics/claude-code-action from 1.0.76 to 1.0.82 (#1203) (705fa57)
  • bump anthropics/claude-code-action from 1.0.82 to 1.0.89 (#1277) (0c77f99)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • bump github/codeql-action from 4.34.1 to 4.35.1 (#1204) (33fb020)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.30.1] - 2026-04-08

Fixed

  • 27 Dependabot vulnerabilities resolved — refreshed lockfiles across all 5 workspaces (root, docs/site, src/hooks, orchestkit-demos, src/mcp-server). Bumped @modelcontextprotocol/sdk ^1.27.1 → ^1.29.0 which resolves hono@4.12.12 and @hono/node-server@1.19.13 natively. Vite resolved to 8.0.7 via vitest peer range.
  • 2 CodeQL alerts — pinned orchestkit-eval.yml actions to SHA (Scorecard Pinned-Dependencies); redacted exception in pool-setup-template.py check_db_health() (py/stack-trace-exposure)
  • 3 secret scanning alerts — dismissed as used_in_tests (fake fixtures in crypto-sanitize.test.ts)
  • 6 Biome lint errors — useOptionalChain in output.ts, mock-common.ts, normalize-command.ts
  • agent-browser file:// URL blocking — added ORCHESTKIT_AGENT_BROWSER_ALLOW_FILE env var (default: allowed) mirroring the localhost bypass pattern

Unreleased

[7.30.0] - 2026-04-08

Added

  • CC 2.1.94 + 2.1.96 integration (minimum CC version: 2.1.92 → 2.1.94)

    • Skill frontmatter hooks unlock (CC 2.1.94 bug fix) — before 2.1.94, skill hooks: blocks in SKILL.md frontmatter were silently ignored by CC. 20 OrchestKit context loaders across 15 skills (assess, implement, verify, brainstorm, review-pr, fix-issue, doctor, explore, cover, setup, commit, quality-gates, visualize-plan, release-checklist, code-review-playbook) were dead code until now. CC 2.1.94 activates them — users now get primed context (PR context, project conventions, assessment baselines, verify scoring rubric, etc.) before every workflow skill runs.
    • hookSpecificOutput.sessionTitle on UserPromptSubmit (CC 2.1.94) — unified prompt dispatcher now sets the CC session title dynamically. Format: {branch} or {branch} · {effort} when effort differs from the default. Visible in the prompt bar, --resume picker, and /remote-control session list. Emitted on every turn (no delta skip) so the title stays current across branch changes.
    • New output builder outputPromptContextWithTitle(ctx, title) in src/hooks/src/lib/output.ts — handles all four combinations: ctx+title, ctx only, title only, neither.
    • sessionTitle field in HookSpecificOutput type (CC 2.1.94).
    • DEFAULT_EFFORT_LEVEL constant in src/hooks/src/lib/effort-detector.ts — set to 'high' to match CC 2.1.94's new default (was 'medium' pre-2.1.94). Used by unified-dispatcher to decide when to log effort and when to suffix the session title.
    • keep-coding-instructions frontmatter field documentation (CC 2.1.94) for future plugin output-style skills.
    • src/__tests__/skill-frontmatter-hooks-invariant.test.ts — guards against dangling skill/<name> references in SKILL.md frontmatter. Every reference must be registered in src/hooks/src/entries/skill.ts, otherwise CC 2.1.94+ will try to spawn a missing handler on every invocation.
    • src/__tests__/prompt/session-title.test.ts — 13 new tests covering outputPromptContextWithTitle semantics, buildSessionTitle format, branch cleanup, effort suffix, branch prefix stripping, truncation, and edge cases (empty branch, oversized prompt, newlines in branch).
    • 13 new entries in cc-version-matrix.ts — 12 for 2.1.94, 1 for 2.1.96 (Bedrock bearer-token 403 hotfix).
  • Task lifecycle enforcement across 21 skills + 33 agents — all multi-phase skills and agents now follow complete TaskCreate → TaskGet → TaskUpdate lifecycle with addBlockedBy dependency chains, activeForm spinner text, and explicit completion markers

    • 8 MANDATORY skills enhanced: brainstorm, explore, verify, assess, implement, create-pr, audit-full, expect
    • 7 skills got new task sections: fix-issue, github-operations, notebooklm, demo-producer, audit-skills, release-management, setup
    • 6 partial skills completed: cover, visualize-plan, write-prd, component-search, design-context-extract, design-to-code
    • 33 agents upgraded to 6-item task management standard (added TaskGet validation)
    • CONTRIBUTING-SKILLS.md updated with task management authoring requirements
  • 34 new tests for 4 untested PreToolUse bash hooks (#1271) — coverage for secret-env-guard, terraform-plan-guard, dangerous-command-guard, docker-safety-guard

  • 10 new tests for stop-uncommitted-check.mjs — output text assertions, staged/modified/untracked classification, conciseness check

  • eval-report.sh — eval duration aggregator (npm run eval:report)

    • Reads *.trigger.json + *.quality.json results, aggregates per-skill durations
    • Flags outliers (>3× median quality duration) with ⚠
    • --json for machine-readable output, --top N to control display
    • Tested against 84 real skill results (413min total, 2 outliers flagged)
  • known-hook-conflicts.json — hook-skill conflict registry for quality eval

    • Quality eval diffs actual hook rejections vs registry
    • HOOK COMPATIBILITY section: COMPATIBLE ✓ / KNOWN CONFLICTS / NEW CONFLICTS ✗
    • Seeded with i18n-date-patterns finding (2 incidental rejections)

Changed

  • Hook dispatcher flattening — all 6 async dispatchers replaced with CC 2.1.92 native async hook entries
    • Stop dispatcher → 9 individual entries (ff75bedd)
    • 4 more async dispatchers flattened (1de40ec3)
    • PostToolUse dispatcher — last of 6 (6a082e32)
    • Hook count: 146 → 169 (100 global + 47 agent-scoped + 22 skill-scoped)
  • 3 dead/low-value hooks removed — gh-issue-creation-guide, license-compliance, pr-merge-gate (−494 LOC) (#1274)
  • Deprecated TaskOutput references removed (#1268)

Fixed

  • 19 stale skill name references across docs, demos, and source
    • brainstorming → brainstorm (6 files), plan-viz → visualize-plan (2), prd → write-prd (1), assess-complexity → quality-gates (4), agent-browser prefix fix (2), git-workflow rows removed (2)
  • 2 phantom demo configs deleted — add-golden-demo.ts, worktree-coordination-demo.ts (skills never existed)
  • 2 demo configs remapped — assess-complexity → quality-gates, run-tests → cover
  • hooks.json description count drift — 171 → 169, 47 agent-scoped → 45
  • api-design trigger eval removed — user-invocable: false skill can't be tested by trigger classifier (0% precision/recall was expected, not a bug)
  • 41 stale "balance is too low" quality result files purged — API credit exhaustion during grading produced fake 0% pass rates
  • stop-uncommitted-check .trim() bug — .trim() on git status --porcelain output stripped leading spaces, misclassifying unstaged modifications as staged changes (#1293)
  • Stop hook message shortened — verbose AI-instruction text replaced with concise human-readable summary

Performance

  • learning-tracker pattern caching — patterns compiled once per process instead of per-invocation (#1265)
  • SessionStart sync hooks gated on input.source — skip sync hooks when source is not relevant (#1269)
  • 3 redundant appendAnalytics JSONL writes removed (#1266)

[7.29.0] - 2026-04-05

Added

  • 72 new tests for P0 coverage gaps:
    • 54 tests for normalizeInput() edge cases: JSON-string unwrapping (12), STRING_FIELDS allowlist (12), tool_name/session_id/hook_event normalization (8), project_dir isValidPath resolution (6), agent_type/agent_id promotion (5), tool_input fallback (6), isValidPath validation (5)
    • 11 tests for crypto sanitization: 3 missing patterns (gho_, github_pat_, xoxp-), 4 boundary/minimum-length tests, 3 false-positive prevention tests, 1 extra short-PAT rejection test
    • 7 tests for HookContext shape: compile-time type assertion, runtime property audit across all 11 HookContext properties, NOOP_CTX callability, buildContext() shape parity
  • CC 2.1.92 version compatibility — 12 new entries in version-compatibility.md, 8 features in cc-version-matrix.ts (#1263)
    • forceRemoteSettingsRefresh policy: fail-closed startup for enterprise managed settings
    • Stop hook preventContinuation:true semantics restored
    • Tool input JSON-string streaming fix
    • Plugin MCP stuck "connecting" fix
    • Write tool 60% faster diff computation
    • Remote Control hostname-based session naming
    • Per-model /cost breakdown
    • Subagent tmux pane count fix
    • /tag and /vim commands removed (no OrchestKit impact)
  • forceRemoteSettingsRefresh in configure/doctor skills — enterprise policy docs, health check outputs for managed settings and MCP connector conflicts
  • Defensive JSON-string unwrap in run-hook.mjs — normalizeInput() auto-parses string-encoded array/object tool_input fields while preserving known string fields via allowlist (defense against CC streaming regression)

Changed

  • HookContext DI migration completed — all 214 hook source files migrated to (input, ctx) signature with NOOP_CTX default fallback
    • buildContext() factory + createTestContext() test helper for dependency injection
    • Shared mock factory replaces inline mocks across 137 test files
    • common.ts god module split into output.ts, env.ts, log.ts
    • Telemetry decoupled — inline webhookForwarder removed from common module
    • All unused testCtx scaffolding removed from lib/utility test files (16 files)
    • All biome lint errors resolved: 0 errors, 0 warnings across 530 files

[7.28.0] - 2026-04-03

Added

  • Telemetry Provider Architecture (M105) — pluggable sink system with Grafana Echo-style interface
    • TelemetrySink interface: { name, supportedEvents, addEvent, flush } for pluggable sinks
    • emit() API: central fan-out to registered sinks with per-sink failure isolation
    • JsonlSink: local JSONL backup via appendFile (async, non-blocking). Always-on safety net — events never lost even when HTTP sink is down
    • HttpSink: HMAC-signed POST with 3x retry (full-jitter exponential backoff) and circuit breaker (5 fails → OPEN, 30s cooldown → HALF_OPEN)
    • telemetry-sync.mjs: batch replay CLI — reads JSONL, POSTs to /batch-ingest as NDJSON, deletes synced rotated files
    • Config-based sink registry: plugin.json and settings.local.json can register custom HTTP sinks
    • Webhook forwarder coverage validator: CI gate ensuring all 27 CC events have forwarder coverage
    • Rotation: files >10MB renamed on SessionEnd, rotated files >7 days cleaned up
  • Payload sanitization — 15 secret patterns redacted before transmission
    • API keys: sk-ant- (Anthropic), sk- (OpenAI), AKIA (AWS), AIza (Google/Firebase)
    • Tokens: GitHub PATs (ghp_, gho_, github_pat_), Slack (xoxb-, xoxp-), Bearer
    • Database URLs: MongoDB, PostgreSQL, MySQL connection strings
    • Environment variable assignments with secret names
    • Recursive sanitization with 500-char truncation
  • 27/27 CC event coverage — all documented (20) and undocumented (7) CC hook events forwarded

Fixed

  • Webhook forwarder coverage: 9 gaps fixed (FileChanged standalone, 8 dispatcher inlines)

  • JsonlSink always registers (was gated behind HTTP config — silent data loss when webhooks disabled)

  • Telemetry directory permissions: 0o755 → 0o700 (world-readable on shared systems)

  • hook_event_name → hook_event normalization in run-hook.mjs (all events logged as "unknown")

  • Plugin/user HttpSink instances now receive config URL/token (were all hitting env-configured default)

  • Stale test assertions: async hook counts, split-bundle counts, description totals, cross-reference prefixes

  • GitHub push protection: replaced realistic-looking test fixtures with obviously-fake tokens

  • CC 2.1.91 integration: MCP result size override, disableSkillShellExecution, plugin bin/ executables, Edit shorter anchors

    • mcp-output-transform "Trust CC's decision" heuristic: results > 50K chars were explicitly kept by CC (likely _meta["anthropic/maxResultSizeChars"]), so skip truncation. Results 2K–50K get token-saving truncation. PII redaction always runs regardless of size. Best-effort _meta extraction kept as fallback.
    • ORCHESTKIT_MCP_TRUNCATION_THRESHOLD env var to override default 2K truncation threshold
    • 7 new unit tests for CC trust heuristic (large result skip, PII on large results, _meta extraction, invalid _meta, env override)
    • Fixed pre-existing test mock: getProjectDir + webhookForwarder missing from mcp-output-transform test mocks
    • disableSkillShellExecution fallback notes added to 4 skills with invocation_hooks (cover, expect, commit, devops-deployment)
    • cc-version-matrix.ts: +8 entries for CC 2.1.91 features
    • mcp-patterns skill: added _meta annotation documentation, code example in server-setup rule
    • cc-version-settings.md: new section covering disableSkillShellExecution, plugin bin/, Edit anchors, auto permission validation
    • version-compatibility.md: +8 feature matrix entries, v7.29.x history row, 2.1.91 compatibility level

Changed

  • webhookForwarder() simplified to thin wrapper around emit() (public API unchanged — dispatchers need zero changes)
  • signPayload() extracted from usage-summary-reporter.ts to shared lib/crypto.ts
  • sanitizePayload() extracted from session-tracker.ts to shared lib/crypto.ts with expanded patterns
  • Hook count: 131 → 132 (added telemetry-sync on SessionEnd)
  • Minimum Claude Code version: 2.1.90 → 2.1.91
  • README badge updated from 2.1.90 to 2.1.91
  • mcp-patterns skill compatibility bumped from 2.1.76+ to 2.1.91+
  • mcp-output-transform uses "Trust CC's decision" heuristic for results > 50K instead of blindly truncating everything to 2K

[7.27.0] - 2026-04-02

Added

  • shadcn/ui v4 style system across 7 design skills: ui-components, design-to-code, component-search, design-context-extract, json-render-catalog, design-system-tokens, ai-ui-generation
    • 6 styles documented: Vega, Nova, Maia, Lyra, Mira, Luma
    • Preset code system (npx shadcn@latest init --preset <code>)
    • Style detection from components.json → "style" field
    • Style-aware component adaptation in design-to-code pipeline
    • Best-fit style recommender in design-context-extract
    • Style-aware catalog overrides in json-render-catalog
    • Luma elevation tokens (shadow-md + ring) in design-system-tokens
  • ui-components/rules/shadcn-v4-styles.md — new rule file for v4 style detection, class mapping, and preset codes (622 total rules)

Fixed

  • P0 BUG denial-notification.ts: in-memory denialTimestamps array reset on every hook invocation (fresh Node.js process per call). Now reads persisted timestamps from permission-denials.jsonl written by denial-logger, with cooldown state in separate JSON file
  • P1 SEC project-write-retry.ts: added resolveRealPath() before isInsideDir() to prevent symlink bypass attacks (replicates ME-001 fix from file-guard.ts)
  • P1 SEC auto-approve-project-writes.ts: added matching resolveRealPath() call to close asymmetry between approval and retry hooks
  • denial-logger.ts: JSDoc incorrectly said "Async" but uses appendFileSync — corrected to "Sync"
  • hasExcludedDir() now catches terminal path segments (e.g., /project/.git without trailing slash)
  • Stale hook count "110 hooks" updated to "112 hooks" across 11 doc pages
  • Stale skill count "100/101 skills" updated to "102 skills" across 11 doc pages

Added

  • CC 2.1.88 integration: 5 PermissionDenied hooks via unified-dispatcher (denial-logger, denial-notification, safe-command-retry, project-write-retry)
  • CC 2.1.89 integration: headless-defer PreToolUse hook returns {decision:"defer"} in headless -p mode for destructive ops (force push, npm publish, terraform apply, kubectl, docker push, gh merge). Configurable via ORCHESTKIT_DEFER_TOOLS env var
  • /ork:release-sync skill: post-release content sync to NotebookLM + HQ Knowledge Base. Reads CHANGELOG, CLAUDE.md, hook README, updates notebook sources
  • lib/path-containment.ts — shared EXCLUDED_DIRS, isInsideDir(), hasExcludedDir(), resolveRealPath() (extracted from 2 duplicating hooks)
  • lib/bash-patterns.ts — unified REJECT_PATTERNS (was duplicated with inconsistencies between auto-approve-safe-bash and safe-command-retry)
  • outputDefer() helper in lib/common.ts for PreToolUse permission deferral
  • types.ts: permissionDecision union extended with 'defer' (CC 2.1.89)
  • cc-version-matrix.ts: +9 entries for CC 2.1.89 features (defer permission, TaskCreated hook, autocompact thrash detection, cleanup validation, MCP nonblocking, named subagent typeahead, hook output disk spill, edit after bash view, symlink permission check)
  • MCP_CONNECTION_NONBLOCKING=true in ork.settings.json for faster headless -p mode startup
  • yarn read-only patterns added to RETRY_SAFE_PATTERNS
  • PermissionDenied section added to src/hooks/README.md
  • 167 new unit tests (152 PermissionDenied + 15 headless-defer)
  • tests/skills/test-upstream-refs.sh — CI freshness check for vendor references
  • Candlekeep batch (CC 2.1.89 architecture insights — #1227-#1236):
    • CwdChanged hook (#1229): Detects tech stack on directory change, returns watchPaths, suggests relevant skills via path_patterns
    • FileChanged hook (#1230): Reacts to watched file changes (CLAUDE.md, package.json, .env, rules/), injects context
    • mcp-output-transform hook: PostToolUse PII redaction (emails, phones) and token-saving truncation for MCP results
    • cache-break-detector hook (#1234): Analytics-only — tracks prompt injection shape changes across turns for cache hit monitoring
    • frustration-detector hook: Analytics-only — privacy-first DX signal (boolean only, no matched text logged)
    • fork() pattern (#1227): Reference doc, explore + brainstorm skills annotated fork-friendly, context-stager skips forks, cache_hit_pct analytics
    • path_patterns (#1228): Added to 15 skills, CwdChanged consumer with 2-level directory scan (94% pattern match rate)
    • critical_system_reminder (#1231): Added to 8 agents, injected at spawn via subagent-context-stager
    • required_mcp_servers (#1232): Added to 4 agents, pre-flight availability check via CLAUDE_MCP_SERVERS env var
    • invocation_hooks (#1235): Shell precondition commands on 4 skills (cover, expect, commit, devops-deployment)
    • MCP channels (#1233): evalApiKey + notebookLmToken sensitive userConfig slots (CC keychain storage)
    • Token budget carry-forward (#1236): Pre/post compact preserves effort level and budget across compaction
    • Transcript quality analysis: SubagentStop bounded JSONL reads for tool counts, errors, token usage
    • SessionStart perf measurement: sync-session-dispatcher logs duration_ms to analytics
    • CC field name diagnostic: Debug-level log of input field names on first SubagentStop for runtime verification
    • 167 new Candlekeep tests (cwd-changed 21, file-changed 14, mcp-output-transform 35, cache-break 21, frustration 28, post-compact 20, analyze-transcript 17, fork detection 4, sync-session perf 3, materialize-rules 4)
    • Cache audit documented in hooks README: only 2 genuinely volatile per-turn hooks, rest migrated or gated

Changed

  • Option E: Replaced 37 copied Vercel Labs skills with upstream references. Their SKILL.md content deposited as references/upstream-*.md inside 6 existing OrchestKit skill directories (browser-tools, json-render-catalog, mcp-visual-output, multi-surface-render, emulate-seed, portless). Skill count: 132 → 102 (101 after Option E + 1 release-sync)
  • scripts/sync-vercel-skills.sh enhanced with JSON mapping file, content-hash dedup (no timestamp-only diffs), SHA pinning in manifest, --check and --dry-run modes
  • vendor/vercel-skills/mapping.json — 37 refs across 4 Vercel repos → 6 skill dirs (committed for offline builds)
  • docs/site/lib/generated/agents-data.ts — now auto-generated from src/agents/*.md frontmatter on every build (was hand-maintained, 8 agents missing, 9 phantom agents). Added taskTypes, keywords, examplePrompts to all 36 agent frontmatter files
  • Removed 32 vercel-*.mdx pages from fumadocs, updated meta.json and category pages
  • Agent descriptions improved for @mention typeahead discoverability: genui-architect, ui-feedback
  • Dead code removed: memory-bridge entity extraction (132→33 lines), antipattern-warning function deleted (was no-op since #1145), speculative mcp_connections HookInput field removed
  • Updated specs: CONTRIBUTING-SKILLS.md (path_patterns, invocation_hooks), agent-authoring.md (critical_system_reminder, required_mcp_servers), frontmatter validator (critical_system_reminder length check)
  • Stale counts fixed across 22 fumadocs pages (102→103 skills, 112→115 hooks)
  • Bounded JSONL read (last 4KB) + atomicWriteSync for crash-safe state persistence
  • Minimum Claude Code version: 2.1.86 → 2.1.90
  • CC 2.1.90 integration: version matrix +11 entries (powerup, plugin-keep-marketplace, husky-protected, exit code 2 fix, format-on-save fix, resume prompt cache fix, 3x perf, resume hides -p sessions, DNS auto-allow removal)
  • .husky/ added to file-guard PROTECTED_PATTERNS (CC 2.1.90 parity — prevent git hook tampering)
  • Path resolution comments updated from "backward compat" to "defense in depth" — fallbacks retained as defensive programming rather than removed
  • README badge updated from 2.1.86 to 2.1.90
  • version-compatibility.md: +25 entries for CC 2.1.88–2.1.90, compatibility levels updated, v7.27.x version history row
  • cc-version-settings.md: added CC 2.1.90 section (PLUGIN_KEEP_MARKETPLACE env var, format-on-save hook pattern, /powerup reference)

[7.26.4] - 2026-03-29

Fixed

  • Flaky analytics-consent-check test (daysAgo boundary jitter at exact 30 days)
  • path-to-regexp ReDoS vulnerability (8.3.0 → 8.4.0)

Added

  • 568 unit tests across 17 files — hook coverage 67% → 82%
  • Tests for: decision-history, problem-tracker, task-integration, guards, decision-flow-tracker, session-parser, setup-repair, first-run-setup, setup-check, setup-maintenance, naming-convention-learner, code-style-learner, issue-subtask-updater, readme-sync, context7-tracker, workflow-preference-learner, model-cost-advisor

[7.26.3] - 2026-03-29

Fixed

  • Agent attribution: race condition in session state (lockedAtomicWriteSync)
  • Agent attribution: path traversal in sanitizeBranch (allowlist-based)
  • Agent attribution: double ork: prefix in Co-Authored-By trailers
  • Agent attribution: filterByThreshold now checks duration only, not success flag
  • Agent attribution: cleanupStaleLedgers uses TTL only (branch name reversal was lossy)

Changed

  • Split agent-attribution.ts into 3 files (types, core, formatters)

Added

  • 35 unit tests for agent attribution (sanitize, ledger, format, deduplicate)
  • 16 integration + E2E tests (session state pipeline, concurrent access, full flow)
  • Agent attribution docs page in fumadocs site

[7.26.2] - 2026-03-28

Changed

  • Refactored run-hook.mjs: extracted SILENT_OK constant replacing 5 hardcoded JSON strings

[7.26.1] - 2026-03-28

Fixed

  • Bumped path-to-regexp to fix ReDoS vulnerabilities (Dependabot alerts #47, #48, #53, #54)

[7.26.0] - 2026-03-28

Added

  • /ork:expect improvements — reliability and DX upgrades (#1185, #1186, #1187):
    • Route-map script: framework-aware file→URL mapping for Next.js App Router, Pages Router, Remix, SvelteKit — replaces Phase 3 prompt guessing with deterministic detection (#1185)
    • --init scaffolder: bash scripts/init.sh creates .expect/ directory with config.yaml template, .gitignore, and flows/ directory in 5 seconds (#1186)
    • Rich expect-agent directive: 194-line agent body (was 61) with full agent-browser command reference, ARIA selector patterns, form interaction workflow, status protocol examples, and failure decision tree (#1187)
    • Report script: parses status protocol to terminal/CI/JSON output, auto-saves to .expect/reports/ (#1189)
    • Example saved flows: login.md, crud.md, navigation.md — ARIA-first templates for common test patterns (#1190)
    • Fingerprint auto-save hook: PostToolUse dispatcher saves fingerprint after successful run, enabling zero-cost skip on next invocation (#1191)
  • CC 2.1.86 adoption: bumped minimum from 2.1.85 → 2.1.86 for config disk write fix (eliminates unnecessary writes on every skill invocation) and Write/Edit fix for files outside project root

[7.25.0] - 2026-03-26

Added

  • /ork:expect — Diff-Aware AI Browser Testing (M99, 18 issues):
    • Skill scaffold: SKILL.md (301 lines), 13 reference files, 5 rules, 3 scripts, 12 test cases (#1165)
    • Git diff scanner: 4 target modes (changes/unstaged/branch/commit), 3 data levels (files/stats/preview), magnitude-based 12-file prioritization (#1166)
    • SHA-256 fingerprint gating: check/save/clear commands, zero-cost skip when unchanged (#1167)
    • Scope-aware test depth: commit=narrow(2-4), unstaged=exact(2-3), changes=combined(3-5), branch=thorough(5-8) (#1168)
    • AI test plan generation: 8-section prompt template with coverage context, ARIA-first interaction, anti-rabbit-hole heuristics (#1169)
    • Machine-parseable status protocol: STEP_START/STEP_DONE/ASSERTION_FAILED/RUN_COMPLETED with parser (#1170)
    • Failure categorization: 6 types (app-bug, env-issue, auth-blocked, missing-test-data, selector-drift, agent-misread) (#1171)
    • ARIA snapshot diffing: semantic UI change detection via accessibility tree comparison (#1172)
    • Saved test flows: Markdown+YAML format_version 1 with adaptive replay (#1173)
    • Test coverage context: cross-reference changed files with existing tests, 4 patterns (#1174)
    • Execution engine: agent-browser orchestration, auth profiles, failure decision tree (#1175)
    • Report generator: terminal/CI/JSON output, GitHub Actions annotations, exit codes (#1176)
    • .expect/config.yaml convention: base_url, route_map, auth, ARIA snapshots, accessibility (#1177)
    • rrweb session recording: DOM event replay without video encoding (#1178)
    • Human-in-the-loop plan review: AskUserQuestion gate before execution (#1179)
    • CI integration: GitHub Actions workflow, pre-push hook, fingerprint-gated zero-cost skip (#1180)
    • Research reference: millionco/expect architecture analysis + feature comparison (#1181)
    • expect-agent: specialized browser test execution subagent (#1182)
  • Skill count 99 → 100, agent count 35 → 36, invocable commands 20 → 21
  • CC 2.1.85 integration:
    • Hook if conditionals: Bash PreToolUse and PermissionRequest now use CC 2.1.85 if field to skip process spawning for irrelevant commands (~40% fewer hook spawns)
    • Headless AskUserQuestion: new pretool/ask/headless-responder hook auto-selects first option when CI=true or ORCHESTKIT_HEADLESS=1, using CC 2.1.85 updatedInput + permissionDecision
    • Hook count 109 → 110 (41 global + 47 agent + 22 skill), CC minimum 2.1.84 → 2.1.85

[7.24.2] - 2026-03-26

Added

  • fal.ai MCP integration — added fal MCP server to configure skill reference (1000+ models: FLUX.2, Kling 3.0, LTX 2.0, Veo 3.1, Chatterbox TTS), new "Generative media" project type
  • Video model pricing refresh — updated stale fal.ai pricing in multimodal-llm skill ($0.90/10s → Kling $0.07/s, Veo 3.1 $0.40/s, Wan 2.5 $0.05/s)

Fixed

  • Scorecard #132 — added nosemgrep annotation for npm pinned-dependencies advisory

[7.24.1] - 2026-03-26

Fixed

  • Stop hook version display — replaced hardcoded ork@7.22.0 with build-time injection from manifests/ork.json via __PLUGIN_VERSION__ placeholder
  • Security: picomatch CVE-2026-33671 + CVE-2026-33672 — bumped picomatch to 4.0.4 across 3 directories; also bumped express-rate-limit (GHSA-46wh-pxpv-q5gq)
  • Stale counts — synced hook count (106→109), skill count (94/98→99), CC version (2.1.75→2.1.84) across 6 docs pages

[7.24.0] - 2026-03-26

Added

  • CC 2.1.84 adoption — Phase 1 quick wins:
    • paths: YAML glob lists on 13 skills — auto-loads relevant context files when skill activates (implement, design-to-code, memory, storybook-mcp, mcp-patterns, database-patterns, cover, fix-issue, devops-deployment, security-patterns, architecture-patterns, configure, explore)
    • CLAUDE_STREAM_IDLE_TIMEOUT_MS=180000 — prevents long-running agents (audit-full, implement) from hitting the 90s default idle timeout
    • WorktreeCreate type: "http" support — returns worktree path via hookSpecificOutput.worktreePath for API-driven orchestration
    • 21 CC 2.1.84 features added to doctor version compatibility matrix
    • MCP 2KB tool description cap documented in CONTRIBUTING-SKILLS.md
  • TaskCreated hook event (CC 2.1.84) — 3 new handlers:
    • creation-tracker — logs task creation events to JSONL + cross-project analytics
    • task-context-injector — enriches tasks with branch/commit context via additionalContext
    • task-progress-initializer — detects [N/M] numbered task patterns and initializes progress bar state
  • 38 unit tests for TaskCreated hooks (creation-tracker, task-context-injector, task-progress-initializer)
  • Fuzzy search with Fuse.js — replaced .includes() substring matching in skill browser and agent selector with Fuse.js fuzzy search:
    • Typo tolerance (threshold 0.35), multi-word queries, word-order-independent matching
    • Relevance ranking with field weighting (name 3x, tags 2x, description 1x)
    • 150ms debounced input to prevent result flashing
    • <Highlight> component renders match ranges with yellow highlights
    • "Did you mean?" smart empty state with relaxed-threshold suggestions
    • Shared useFuzzySearch hook and createSearch() / createRelaxedSearch() in lib/search.ts
    • 23 unit tests covering fuzzy matching, relevance ranking, match metadata, and relaxed suggestions
  • RTK (Rust Token Killer) documentation — added Git Validator section to safety-hooks docs covering stripProxyPrefix() flow, RTK v0.33.1 coexistence guide, --hook-only setup, and git status output change warning; added RTK FAQ entry

Changed

  • CC version requirement bumped to >= 2.1.84 (from >= 2.1.83)
  • Hook count 106 → 109 (37 → 40 global hooks)
  • HookEvent type updated: added TaskCreated event
  • HookInput type updated: added type field for WorktreeCreate HTTP, task_description for TaskCreated
  • HookSpecificOutput type updated: added worktreePath field for WorktreeCreate HTTP

[7.23.0] - 2026-03-25

Added

  • CC 2.1.83 adoption — Tier 1 quick wins:
    • sandbox.failIfUnavailable: true — fail-fast when sandbox runtime unavailable instead of running unsandboxed
    • CLAUDE_CODE_SUBPROCESS_ENV_SCRUB=1 — strip Anthropic/cloud credentials from all subprocess environments (Bash, hooks, MCP servers)
    • initialPrompt for all 15 background agents — zero-wasted-turn bootstrap on spawn
    • userConfig with sensitive:true — keychain-backed webhook token storage via manifest
  • RTK proxy compatibility — git-validator and guardGitCommand now strip rtk prefix before pattern matching, ensuring commit/branch validation works with RTK-rewritten commands

Changed

  • Deprecated TaskOutput removed from 6 agents (eval-runner, release-engineer, monitoring-engineer, data-pipeline-engineer, market-intelligence, web-research-analyst) — replaced by Read on background task output file per CC 2.1.83 deprecation
  • CC version requirement bumped to >= 2.1.83 (from >= 2.1.81)

[7.22.1] - 2026-03-24

Fixed

  • YAML parser >- folded scalar support: parse-frontmatter.js now handles >-, |-, >, | multiline scalars — prevents "description too short (2 chars)" CI failures for new skills/agents
  • test-cases.json 100% coverage: created missing test-cases.json for 8 skills (component-search, cover, design-context-extract, testing-e2e, testing-llm, testing-patterns, testing-perf, testing-unit)

Added

  • Pre-commit frontmatter schema validator: checks agent required fields (category, context, mcpServers, description ≤250 bytes) and skill required fields before commit
  • Pre-commit test-case rule validator: verifies each rule: reference in test-cases.json has a matching rules/*.md file
  • MDX count drift scanner (bin/validate-mdx-counts.sh): detects stale hardcoded total counts in docs
  • Parser test suite: 10 tests covering all YAML scalar types

[7.22.0] - 2026-03-23

Added

  • json-render + emulate integration (Milestone #98): 4 new skills, 2 agents, 3 skill upgrades wrapping Vercel Labs' json-render (AI-safe GenUI) and emulate (stateful API emulation)
    • json-render-catalog skill: Zod-typed component catalogs for AI-safe UI generation, 36 shadcn components, YAML token optimization
    • emulate-seed skill: stateful API emulation config generation for GitHub (:4001), Vercel (:4000), Google OAuth (:4002)
    • mcp-visual-output skill: interactive dashboards via @json-render/mcp in Claude/Cursor conversations
    • multi-surface-render skill: same JSON spec renders to React, PDF, email, Remotion video, OG images
    • genui-architect agent: json-render catalog design specialist
    • emulate-engineer agent: stateful API emulation specialist

Changed

  • ai-ui-generation: json-render added as first choice option alongside v0/Bolt/Cursor
  • testing-integration: emulate added as first choice for GitHub/Vercel/Google API tests (new emulate-stateful-testing rule)
  • testing-e2e: emulate backends for Playwright E2E (new emulate-e2e rule with port isolation)
  • design-to-code: Stage 4 RENDER added — register adapted components as json-render catalog entries
  • Component counts: 94→98 skills, 33→35 agents

Fixed

  • ai-review-checklist.md: impact value CRITICAL → HIGH (invalid enum)
  • integration-api.md: trimmed from 165 to 140 lines (was over 150 limit)
  • Added missing test-cases.json for testing-integration and design-to-code skills
  • Quoted impactDescription in 3 rule files for YAML consistency

[7.21.1] - 2026-03-22

Fixed

  • Brace expansion false positive on jq syntax: --jq '{name, os, status}' was blocked as "brace expansion with command-like pattern". Bash brace expansion requires NO spaces between elements ({cat,/etc/passwd}), so spaced patterns ({name, os}) and JSON/jq colon syntax ({key: .value}) are now correctly allowed.

[7.21.0] - 2026-03-21

Added

  • Portless + agent-browser in debug workflows: all debug-related skills/agents now instruct to use Portless named URLs and agent-browser for visual inspection
    • debug-investigator agent: service discovery step, visual inspection, updated examples
    • fix-issue skill: service discovery + agent-browser phase before hypothesis formation
    • errors skill: ECONNREFUSED/connection refused pattern with Portless fix
    • doctor skill: Portless health check in external dependencies
    • performance skill: local profiling target section with Portless URLs
    • browser-tools skill: new portless-local-dev rule file with incorrect/correct patterns
  • CC 2.1.81 integration: --bare mode for eval pipeline, version-compatibility matrix, configure skill docs
  • bare-eval skill: new skill for isolated eval/grading calls using --bare (SKILL.md, 3 references, 3 rules, test-cases.json)
  • triggers: frontmatter: all 20 user-invocable skills declare keywords, examples, and anti-triggers
  • test-trigger-keywords.sh: deterministic keyword-based trigger matching — $0, 5 seconds, 17/17 pass (replaces $3/45min LLM classifier)
  • Content hash cache: skip re-eval if SKILL.md + .eval.yaml unchanged (--force to bypass)
  • --changed mode: git-diff eval — only eval skills changed vs main (ideal for CI)
  • --skip-baseline --force-skill: fastest quality eval mode (1 call per prompt instead of 3)

Changed

  • Default eval model: Haiku for all generation AND grading (was Sonnet — 12x cost reduction)
  • Trigger reps: 5→3 (95%+ detection with 40% fewer calls)
  • Trigger eval: parallel prompt execution (6 workers via EVAL_MAX_PARALLEL), skip non-invocable skills
  • Trigger timeout: 120s→60s for classification calls
  • agent-browser safety: allow localhost, 127.0.0.1, and *.localhost URLs (dev servers should never be blocked)
  • Min CC version: >= 2.1.81 (was 2.1.80)

Fixed

  • Hook resilience — 5 fail-silent bugs made visible:
    • run-hook.mjs: stderr warning when stdin >512KB truncated (was silent data loss)
    • run-hook.mjs: stderr warning when truncated JSON falls back to {} (was silent no-op)
    • run-hook.mjs: appendFile error callbacks replace fire-and-forget () => {} (was silent write failure)
    • run-hook.mjs: session ID validation strengthened to UUID + smart-ID structural patterns (was permissive regex)
    • stop-failure-handler: tries fallback field names, logs available keys when reason unknown (was always "unknown")
    • unified-dispatcher: webhook health check on SessionStart — warns if endpoint unreachable
  • agent-browser blocking localhost:PORT and 127.0.0.1:PORT URLs
  • Keyword collision: "grade" removed from verify (owned by assess)
  • set -e crash on ((x++)) when x=0 in trigger test
  • Agent frontmatter test labels updated from stale "CC 2.1.6 compliant" to "valid"

[7.20.0] - 2026-03-21

Added

  • feat(eval): unified npm run eval:skill command — trigger + quality in one command per skill
  • feat(eval): npm run eval:optimize-desc — iterative description improvement with train/test split
  • feat(eval): check-eval-regression.sh — CI regression gate comparing eval baselines (no API calls)

7.19.0 (2026-03-21)

Features

  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)
  • skills: version management — renamed_from, targets, ref validation (#1132) (2766fd9)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Documentation

  • update README What's New to v7.18.0, link to docs site changelog (#1134) (d8aebef)

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.18.0] - 2026-03-21

Added

  • feat(skills): targets frontmatter with library version ranges on 14 skills
  • test: validate all skills: cross-references point to existing skills (93 refs)
  • test: validate targets: entries have library declarations

7.17.0 (2026-03-21)

Features

  • skills: update notebooklm skill for v0.5.2 + cinematic video (#1130) (28d3726)

7.16.2 (2026-03-20)

Bug Fixes

  • docs: parse release-please changelog format in data generator (#1128) (6efbf19)

7.16.1 (2026-03-20)

Miscellaneous

7.16.0 (2026-03-20)

Features

  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • compat: update OrchestKit for CC 2.1.80 features (#1124) (1f0e578)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)
  • docs: fix hooks spotlights generation in build script (#1126) (c74ebf4)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.15.0] - 2026-03-20

Added

  • feat(skills): effort frontmatter on 40 skills — CC 2.1.80 overrides model effort level (18 high, 22 low)
  • feat(hooks): rate_limits display in statusline — shows rate limit usage alongside context percentage
  • feat(compat): version-compatibility matrix updated with 13 CC 2.1.80 entries
  • docs: simplified plugin install commands (/plugin install single-step flow)

7.14.1 (2026-03-19)

Bug Fixes

  • ci: auto-bump docs site version on release (#1117) (ddf5f3b)

7.14.0 (2026-03-19)

Features

  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • CC 2.1.78 adoption — StopFailure, PLUGIN_DATA, effort frontmatter, worktree (#1104) (de50151)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: docs site overhaul — CTA, categories, sidebar, dep graph, quiz (#1114) (edd84e5)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.13.0] - 2026-03-18

Added

  • Invocation CTA on all 93 skill reference pages — copy-to-clipboard /ork:skillname for commands, auto-activated notice for reference skills (#1082)
  • 8 category index pages at /skills/by-category/ — backend, frontend, testing, security, AI/LLM, devops, product, workflows (#1082)
  • ContextualSkillSidebar on all skill reference pages — shows dependencies, reverse deps, related skills, agent link (#1080)
  • Skill dependency graph data generation from frontmatter skills: field — 93 nodes, 101 edges (#1084)
  • SkillRecommender quiz on skills overview — 3-step role/task/stack flow with tag-based scoring (#1079)

Fixed

  • Dependency graph readability — 800px viewport, LR layout, focus dropdown, auto-focus on hub nodes, node dimming on hover
  • Category false positives — removed overly broad tags from AI/LLM and product categories
  • Quiz UX — added 200ms selection feedback delay before step advancement

[7.12.0] - 2026-03-18

Features

  • StopFailure hook (CC 2.1.78): new stop/stop-failure-handler.ts flushes state and writes emergency handoff on API error/rate limit — error-loop safe by design (#1106)
  • CLAUDE_PLUGIN_DATA persistence (CC 2.1.78): getPluginDataDir/getSessionStorageDir/getAnalyticsStorageDir in paths.ts; session storage migrates to persistent directory that survives plugin updates (#1107)
  • effort frontmatter defaults: 21 agents annotated with effort: low/medium; remaining 12 default to high (#1108)
  • worktree hook validation: tests confirming WorktreeCreate/WorktreeRemove registration and handler correctness after CC 2.1.78 --worktree flag fix (#1109)

Changed

  • Minimum Claude Code engine bumped to >= 2.1.78
  • Hook count: 105 → 106

[7.11.2] - 2026-03-17

Fixed

  • TODO: Describe your changes here

7.11.1 (2026-03-16)

Bug Fixes

  • security: resolve yauzl CVE and pinned-dependencies alert (#1101) (d96d68d)

Miscellaneous

  • remove release-as pin after 7.11.0 release (#1103) (5fd681c)

7.11.0 (2026-03-16)

Features

  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • cover: add /ork:cover skill + bump to 7.11.0 (#1096) (36f14c6)
  • docs: integrate @yonatan-hq/analytics for cross-project tracking (#1090) (3e294dc)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • notebooklm: update skill for notebooklm-mcp-cli v0.4.8 (#1094) (0b72024)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/download-artifact from 8.0.0 to 8.0.1 (#1097) (ac52e6a)
  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)
  • update anthropics/claude-code-action requirement to 26ec041249acb0a944c0a47b6c0c13f05dbc5b44 (#1098) (8a7600e)

[7.9.1] - 2026-03-15

Added

  • Integrate @yonatan-hq/analytics for cross-project page view tracking in docs site
  • Add Edge route handler for HMAC-signed analytics event forwarding to HQ API

7.9.0 (2026-03-15)

Features

  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • eval: Sprint 1 — trigger runner + CONTRIBUTING-SKILLS eval docs (#1065) (8d845c7)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)
  • Sprint 2 Wave A — eval runner, 1M context GA, RFC 9457 agent errors (v7.6.0) (#1067) (12968a5)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

Miscellaneous

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)

[7.9.0] - 2026-03-15

Added

  • monitoring-observability: Langfuse Python SDK v4 coverage — as_type semantic types (generation, retriever, chain, embedding, agent, guardrail, evaluator), score_current_span() / score_current_trace(), should_export_span filter, LangfuseMedia attachments, run_experiment() batch evaluation
  • monitoring-observability: Python v3→v4 migration guide section with breaking changes checklist, code examples, and version pin guidance (langfuse>=4.0.0)

Changed

  • monitoring-observability: v2.0.0→3.0.0 — all code examples updated from v3 to v4 patterns across rules, scripts, checklists, and references
  • monitoring-observability: Setup checklist updated with v4 version pins and migration steps
  • brainstorm: v4.4.0→4.6.0 — added product/event-driven/devops/data-pipeline domain detection, design pipeline agents, /effort scaling, CC 2.1.76 partial results + PostCompact recovery

[7.8.0] - 2026-03-14

Added

  • hooks: PostCompact recovery hook — re-injects branch, tasks, decisions after compaction using CC-provided compaction_count and context_size_after fields
  • hooks: Elicitation guard — blocks form-mode MCP elicitations requesting secret fields (api_key, password, token), forces URL mode
  • hooks: ElicitationResult logger — tracks accept/decline/cancel outcomes for analytics, injects alternative suggestions on decline
  • hooks: Effort-aware context budgeting — detectEffortLevel() reads /effort setting, adjusts prompt dispatcher token budget (low=200t, medium=800t, high=1200t), skips heavy context at low effort
  • hooks: Monorepo sparse paths advisory — WorktreeCreate detects monorepo (pnpm-workspace, nx, lerna, turbo, workspaces) and suggests worktree.sparsePaths config
  • types: PostCompact, Elicitation, ElicitationResult added to HookEvent union with typed input fields
  • version-matrix: 23 new feature entries covering CC 2.1.75 and 2.1.76 (elicitation, sparse paths, /effort, bg agent partial results, compaction circuit breaker, etc.)

Changed

  • hooks: MIN_CC_VERSION bumped from 2.1.74 to 2.1.76 — backward compatibility dropped
  • hooks: SubagentStop retry-handler now returns early with [PARTIAL RESULT] tag for killed background agents instead of entering retry logic (CC 2.1.76 partial results)
  • hooks: PostCompact and Elicitation hooks registered in entries/lifecycle.ts split bundle (lifecycle.mjs grew 77→80KB)
  • compatibility: All 78 skill compatibility: fields updated from 2.1.74+ to 2.1.76+
  • counts: 99→104 hooks (33→36 global), MIN_CC_VERSION 2.1.74→2.1.76

Fixed

  • docs: Missing v7.7.x row in version history table (between v7.8.x and v7.5.x)
  • docs: Stale "CC 2.1.74 format" agent reference in CLAUDE.md directory structure
  • docs: README badge and requirement updated from ≥2.1.74 to ≥2.1.76
  • docs: hooks README footer CC requirement updated to >= 2.1.76

[7.7.0] - 2026-03-14

Added

  • design-to-code: New user-invocable skill (/ork:design-to-code) — mockup-to-component pipeline using Google Stitch MCP and 21st.dev component registry
  • component-search: New user-invocable skill (/ork:component-search) — search 21st.dev registry for production-ready React components
  • design-context-extract: New user-invocable skill (/ork:design-context-extract) — extract design DNA (colors, typography, spacing) from screenshots or URLs
  • design-context-extractor: New agent for autonomous design context extraction with stitch-mcp integration
  • component-curator: New background agent that audits project component usage and recommends 21st.dev upgrades
  • MCP: Added stitch-mcp (Google Stitch) and 21st-dev-magic (21st.dev component registry) to MCP configuration docs

Changed

  • frontend-ui-developer: Added 21st-dev-magic MCP scope + component-search, design-to-code, design-context-extract skills
  • design-system-architect: Added stitch-mcp MCP scope + design-context-extract, component-search skills
  • counts: 89→92 skills, 31→33 agents, 99→101 hooks, 16→19 user-invocable

[7.6.0] - 2026-03-14

Added

  • api-design: Agent-facing RFC 9457 error patterns with structured retry signals — retryable, error_category, retry_after, owner_action_required extensions for deterministic AI agent error handling
  • api-design: Content negotiation middleware pattern (JSON/Markdown/HTML via Accept header) with token efficiency budget (<300 tokens)
  • api-design: Error category taxonomy (10 categories) with agent action mapping table
  • types: StructuredError and ErrorCategory types in orchestration-types.ts for structured error handling in hooks
  • testing-integration: RFC 9457 error response assertion patterns (Python + TypeScript)

Changed

  • hooks: retry-manager now extracts RFC 9457 structured errors before regex fallback — uses server-provided retry_after instead of calculated backoff
  • hooks: failure-handler detects structured errors for richer context injection with error_category and what_you_should_do

Fixed

  • python-backend: Updated stale RFC 7807 reference to RFC 9457 (the active standard)

[7.5.3] - 2026-03-13

Added

  • eval: Quality evaluation runner (npm run eval:quality) — A/B skill vs baseline grading with Claude-as-judge
  • eval: Seed eval YAMLs for all 16 user-invocable skills with trigger + quality assertions
  • eval: Shared eval library (lib/eval-common.sh) — colors, cleanup, deps, portable timeout
  • eval: --grade-only, --tag, --max-turns, --timeout flags for both eval runners
  • eval: Batch assertion grading (A3) reduces Claude calls ~60%
  • eval: Parallel with-skill + baseline execution (A4) for faster evals
  • eval: Scaffold field support — quality evals can declare project scaffolds (e.g., typescript-nextjs)

Fixed

  • eval: SEC-001 path traversal in skill name input validation
  • eval: SEC-002 JSON injection via unsanitized YAML-sourced strings
  • eval: PASS/FAIL grading anchored to first token to prevent false matches
  • eval: Silent claude CLI crashes now emit warnings
  • eval: grep -w with multi-word patterns replaced by ERE

Changed

  • eval: Trigger runner refactored to source shared eval-common.sh
  • eval: All 17 eval YAMLs hardened with adversarial prompts, confusion pairs, and negative assertions

[7.5.2] - 2026-03-12

Added

  • eval: Trigger evaluation runner (npm run eval:trigger) — measures skill trigger precision and recall with flaky detection
  • docs: Expanded CONTRIBUTING-SKILLS.md with mandatory eval requirements, writing guidance, and pre-submit checklist

7.5.1 (2026-03-12)

Bug Fixes

  • docs: update CC version requirement from 2.1.72 to 2.1.74 (#1062) (622ba4c)

7.5.0 (2026-03-12)

Features

  • skills: adopt CC 2.1.72 features — ExitWorktree, model override (#1036) (#1060) (96fd3a0)

Bug Fixes

  • docs: update generated plugins-data version to 7.4.0 (#1058) (e2693b1)
  • skills: add ExitWorktree + model override to Agent Teams teardown paths (#1061) (c326666)

7.4.0 (2026-03-12)

Features

  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)
  • skills: skills health — CC 2.1.72 alignment (all phases) (#1040) (b9db695)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)
  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)

Miscellaneous

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)

[7.5.0] - 2026-03-12

Changed

  • platform: bump minimum Claude Code version from >= 2.1.73 to >= 2.1.74
    • SessionEnd hook.timeout now respected (was hardcoded to 1.5s — hooks silently killed)
    • New CLAUDE_CODE_SESSIONEND_HOOKS_TIMEOUT_MS env var for timeout override
    • Managed policy ask rules now correctly override user allow + skill allowed-tools
    • Full model IDs (claude-opus-4-6) accepted in agent frontmatter (symbolic names still recommended)
    • /context command surfaces actionable optimization suggestions
    • autoMemoryDirectory setting for custom auto-memory storage path
    • --plugin-dir local overrides take precedence over marketplace plugins
    • Streaming API + bash prefix caching memory leaks fixed
  • skills: update compatibility field to "Claude Code 2.1.74+" across 78 skills
  • docs: update version matrix with 12 new features (4 for 2.1.73, 8 for 2.1.74) and "Full+++++" compatibility level

[7.4.0] - 2026-03-11

Changed

  • platform: bump minimum Claude Code version from >= 2.1.72 to >= 2.1.73
    • Fixes skill-file deadlock on git pull (critical for 89-skill plugin)
    • Fixes SessionStart hooks double-firing on --resume/--continue
    • Fixes no-op system reminder injection (~2K tokens/turn recovered across 99 hooks)
    • Opus 4.6 now default on Bedrock/Vertex/Foundry (6 opus-tier agents auto-upgrade)
  • skills: update compatibility field to "Claude Code 2.1.73+" across 78 skills
  • docs: update version matrix with 5 new 2.1.73 features and "Full++++" compatibility level

Fixed

  • docs: fix broken light mode on documentation site — split dark-only @theme into proper light/dark token sets following fumadocs convention (@theme for light defaults, .dark {} for dark overrides)

[7.3.3] - 2026-03-11

Added

  • skills: shared unified scoring framework in quality-gates for assess/verify
  • skills: 33 rule files across 12 skills (setup, implement, brainstorm, remember, help, audit-full, configure, errors, memory-fabric, release-management, scope-appropriate-architecture, testing-patterns)
  • skills: 12 _sections.md catalog files for new rule directories
  • hooks: task-existence-gate for task tracking enforcement in sync-task-dispatcher
  • agents: background: true on 8 read-only analysis agents

Changed

  • skills: compress 12 bloated skills under 300 lines via reference extraction (setup, write-prd, code-review-playbook, zustand-patterns, release-management, vite-advanced, async-jobs, devops-deployment, audit-full, implement, react-server-components-framework, remember)

Fixed

  • hooks: improve subagent-quality-gate and unified-dispatcher

7.3.2 (2026-03-11)

Bug Fixes

  • security: resolve CodeQL and Scorecard alerts (#1027) (ab9136c)

Miscellaneous

  • fix hono CVE + organize playgrounds by date (#1039) (b43cf2d)

[7.3.2] - 2026-03-11

Security

  • hooks: fix second-order command injection in gitExec() (CodeQL #124, High) — add assertSafeGitArgs() blocklist for --upload-pack, --receive-pack, --exec, -c options that can execute arbitrary commands via git
  • ci: fix unpinned npm dependency in claude-health.yml (Scorecard #125, Medium) — add --ignore-scripts to npm install and bump to @2.1.72

7.3.1 (2026-03-11)

Miscellaneous

  • deps-dev: bump the npm_and_yarn group across 2 directories with 1 update (#1025) (f924b27)

7.3.0 (2026-03-10)

Features

  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)
  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)

Miscellaneous

Code Refactoring

  • split testing-patterns, CC 2.1.72 upgrade, HTTP hooks fix (#1015) (7bcab43)

CI/CD

  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)

[7.3.0] - 2026-03-10

Fixed

  • agents: remove stale git-workflow references from git-operations-engineer, frontend-ui-developer, release-engineer
  • agents: remove stale ci-automation reference from ci-cd-engineer
  • skills: fix broken hook refs in commit skill (2 non-existent hooks → git-validator)
  • skills: remove broken audit-context-loader hook section from audit-full skill
  • skills: fix stale cross-references in help, storybook-testing, release-management, verify
  • hooks: fix verify scoring rubric weights (6→8 dimensions matching quality-model.md)
  • tests: fix test-git-enforcement-hooks.sh references to deleted skills
  • tests: fix trigger-cases.yaml — replace git-workflow with github-operations
  • tests: delete always-failing test-git-workflow-completeness.sh

Changed

  • hooks: skill-scoped hooks 23→21 (removed 2 non-existent refs), total 101→99
  • verify: add cross-route findings panel to gallery template (common issues, strengths, breakdown)
  • verify: enhance visual-capture.md with 6-criteria structured AI evaluation prompt
  • docs: update all hook count references across 10+ doc files

[7.2.2] - 2026-03-10

Removed

  • skills: remove git-workflow skill (5 best rules migrated to commit skill)
  • skills: remove ci-automation skill (ci-cd-engineer agent covers this)
  • evals: remove orphaned git-workflow eval files

Changed

  • skills: upgrade write-prd to v2.0.0 — AskUserQuestion, Task Management, memory integration, disk output
  • skills: upgrade commit to v1.2.0 — absorb 5 git-workflow rules (branch-protection, merge-strategy, history-hygiene, stacked-pr-workflow, stacked-pr-rebase)
  • skills: update compatibility field to 2.1.72+ across 77 SKILL.md files
  • marketplace: bump engine from >=2.1.69 to >=2.1.72

Added

  • skills: split testing-patterns into 5 focused sub-skills — testing-unit, testing-e2e, testing-integration, testing-llm, testing-perf (#669)
  • skills: add visual verification portfolio to verify skill — Phase 2.5 (screenshot gallery) + Phase 8.5 (agentation feedback loop)
  • skills: gallery-template.html — self-contained HTML with dark/light mode, base64-embedded screenshots, AI vision evaluations
  • hooks: 4 once:true standards loaders for implement, review-pr, verify, brainstorm skills (CC 2.1.72 prompt cache optimization)
  • hooks: implement-standards-loader — dynamic project stack detection via package.json/pyproject.toml
  • hooks: review-dimensions-loader — static 6-dimension review criteria + JSON output contract
  • hooks: verify-scoring-rubric-loader — static 0-10 scale with dimension weights
  • hooks: brainstorm-instructions-loader — static divergent mode rules + evaluation dimensions

Fixed

  • evals: update help.eval.json from "5 GIT skills" to "4 GIT skills"
  • tests: update split-bundles expected hook count to 176

7.2.1 (2026-03-09)

Bug Fixes

  • docs: update generated plugins-data version to 7.2.0 (#1012) (dca7243)

CI/CD

  • bump actions/setup-node from 6.2.0 to 6.3.0 (#1008) (61f73f5)
  • bump github/codeql-action from 4.32.4 to 4.32.6 (#1009) (13a0975)

7.2.0 (2026-03-09)

Features

  • add ci-automation skill + GitHub Actions workflows (#1006) (8b328a7)
  • skills: add 13 cognitive-science UX rules across 5 existing skills (#1011) (91b62ac)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)

Miscellaneous

[7.1.14] - 2026-03-08

Fixed

  • skills: add 234 test case entries for 222 previously untested rules across 63 skills
  • skills: fix 18 term mismatches in expectedBehavior strings for rule traceability
  • skills: scaffold 49 .eval.json files and regenerate all 82 eval files from current test-cases
  • skills: fix remaining short behavior in animation-motion-design view-transitions

[7.1.13] - 2026-03-08

Fixed

  • ci: resolve TypeScript type error in home-environment test (tuple annotation mismatch)
  • ci: update stale skill names in trigger tests — brainstorming→brainstorm, plan-viz→visualize-plan
  • skills: add skill name fragments to 6 descriptions for semantic discovery (help, feedback, business-case, prioritization, setup, write-prd)
  • skills: add chain-patterns Overview section and trigger phrasing
  • skills: expand 546 short expectedBehavior entries (≤5 words) to 6+ words across 43 skills
  • skills: add context exhaustion handling docs to brainstorm and task-dependency-patterns
  • ci: add id-token: write to claude-review and claude-triage workflows for OIDC
  • ci: fix SessionStart hook count assertion (4→3 after HTTP webhook removal)
  • skills: strip ork: prefix from subagent_type in chain-patterns, fix-issue, implement

[7.1.12] - 2026-03-08

Fixed

  • hooks: InstructionsLoaded dispatcher — shared content cache eliminates 3x redundant file reads
  • hooks: classifySource uses basename() check to prevent substring misclassification
  • hooks: rule-conflicts false-positive prevention — files matching both sides excluded
  • hooks: drift-detection conditional cache write — only writes when hashes differ
  • hooks: token-budget-tracker simplified threshold to percentage-only

Changed

  • hooks: all 6 InstructionsLoaded handlers now receive shared Map<string, string> content cache
  • hooks: dispatcher validates files_loaded elements with runtime type guard

Added

  • tests: 12 new tests for edge cases, all 6 conflict patterns, false-positive prevention

[7.1.11] - 2026-03-07

Added

  • skill: ci-automation skill with SKILL.md + 4 reference docs (cost-model, gh-action-patterns, headless-cli, security)
  • ci: GitHub Actions workflows for claude-health, claude-review, claude-triage
  • agent: ci-cd-engineer skill index updated with ci-automation

7.1.10 (2026-03-06)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)
  • ci: add plugins[0].version to release-please extra-files (#1001) (8a4faad)
  • docs: replace broken ASCII diagrams with fumadocs Steps (#999) (2c1b6fe)

Miscellaneous

[7.1.9] - 2026-03-06

Fixed

  • Add $.plugins[0].version to release-please extra-files so release PRs update both top-level and plugin versions in marketplace.json

[7.1.8] - 2026-03-06

Fixed

  • Replace broken ASCII box-drawing diagrams with fumadocs Steps components and markdown tables across 12 docs pages
  • Add missing /ork:prd command to command-skills page
  • Fix stale skill counts and clarify "listed vs invocable" text in skills overview
  • Fix skill.mjs bundle hook count from "1+" to "19" in hooks overview

7.1.7 (2026-03-06)

Bug Fixes

  • ci: add bootstrap-sha to release-please config (#997) (0d46843)

[7.1.6] - 2026-03-06

Fixed

  • TODO: Describe your changes here

[7.1.5] - 2026-03-06

Fixed

  • docs: Full fumadocs sync — fix 40+ stale references across 37 MDX files
    • All counts corrected: 79 skills (18 cmd + 61 ref), 30 agents, 105 hooks
    • Remove 8 demoted agent references from 15 files (replaced with current agents)
    • Remove deleted hook references (graph-memory-inject, agent-memory-store, capture-user-intent, memory-capture) from 8 files
    • Rewrite memory docs for graph-first architecture (session-summary replaces 4 hooks)

Added

  • docs: Document CC 2.1.69/70 hook features in fumadocs
    • ConfigChange, InstructionsLoaded, WorktreeCreate/Remove events in hook timeline
    • Native HTTP hooks section (12 type:"http" hooks)
    • once:true auto-removal pattern (13 skill context loaders)
    • statusMessage spinner UX documentation

[7.1.4] - 2026-03-06

Fixed

  • docs: Fix 30+ stale counts across fumadocs site, skills, and source files
    • 71 skills → 79 skills, 83/98/106 hooks → 105 hooks (42 global + 44 agent + 19 skill)
    • CC version badge ≥2.1.59 → ≥2.1.69, user-invocable 17 → 18
    • Fix count-hooks.sh to exclude CONTRIBUTING-SKILLS.md example (was inflating skill count by 1)
    • Regenerated docs data + rebuilt plugins to match

[7.1.3] - 2026-03-05

Added

  • feat(skills): Enable CC native matching on 10 knowledge skills (#960)
    • security-patterns, api-design, testing-patterns, database-patterns, python-backend, architecture-patterns, performance, quality-gates, react-server-components-framework, devops-deployment
    • Changed disable-model-invocation: true → false so CC auto-selects these based on prompt/description matching
    • Skills visible to CC: 17 → 27 (17 user-invocable + 10 model-invocable)
  • docs(skills): Document model-invocable pattern in CONTRIBUTING-SKILLS.md

Removed

  • perf(hooks): Delete ~1,550 lines dead routing code (#960)
    • Deleted: intent-classifier.ts (669 lines), calibration-engine.ts, skill-nudge.ts, calibration-tracker.ts, calibration-persist.ts, + 4 test files
    • These modules were superseded by CC native skill matching and never ran in production
    • Prompt pipeline hooks: 5 → 4 (removed skill-nudge-prompt)
    • Split-bundles test count: 170 → 167

Changed

  • perf(agents): Demote 8 redundant agents to reduce routing overhead (#863)
    • Removed: requirements-translator, documentation-specialist, prompt-engineer, ux-researcher, rapid-ui-designer, business-case-builder, metrics-architect, prioritization-analyst
    • Agent count: 38 to 30, Hook count: 98 to 88 (10 agent-scoped hooks removed)
    • Reduces routing complexity and eliminates ~98k tokens/invocation for unnecessary agent spawns
  • docs(build): Document skills/commands duplication token waste (#889) with upstream CC bug tracking

Removed

  • CC-v3 feature issues (#930-933) removed from perf milestone — already built, architecturally impossible, or superseded by CC native tools

[7.1.2] - 2026-03-04

Added

  • browser-tools skill v3.0.0: 7 new sections documenting agent-browser v0.16 features (semantic locators, mouse commands, tab management, debug/recording, mobile testing, configuration flags, auth vault)
  • 2 new rules: browser-debug-recording.md (trace/profiler safety), browser-mobile-testing.md (device emulation verification)
  • 3 new safety hook checks: --allow-file-access warning, AGENT_BROWSER_ENCRYPTION_KEY leak prevention, --user-agent spoofing warning
  • 4 new hook test cases (14/14 passing)
  • Updated 4 agents (web-research-analyst, test-generator, accessibility-specialist, frontend-ui-developer) with v0.16 capabilities
  • Expanded 4 existing rules with v0.16 flags and patterns

[7.1.1] - 2026-03-04

Fixed

  • NotebookLM MCP bloat (#890): Disabled notebooklm-mcp at project level (disabled: true in .mcp.json), saving ~6k tokens/session from 25 unused tools
  • Tool name mismatch: Fixed mcp__notebooklm__* → mcp__notebooklm-mcp__* across release-notebook skill, notebooklm rules, and create-release-notebook.sh
  • Release-notebook flow: Wired release-notebook skill into release-engineer agent; fixed shell script instructions (removed non-existent /ork:release-notebook command reference)
  • Perf test allowlist: Updated test-mcp-overhead.sh to reference release-engineer instead of phantom release-notebook agent

[7.1.0] - 2026-03-04

Added

  • Full agent-browser CLI adoption (#947 #948 #949): 64 commands across 12 categories documented in browser-tools skill
  • Interaction commands: click, dblclick, focus, fill, type, keyboard, press, hover, check/uncheck, select, scroll, scrollintoview, drag, upload (22 commands)
  • Storage API (v0.13): localStorage/sessionStorage get, set, clear
  • Enhanced capture: screenshot --full, screenshot --annotate, pdf export
  • Extraction: eval, eval -b, eval --stdin
  • Action confirmation (v0.15): native --confirm-interactive with env-gated AGENT_BROWSER_NATIVE_CONFIRM
  • Network route validation: agent-browser-safety hook Check 5 validates route URLs against blocklist
  • 6 agents updated: web-research-analyst, test-generator, accessibility-specialist, frontend-ui-developer, rapid-ui-designer, frontend-performance-engineer
  • 3 rules updated: browser-auth-security, browser-rate-limiting, browser-snapshot-workflow
  • Configurable settings wizard (#951): setup skill writes to .claude/settings.json for non-orchestkit projects
  • Performance sprint: injection gating, CI gate, perf-compare, perf-snapshot hook (#865 #944 #945 #946)

[7.0.3] - 2026-03-04

Fixed

  • Assessment quality gaps (#946): address code-quality-reviewer blockers in verify skill

[7.0.2] - 2026-03-03

Added

  • Session handoffs (#464): Structured YAML handoffs on SessionEnd with auto-injection on SessionStart for cross-session continuity
  • Release notebook skill (#836): release-notebook skill creates versioned NotebookLM notebooks per release with changelog, manifest, and CLAUDE.md sources
  • Release notebook script (#836): scripts/create-release-notebook.sh for manual/CI invocation

Changed

  • Native HTTP hooks (#896): Converted 3 HQ reporters (session-end, worktree-create, worktree-remove) from type:command to type:http — eliminates ~200ms Node spawn per fire
  • Hook count: 95 -> 97 (36 global + 54 agent-scoped + 7 skill-scoped)
  • Skill count: 69 -> 70

[7.0.1] - 2026-03-03

Fixed

  • Add agents/hooks paths to plugin.json and build script (#900)
  • Add metadata.description and tags to marketplace.json (#900)
  • Fix 3 agents with empty mcpServers that use memory skills (#887)
  • Fix localhost regex false positive on *.localhost TLD (#935)

[7.0.0] - 2026-02-26

Breaking Changes

  • plugins: Unified plugin architecture — orkl and ork-creative removed, consolidated into single ork plugin (69 skills, 38 agents, 96 hooks). Migration: claude plugin uninstall orkl && claude plugin install ork.

Added

  • setup: New setup wizard skill (/ork:setup) — scans codebase, detects stack, recommends skills and MCPs, generates readiness score
  • notifications: Satisfaction detector bug nudge — suggests /ork:feedback bug after 2+ negative signals per session
  • mcp: NotebookLM MCP integration — skill, hooks, rules for NotebookLM workflows
  • release: Release channels (alpha/beta/stable) with branch-based CI workflows and --ref marketplace flag
  • settings: 8 chord keybindings for top skills (ctrl+k ctrl+f → fix-issue, ctrl+k ctrl+i → implement, etc.)
  • settings: StatusLine showing ork version, git branch, and uncommitted file count
  • settings: 10 spinner tips for skill discovery (e.g., "/ork:memory search to find past decisions")
  • skills: AskUserQuestion markdown previews on 12 skills — side-by-side ASCII diagrams showing what each mode does (assess, verify, explore, review-pr, fix-issue, audit-full, help, memory, brainstorming, implement, plan-viz, setup)
  • skills: multiSelect: true on audit-full (combinable audit modes), verify (combinable checks), setup (combinable focus areas)
  • skills: EnterPlanMode "Plan first" option on implement, fix-issue, brainstorming — research before coding
  • skills: Skill listing demotion — 58 → 16 listed skills via disable-model-invocation: true (~2,630 tokens/turn saved)
  • tests: test-settings-validation.sh — validates keybindings, statusLine, spinnerVerbs, spinnerTips, permissions
  • tests: test-skill-cc-features.sh — validates AUQ previews, multiSelect, EnterPlanMode, argument-hints, listing balance

Changed

  • settings: Unified all plugin settings: identical spinner verbs and permissions across all plugins
  • notifications: Desktop notification sound handling delegated to dedicated sound.ts module with command cache for performance
  • hooks: Unified notification dispatcher with simplified error filtering and HookFn type usage
  • hooks: Replaced 2 HTTP session-end hooks with command hook (HMAC-signed POST)
  • hooks: Consolidated 55 → 37 global hooks via 6 sync dispatchers, then expanded to 96 total (35 global + 54 agent-scoped + 7 skill-scoped)
  • manifests: Composite index budget bumped from 16KB to 20KB to accommodate unified plugins

Fixed

  • Sound notifications now use detached spawn to survive async hook exit
  • Bump Hono from 4.12.0 to 4.12.2 (CVE-2026-27700)
  • 6 Bash security hook vulnerabilities fixed (#879-#883)

[6.7.2] - 2026-02-26

Fixed

  • Bump skill count 68 → 69 across manifests and docs
  • Bump Hono 4.12.0 → 4.12.2 (CVE-2026-27700)
  • Bump to v6.7.2

[6.7.1] - 2026-02-25

Breaking Changes

  • skills: Remove worktree-coordination skill — fully superseded by native CC 2.1.56 EnterWorktree tool and Task(isolation:"worktree") (#830)

Features

  • eval: Overhaul eval skills, agents, and LangGraph/Langfuse compatibility
  • skills: Extract create-pr into 7 rules + 4 references (274→156 lines)
  • skills: Compress 8 bloated skills (demo-producer, plan-viz, doctor, fix-issue, skill-evolution, verify, memory, assess) with content preserved in rules/references
  • testing: Add 3 new test suites: agent skill refs, build drift, cross-skill refs

Fixed

  • Fix 20 broken cross-skill references across 12 skills
  • Fix 5 weak skill descriptions (configure, feedback, help, release-checklist, skill-evolution)
  • Bump minimum Claude Code version to 2.1.56
  • Resolve all verification findings (Langfuse v2 remnants, credential warnings, sync guards)

6.7.0 (2026-02-23)

Features

  • dead code cleanup & PostToolUse consolidation (#684) (#790) (da245ac)
  • docs: add Vercel Analytics, Speed Insights, and dynamic social proof (#809, #810) (#812) (714af85)
  • integrate Claude Code 2.1.50 features (#805) (60a16d2)
  • security hardening & hook consolidation (Milestone #72) (#788) (5f014b4)
  • skills: add presentation-builder skill (#817) (93cacda)

Bug Fixes

  • add plugins/ plugin.json to release-please extra-files (#822) (8eb27cb)
  • eliminate 32 code scanning alerts (19 CodeQL + 13 Scorecard) (#792) (12e796b)
  • eliminate 91 CodeQL alerts + fix 30 pre-existing test failures (#782) (ca47641)
  • release: scope release-please from v6.1.0 to prevent 7.0.0 (#786) (83a248b)
  • remove 35 broken symlinks from .claude/ (#813) (3bd179c)
  • resolve 21 code scanning alerts (17 CodeQL + 4 Scorecard) (#825) (00e9ed8)
  • sanitize shell arg in security-scan-aggregator bandit command (#828) (18c0415)
  • sync manifests and pyproject.toml to v6.3.0 (#806) (d06661d)
  • use jq for plugin.json generation to match release-please format (#824) (0a37a07)

Miscellaneous

CI/CD

  • bump actions/attest-build-provenance from 2.4.0 to 3.2.0 (#816) (04a2cee)
  • bump github/codeql-action from 4.32.3 to 4.32.4 (#815) (e31096e)
  • pin package manager dependencies for Scorecard compliance (#827) (45dec20)
  • restrict workflow-level permissions to read-all in release-please (#820) (8e03fe5)

[6.6.3] - 2026-02-23

Fixed

  • Sanitize resultsDir in security-scan-aggregator bandit command (CodeQL alert #43)

[6.6.2] - 2026-02-23

Changed

  • Use npm ci instead of npm install in CI for lockfile-based integrity hashes
  • Remove unnecessary Python/pip dependency from schema validation (uses built-in basic validator)

[6.6.1] - 2026-02-23

Fixed

  • TODO: Describe your changes here

6.6.0 (2026-02-23)

Features

  • dead code cleanup & PostToolUse consolidation (#684) (#790) (da245ac)
  • docs: add Vercel Analytics, Speed Insights, and dynamic social proof (#809, #810) (#812) (714af85)
  • integrate Claude Code 2.1.50 features (#805) (60a16d2)
  • security hardening & hook consolidation (Milestone #72) (#788) (5f014b4)
  • skills: add presentation-builder skill (#817) (93cacda)

Bug Fixes

  • add plugins/ plugin.json to release-please extra-files (#822) (8eb27cb)
  • eliminate 32 code scanning alerts (19 CodeQL + 13 Scorecard) (#792) (12e796b)
  • eliminate 91 CodeQL alerts + fix 30 pre-existing test failures (#782) (ca47641)
  • release: scope release-please from v6.1.0 to prevent 7.0.0 (#786) (83a248b)
  • remove 35 broken symlinks from .claude/ (#813) (3bd179c)
  • sync manifests and pyproject.toml to v6.3.0 (#806) (d06661d)
  • use jq for plugin.json generation to match release-please format (#824) (0a37a07)

Miscellaneous

CI/CD

  • bump actions/attest-build-provenance from 2.4.0 to 3.2.0 (#816) (04a2cee)
  • bump github/codeql-action from 4.32.3 to 4.32.4 (#815) (e31096e)
  • restrict workflow-level permissions to read-all in release-please (#820) (8e03fe5)

[6.5.3] - 2026-02-23

Fixed

  • TODO: Describe your changes here

[6.5.2] - 2026-02-23

Fixed

  • TODO: Describe your changes here

[6.5.1] - 2026-02-23

Features

  • Add presentation-builder skill — zero-dependency HTML slide presentations with viewport fitting, 12 style presets, PPT conversion, and keyboard/touch navigation (adopted from zarazhangrui/frontend-slides, restructured into OrchestKit format)
  • Skill count: 67 → 68 skills, 28 → 29 user-invocable

6.5.0 (2026-02-22)

Features

  • dead code cleanup & PostToolUse consolidation (#684) (#790) (da245ac)
  • docs: add Vercel Analytics, Speed Insights, and dynamic social proof (#809, #810) (#812) (714af85)
  • integrate Claude Code 2.1.50 features (#805) (60a16d2)
  • security hardening & hook consolidation (Milestone #72) (#788) (5f014b4)

Bug Fixes

  • eliminate 32 code scanning alerts (19 CodeQL + 13 Scorecard) (#792) (12e796b)
  • eliminate 91 CodeQL alerts + fix 30 pre-existing test failures (#782) (ca47641)
  • release: scope release-please from v6.1.0 to prevent 7.0.0 (#786) (83a248b)
  • remove 35 broken symlinks from .claude/ (#813) (3bd179c)
  • sync manifests and pyproject.toml to v6.3.0 (#806) (d06661d)

Miscellaneous

[6.4.1] - 2026-02-22

Fixed

  • TODO: Describe your changes here

6.4.0 (2026-02-22)

Features

  • docs: add Vercel Analytics, Speed Insights, and dynamic social proof (#809, #810) (#812) (714af85)

Bug Fixes

  • sync manifests and pyproject.toml to v6.3.0 (#806) (d06661d)

6.3.0 (2026-02-21)

Bug Fixes

  • sync manifests and pyproject.toml to v6.3.0 (release-please gap)

Features

Bug Fixes

  • eliminate 32 code scanning alerts (19 CodeQL + 13 Scorecard) (#792) (12e796b)
  • eliminate 91 CodeQL alerts + fix 30 pre-existing test failures (#782) (ca47641)
  • release: scope release-please from v6.1.0 to prevent 7.0.0 (#786) (83a248b)

Miscellaneous

[6.2.1] - 2026-02-21

Added

  • WorktreeCreate and WorktreeRemove hook events with lifecycle logger handler (CC 2.1.50)
  • isolation: worktree frontmatter on test-generator, demo-producer, git-operations-engineer agents
  • CC feature matrix entries for 2.1.49 (EnterWorktree tool) and 2.1.50 (7 features)
  • Doctor skill memory leak warning for CC < 2.1.50 (8 leaks fixed)
  • Doctor agent registration health check using claude agents CLI (CC 2.1.50)
  • audit-full guard for CLAUDE_CODE_DISABLE_1M_CONTEXT env var

Changed

  • Hook counts: 52 → 54 global, 75 → 77 total, 15 → 17 event types
  • upgrade-assessment skill updated with 2.1.50 breaking changes and memory leak note
  • worktree-coordination skill updated with lifecycle hooks section and compatibility bump to 2.1.50

[6.1.6] - 2026-02-20

Security

  • Fix 16 CodeQL shell injection alerts — convert execSync template strings to execFileSync array form
  • Fix 2 CodeQL incomplete string escaping alerts (issue-work-summary, session-cleanup)
  • Fix 1 CodeQL DOM XSS alert in coverage report (innerHTML → createElement)
  • Add shell.ts helpers: shellEscape(), isValidGitRef(), safeGitExec()
  • Harden GitHub Actions workflows: add explicit permissions blocks (least-privilege)
  • Pin jsonschema dependency version in plugin-validation workflow

[6.1.5] - 2026-02-20

Fixed

  • Harden Scorecard workflow security: add explicit job-level permissions to release-please job (least-privilege)
  • Pin jsonschema dependency to v4.23.0 in plugin-validation workflow (Pinned-Dependencies alert)
  • Add --no-audit flags to npm commands in ci-setup.sh for reproducible builds

6.2.0 (2026-02-20)

Features

Bug Fixes

  • eliminate 91 CodeQL alerts + fix 30 pre-existing test failures (#782) (ca47641)
  • release: scope release-please from v6.1.0 to prevent 7.0.0 (#786) (83a248b)

Miscellaneous

[6.1.4] - 2026-02-20

Changed

  • Consolidate PostToolUse hooks.json entries from 4 to 1 async dispatcher (#684)
  • Move redact-secrets, config-change-auditor, team-member-start, error-logger into unified-dispatcher
  • Strip solution-suggestion machinery from unified-error-handler (~250 LOC)

Removed

  • Dead hooks: context-budget-monitor (225 LOC), posttool/skill-nudge (47 LOC), context-compressor (159 LOC)
  • Dead context/ files: 7 agent JSONs, 1 orphaned archive session
  • Dead path helpers: getMcpDeferStateFile, getErrorSuggestionsDedupFile

Fixed

  • skill-edit-tracker path bug (missing context/ segment in session state path)

6.2.0 (2026-02-20)

Features

  • security hardening & hook consolidation (Milestone #72) (#788) (5f014b4)

6.1.3 (2026-02-20)

Bug Fixes

  • eliminate 91 CodeQL alerts + fix 30 pre-existing test failures (#782) (ca47641)
  • release: scope release-please from v6.1.0 to prevent 7.0.0 (#786) (83a248b)

[6.1.2] - 2026-02-20

Fixed

  • Stale model name in cost-estimation.md (claude-sonnet-4-5 → claude-sonnet-4-6)
  • CC version badge and FAQ updated from >=2.1.34 to >=2.1.49

[6.1.1] - 2026-02-20

Fixed

  • Release-please configuration scoped to prevent major version bump

6.1.0 (2026-02-20)

Features

  • CC 2.1.49 full adoption — hooks, agents, MCP overhaul (#780) (1234c46)

[6.0.27] - 2026-02-20

Security

  • Eliminate 30 CodeQL ReDoS vulnerabilities (js/polynomial-redos) across 12 hook files
  • Replace polynomial-time regex patterns with O(n) string operations
  • Dismiss 1 Dependabot alert (ajv ReDoS in demo project — tolerable risk)

6.0.26 (2026-02-20)

Miscellaneous

CI/CD

  • 2026 best practices — permissions, CodeQL, SLSA, merge queue (57ca595)

6.0.25 (2026-02-19)

Bug Fixes

  • ci: skip version check for Dependabot PRs (#758) (45c850f)
  • ci: use PAT for release-please to trigger CI on release PRs (#760) (5e8c700)

CI/CD

  • bump actions/download-artifact from 5.0.0 to 7.0.0 (#755) (c28f49b)
  • bump actions/upload-artifact from 5.0.0 to 6.0.0 (#753) (4a4f9e0)
  • bump gitleaks/gitleaks-action (#756) (b929e1b)
  • bump googleapis/release-please-action (#754) (8028579)

[6.0.24] - 2026-02-19

Fixed

  • Release automation — rewrite release-please for v4 manifest mode; was failing on every push to main since v6.0.17 due to invalid inline params (package-name, changelog-types, extra-files silently ignored by v4)
  • Version sync — create version.txt + .release-please-config.json with jsonpath for package.json and marketplace.json auto-update
  • Version drift — sync package.json 6.0.20 and manifest 4.27.5 to match CHANGELOG 6.0.23
  • version-check — skip validation for release-please automated PRs

[6.0.23] - 2026-02-19

Added

  • CC 2.1.47 Full Adoption — Milestone #75 (19 issues: #710–#728)
  • cc-version-matrix.ts — runtime feature compatibility matrix for 18 CC 2.1.47 features (last_assistant_message, added_dirs, deferred SessionStart, agent model in teams, worktree discovery, etc.)
  • added_dirs support — session-tracking, session-cleanup, session-context-loader, memory-capture, and monorepo-detector hooks now read and log the new added_dirs hook input field
  • last_assistant_message support — memory-capture uses classifySessionOutcome() for richer analytics; session-cleanup and unified-dispatcher log last_msg_len
  • Prefill-guard caching — scan results cached for 24 hours with CACHE_VERSION invalidation, skipping expensive filesystem scans on repeat sessions
  • Worktree discovery tests — 15 vitest path-resolution tests + 17 shell tests with real git worktree add creation/cleanup
  • Agent model field validation — shell test validates all 37 agents have valid model: field for CC 2.1.47 team spawns
  • hook-input-fields.test.ts — validates added_dirs type contract and confirms enabledPlugins is NOT a hook field
  • configure skill — documents added_dirs, clarifies enabledPlugins is CC-internal (not hook-accessible)
  • help skill — CC 2.1.47 keyboard shortcuts section (Ctrl+F, Shift+Down)
  • upgrade-assessment — CC 2.1.47 upgrade guide reference

Fixed

  • Cross-platform /tmp hardcode — 13 hook source files migrated from /tmp/claude-* to paths.ts helpers using os.tmpdir() and path.join() (#720)
  • ork: prefix standardization — ~120 violations fixed across 60 skill files for consistent cross-referencing (#716)
  • Test assertions — 8 test files updated for os.tmpdir() compatibility (macOS /var/folders/.../T/ vs Linux /tmp)
  • node:path mock — test files now include proper default export when mocking node:path
  • User-invocable count — EXPECTED_USER_INVOCABLE updated 24→28 in skill validation test

[6.0.22] - 2026-02-18

Added

  • Hook: memory-capture — new Stop hook that auto-captures session summaries to ~/.claude/memory/decisions.jsonl for sessions with >20 tool calls; nudges /ork:remember for sessions with >50 tool calls (#708)
  • Hook: skill-nudge (PostToolUse/Bash) — nudges /ork:create-pr after a successful git push (#705)
  • Hook: skill-nudge (UserPromptSubmit) — nudges /ork:fix-issue when a GitHub issue URL is detected in the prompt (#705)
  • Hook: task-agent-advisor — PreToolUse[Task] hook that suggests curated ork agents for 14 common ad-hoc agent names and corrects 4 built-in casing errors (#704 #706)

Fixed

  • Analytics: zero-tool sessions — session-cleanup no longer writes to session-summary.jsonl when total_tools == 0, eliminating 57% noise from short/failed sessions (#707)
  • Cross-platform: /tmp hardcode — session-cleanup and memory-capture now use getMetricsFile() (via os.tmpdir()) instead of hardcoded /tmp/claude-session-metrics.json (#704)
  • Analytics: jq-queries — session count query now filters total_tools > 0 to match corrected analytics data (#707)
  • Tests: agent casing — tests/unit/test-pretool-all-hooks.sh corrected "explore" → "Explore" per CC 2.1.45 built-in naming spec (#704)

[6.0.20] - 2026-02-18

Added

  • Hook: mcp-health-check — new SessionStart hook that silently detects MCP misconfigurations at session start: warns if Tavily is enabled but TAVILY_API_KEY is unset, or if agentation is enabled but agentation-mcp package is not installed. Respects ORCHESTKIT_SKIP_SLOW_HOOKS=1.
  • Skill: github-operations — new references/cli-vs-api-identifiers.md mapping gh CLI identifiers (NAME) to REST API identifiers (NUMBER/node_id) for milestones, issues, PRs, and Projects v2 (#701)
  • Static analysis: Section E — scripts/eval/static-analysis.sh now enforces CLI-vs-API identifier documentation for any skill mixing --milestone CLI flags with REST API milestone paths

Fixed

  • Windows: console flashing — fix fire-and-forget hooks spawning visible cmd.exe windows (detached: true → detached: false + unref()) (#644)
  • Windows: ENAMETOOLONG — use os.tmpdir() for hook work files instead of deep project-dir paths (79 errors/session eliminated)
  • MCP defaults — Tavily and agentation are now "disabled": true in .mcp.json; users opt in explicitly to avoid surprise API costs
  • MCP docs — configure skill, doctor skill, installation.mdx, configuration.mdx, faq.mdx all updated to reflect accurate MCP status and setup instructions (#702)
  • github-operations: --milestone footgun — explicit warning that gh issue edit --milestone takes a NAME (string), not a number; REST API uses NUMBER (#699)
  • issue-progress-tracking: close-on-merge rule — added as Common Mistake #5; issues close only via Closes #N in PR body on merge, never via gh issue close directly

Improved

  • Skill: github-operations — batch issue creation pattern with array-driven loop and captured issue number (#700); Best Practice #9 (never close issues directly)
  • Skill: issue-progress-tracking — close-on-merge workflow documented in Common Mistakes
  • Extract spawn-worker.mjs — shared helper reduces 7 duplicated entry points (7 × 57 → 7 × 17 lines)
  • Cursor compat note — marketplace.json documents why Cursor misreads author field as MCP server (#698)

[6.0.19] - 2026-02-17

Fixed

  • Hook: memory-capture — new Stop hook that auto-captures session summaries to ~/.claude/memory/decisions.jsonl for sessions with >20 tool calls; nudges /ork:remember for sessions with >50 tool calls (#708)
  • Hook: skill-nudge (PostToolUse/Bash) — nudges /ork:create-pr after a successful git push (#705)
  • Hook: skill-nudge (UserPromptSubmit) — nudges /ork:fix-issue when a GitHub issue URL is detected in the prompt (#705)
  • Hook: task-agent-advisor — PreToolUse[Task] hook that suggests curated ork agents for 14 common ad-hoc agent names and corrects 4 built-in casing errors (#704 #706)

Fixed

  • Analytics: zero-tool sessions — session-cleanup no longer writes to session-summary.jsonl when total_tools == 0, eliminating 57% noise from short/failed sessions (#707)
  • Cross-platform: /tmp hardcode — session-cleanup and memory-capture now use getMetricsFile() (via os.tmpdir()) instead of hardcoded /tmp/claude-session-metrics.json (#704)
  • Analytics: jq-queries — session count query now filters total_tools > 0 to match corrected analytics data (#707)
  • Tests: agent casing — tests/unit/test-pretool-all-hooks.sh corrected "explore" → "Explore" per CC 2.1.45 built-in naming spec (#704)

6.0.18 - 2026-02-17

Fixed

  • Sync package.json version from 6.0.12 to 6.0.18 (was drifted from manifests)
  • Create 19 missing git tags for all historical versions (v5.2.4 through v6.0.17)
  • Fix all changelog compare links to resolve to valid tag comparisons
  • Update [Unreleased] link to point to v6.0.18...HEAD

6.0.17 - 2026-02-17

Added

  • CI: Add manifest-schema-tests job (manifests, schemas, hooks, indexes)
  • CI: Add eval-static job running npm run eval:static with artifact upload
  • CI: Add build drift detection (git diff --quiet plugins/) to build-plugins job

Fixed

  • Rename validate-evaluations.sh → test-validate-evaluations.sh so CI test discovery (test-*.sh pattern) finds it
  • Fix test-agent-categories.sh: add research to valid categories list
  • Fix test-category-grouping.sh: update stale agent name, add missing agent/category mappings, fix category order

Removed

  • Delete stale test-json-output-compliance.sh (hooks migrated to TypeScript, tested via vitest)
  • Delete stale test-build-marketplace-sync.sh (marketplace sync logic removed from build script)

6.0.16 - 2026-02-16

Added

  • Add impactDescription and tags metadata to 98 rule files across 9 skills
  • Add **Incorrect**/**Correct** code example pairs to all 258 rules missing them
  • Rule example coverage now 412/412 (100%)

Fixed

  • Fix 5 skill structure warnings (missing Overview/Related Skills sections)
  • Resolve all 459 skill test warnings (196 metadata + 258 code examples + 5 structure)

6.0.15 - 2026-02-16

Added

  • Rule traceability test: validates test case → rule file → content chain across 28 skills
  • Eval completeness test: validates consistency between test-cases.json and .eval.json formats
  • Skill efficiency scorecard: JSON metrics (rule count, imperative instructions, code examples, specificity ratio) for all 62 skills
  • npm run test:skills:functional script for running functional tests

Fixed

  • Corrected rule field mappings in 8 test-cases.json files (39 broken references → actual rule filenames or null)

6.0.14 - 2026-02-16

Fixed

  • Remove all Mem0 Cloud references from 24 docs files — 4-tier → 3-tier memory architecture
  • Delete leftover mem0 artifacts (.claude/mem0-webhooks.json, mem0-queue.jsonl, logs)
  • Clean .gitignore and mypy.ini of mem0 entries
  • Fix stale hook count in choosing-a-plugin.mdx (86 → 89)

6.0.13 - 2026-02-15

Fixed

  • Version sync: package.json 6.0.9 → 6.0.13 to match manifests/pyproject.toml
  • CLAUDE.md: version 6.0.8 → 6.0.13, hook bundles 11 → 12
  • marketplace.json: engine >=2.1.33 → >=2.1.34
  • README: orkl skill count 62 → 45, ork-creative 62 → 3
  • CONTRIBUTING: orkl 44 → 45 skills, hooks 98 → 89, bundles 11 → 12
  • Doctor skill: version 5.4.0 → 6.0.13, CC min 2.1.16 → 2.1.34, hooks 22 → 66 entries, bundles 11 → 12, agents 35 → 36, skills 186 → 62, all reference files updated
  • Hooks README: 93 → 89 hooks, 11 → 12 bundles, CC 2.1.33 → 2.1.34
  • Docs site: installation orkl 44 → 45, command-skills 23 → 24, troubleshooting bundles 11 → 12
  • Demo files: all 6 component files updated with current counts (62 skills, 36 agents, 89 hooks)

6.0.11 - 2026-02-15

Fixed

  • Guard against image paste killing context window — 3-layer defense: stdin cap (512KB), prompt length guard (50K), image/binary detection (#620)
  • Correct orkl skill count in marketplace.json (62 → 45)

Changed

  • Rewrote mcp-patterns SKILL.md for 2025-11-25 MCP spec + AAIF governance (#613)
  • Updated _sections.md with 7 categories (14 rules) and priority levels
  • Updated agent skill-indexes for ai-safety-auditor, llm-integrator, security-auditor
  • Regenerated docs site pages and generated data files

6.0.9 - 2026-02-15

Fixed

  • Sync all component counts across docs, README, CLAUDE.md, marketplace.json, and tests
  • Hook count 86→89 (3 new hooks from issue-driven git workflow)
  • Plugin skill counts: orkl 44→45, ork 61→62
  • Version reference 6.0.6→6.0.9 in CLAUDE.md
  • Fix "0 hooks" in README and CLAUDE.md hero line
  • Update docs-data.test.ts hardcoded assertions to match current counts
  • Close milestones #56 (Documentation Site) and #68 (Issue-Driven Git)

6.0.7 - 2026-02-14

Added

  • 9 test-cases.json files for orchestration skills (commit, review-pr, fix-issue, implement, explore, create-pr, verify, assess, git-workflow) — 39 test cases total (#563)
  • Skill triggering test suite with trigger-cases.yaml + keyword overlap scoring (#571)
  • Rule validation test for YAML frontmatter compliance (#543)
  • "When NOT to Use" tier matrices for architecture-patterns, DDD, distributed-systems (#532)
  • Interview/take-home mode (STEP 0c) in implement skill (#533)
  • 5 new scripts: validate-conventional.sh, check-plugin-health.sh, scaffold-test.sh, scan-vulnerabilities.sh, run-lint-check.sh (#565)
  • Edit > Write memory pattern in memory + remember skills (#546)
  • Lazy loading wrappers for SkillBrowser, DemoGallery, AgentSelector (#438)
  • Docs CI workflow for auto-rebuild on src/ changes (#402)
  • Shared category-colors.ts palette for consistent UI (#439)
  • Database selection rule + 4 references in database-patterns (#535)
  • Railway deployment rule + 3 references in devops-deployment (#550)
  • Messaging integrations rule + 3 references in api-design (#551)
  • Payload CMS rule + 3 references in api-design (#552)

Fixed

  • Replaced ~50 hardcoded hex colors with fd-* design tokens in docs site (#440)
  • Fixed category color inconsistencies: ai (emerald→cyan), devops (violet→orange), research (indigo→teal) (#439)
  • Added aria-live="polite" on copy button feedback (#440)

6.0.6 - 2026-02-14

Added

  • Hooks: Add type-error-indexer SessionStart hook — caches tsc --noEmit errors for agent awareness (#304)
  • Skills: Extract 27 rules from 7 reference-only tech skills (#559)
  • Skills: Add YAGNI gate to quality-gates, context-aware architecture-patterns (#530, #531)
  • Skills: Add scope-appropriate-architecture skill (#528, #529)

Changed

  • Agents: Rename performance-engineer → frontend-performance-engineer to avoid confusion with python-performance-engineer (#587)
  • CLAUDE.md: Update CC format reference from 2.1.6 to 2.1.34 (#515)

Fixed

  • CI: Raise vitest coverage thresholds (40/45/30/40 → 70/72/65/70), target 80%
  • CI: Raise hook coverage threshold (20% → 75%), fix detection to match .js imports
  • CI: Pin 6 unpinned actions in eval-index-effectiveness.yml to SHA
  • CI: Fix version-check.yml to detect manifest/workflow file changes
  • CI: Add graduated coverage penalties and skip/lint penalties to CI report scoring
  • CI: Fix Job Summary to render ci-report.md inline
  • CI: Include dispatchers in hook coverage count
  • Tests: Delete dead test-hooks-unit.sh (all 6 hooks migrated to TS, every test SKIPped)
  • Tests: Remove 21 aspirational test.skip blocks from user-tracking-wiring.test.ts
  • Tests: Delete unused parallel mode from run-tests.sh (had bug assuming 100% pass)
  • Tests: Report vitest pass/fail/skip counts in run-all-tests.sh
  • Docs: Fix CLAUDE.md hook count (65 → 63 global, matches hooks.json)
  • Docs: Fix markdownlint warnings in golden-dataset and release-management skills
  • Docs: Replace unsupported gitattributes language with text for Shiki

6.0.5 - 2026-02-13

Fixed

  • Bugs: Close 11 open issues (#417, #418, #437, #450, #451, #452, #453, #454, #455, #456, #534)
  • Skills: Fix doctor "health diagnostics for health diagnostics" redundancy (#456)
  • Skills: Add skill name to explore description for semantic discovery (#456)
  • Tests: Rewrite MCP pretool hook tests for TypeScript — removes 14 dead skip() calls (#417)
  • Tests: Rewrite hook timing tests to measure TypeScript hooks via run-hook.mjs (#455)

Changed

  • CLAUDE.md: Prune from 275 to 75 lines, link to docs instead of inline (#452)
  • Hooks: Consolidate 4 PostToolUse Write|Edit hooks into unified-write-quality-dispatcher (#453)
    • auto-lint, readme-sync, merge-conflict-predictor, coverage-predictor → 1 dispatcher
    • Hook count: 88 → 86 (63 global + 22 agent + 1 skill)
  • README: Fix stale counts (200→60 skills, 98→86 hooks), remove playground-demo.gif
  • GitHub: Update repo description with accurate component counts

6.0.4 - 2026-02-13

Fixed

  • CI: Remove deleted evidence-verification skill from test-specific-skills.sh (#555 aftermath)
  • CI: Add missing version field and checklists/ directory to mcp-patterns skill for test-ai-ml-skills.sh

Unreleased

Changed

  • Skill Consolidation (#536): Restructured 200 skills into 103 through 16 consolidation batches
    • Batches 1-6: LangGraph (10→1), RAG (9→1), Testing (13→1), Caching (4→1), Performance (6→1), Video (14→1)
    • Batches 7-16: Event-Driven (3→1), Golden-Dataset (3→1), Accessibility (3→1), Database-Patterns (4→1), LLM-Integration (7→1), API-Design (3→1), Distributed-Systems (4→1), Agent-Orchestration (3→1), Security-Patterns (6→1), Product-Frameworks (5→1)
    • Additional groups: Monitoring-Observability (2→1), Frontend-Animation (3→1), UI-Components (4→1), Data-Visualization (2→1), Python-Backend (5→1), Architecture-Patterns (2→1), Browser-Tools (2→1), Context-Optimization (2→1), Async-Jobs (2→1)
    • Total: 200 → 103 skills (76 internal, 27 user-invocable)
    • ork-creative: 16 → 3 skills (demo-producer, video-production, ascii-visualizer)
    • Updated 33 agents, hooks, manifests, tests, and CLAUDE.md with corrected counts
    • Hook count corrected: 93 → 88 (65 global + 22 agent + 1 skill)

Added

  • TLDR-Lite File Summaries (#463): New PreToolUse[Read] hook injects structural summaries for large files (>500 lines or >2000 tokens)
    • Regex-based extractors for TypeScript/JS, Python, Go, Rust, Shell, Markdown
    • Extracts imports, functions, classes, types, exports as a navigation roadmap
    • ~500 token summary injected as additionalContext alongside full file content
    • 7 guard conditions: skip targeted reads, unsupported extensions, small files, >2MB files
    • 42 new tests (25 library + 17 hook)
    • Hook count: 97 → 98 (70 global + 22 agent + 6 skill)
    • pretool bundle: 57.42 → 63.36 KB (+10%)

Fixed

  • Docs: Corrected stale component counts across README, marketplace.json, CONTRIBUTING.md, Fumadocs site pages, and skill references (199→200 skills, 119→98 hooks)

6.0.3 - 2026-02-07

Added

  • Langfuse v3 Rewrite (Milestone #58): Complete rewrite of langfuse-observability skill from deprecated SDK v2 to v3/v4 (OTEL-native)
    • SKILL.md bumped to v2.0.0 with 3 new capability sections (agent-graphs, mcp-prompt-management, framework-integrations)
    • 3 new reference files: agent-observability.md, framework-integrations.md, migration-v2-v3.md
    • 7 existing reference files rewritten with v3 imports (from langfuse import observe, get_client)
    • All langfuse_context → get_client(), all langfuse.decorators → langfuse imports
    • New coverage: Agent Graphs, MCP Server, Experiment Runner SDK, dataset versioning, spend alerts, natural language filtering, evaluator execution tracing
    • 18 tracking issues (#419-#436) under Milestone #58

Fixed

  • Mem0: Add --no-infer flag to add-memory.py — passes infer=False to mem0's client.add(), disabling semantic dedup in batch tests (Test 6, Test 18) so counts are deterministic across parallel CI runners
  • CI: Quote workflow names containing colons (Validate: Plugins, Validate: Version, Eval: Agent Routing, Visualize: Memory) — unquoted colons caused YAML parse errors resulting in 0 jobs
  • CI: Fix test-git-enforcement-hooks.sh failing on main by setting ORCHESTKIT_BRANCH env override so git-validator hook doesn't block test commits on protected branches
  • CI: Rewrite test-hook-chains.sh to read from committed hooks.json instead of gitignored settings.json

6.0.2 - 2026-02-06

Added

  • #328 (P1-C): complexity: low|medium|high field added to all 199 skill frontmatters for Opus 4.6 adaptive thinking alignment

  • #337 (P2-E): New upgrade-assessment user-invocable skill — 6-phase readiness evaluation with structured JSON scoring across 6 dimensions

  • #338 (P2-F): New platform-upgrade-knowledge reference skill with scoring rubrics and compatibility matrices

  • #333 (P2-D): 128K output token guidance added to implement skill, context-engineering, and 3 agent definitions

  • #331 (P2-B): New model-cost-advisor SubagentStart hook — analyzes task complexity and recommends optimal model for cost savings

  • #325 (P0-B): Prefill-guard SessionStart hook warns about Opus 4.6 breaking change (prefilled assistant messages return 400 errors)

  • #346 (P1-E): Agent memory frontmatter — all 36 agents (31 project scope, 5 local scope) (CC 2.1.33)

  • #347 (P1-F): New TeammateIdle and TaskCompleted hook events with progress-reporter and completion-tracker handlers (CC 2.1.33)

  • #335 (P3-B): New /ork:audit-full user-invocable skill — single-pass whole-codebase audit (security, architecture, dependencies) leveraging 1M context window with 4 references, 2 assets, 1 checklist, 1 script

  • Batch script scripts/add-complexity.mjs for applying complexity classifications

  • #334 (P3-A): Agent Teams dual-mode orchestration — /ork:implement and 5 other user-invocable skills (assess, brainstorm, explore, fix-issue, review-pr, verify) support both Task tool (star topology) and Agent Teams (mesh topology) via ORCHESTKIT_PREFER_TEAMS env var

  • #405: TeamCreate, SendMessage, TeamDelete tools added to all 36 agents

  • #406: task-dependency-patterns skill updated with Agent Teams coordination patterns

  • #407: multi-agent-orchestration skill updated with mesh topology patterns

  • #362: 4 Agent Teams lifecycle hooks (team-formation-advisor, teammate-progress-reporter, teammate-completion-tracker, team-coordination-advisor)

  • #391 (P2-B): Interactive Agent Selector playground with search, category/task filters, quiz wizard, and 10 scenario suggestions

  • Fumadocs site scaffold (Milestone #56): Fumadocs v16.5 + Next.js + MDX + Orama search, reference pages auto-generated for all 199 skills, 36 agents, 15 hook categories

  • Tavily Integration: 3-tier web research workflow (WebFetch → Tavily → agent-browser) with curl patterns for search/extract/map APIs, graceful degradation when TAVILY_API_KEY is unset

  • Tavily Site Discovery: competitive-monitoring skill gains Tavily map+extract pre-step for competitor URL enumeration

  • Tavily Agent Awareness: web-research-analyst, market-intelligence, and product-strategist agents updated with Tavily directives

Fixed

  • SEC-001: SQL injection prevention — multi-instance-cleanup and cleanup-instance now validate instance IDs with /^[a-zA-Z0-9_\-.:]+$/ before SQLite interpolation
  • SEC-003: Atomic file writes — multi-instance-lock uses write-to-temp + renameSync to prevent TOCTOU race conditions in lock files
  • README.md hook count corrected (120 → 119)

Changed

  • #348 (P2-G): Task(agent_type) restrictions on python-performance-engineer and demo-producer (CC 2.1.33)
  • #349 (P1-G): CC minimum version bumped to >= 2.1.33 (from 2.1.32) for agent memory and new hook events
  • #330 (P2-A): 13 agents migrated from mcp__sequential-thinking to Opus 4.6 native adaptive thinking
  • #329 (P1-D): TOKEN_BUDGETS now scale dynamically with CLAUDE_MAX_CONTEXT (2% of context window per CC 2.1.32)
  • #332 (P2-C): Enhanced pre-compact-saver v2.0 — preserves decision logs, memory tier snapshots, compaction frequency analytics
  • #324 (P0-A): Replace hardcoded model string in multi-instance-init.ts with dynamic detectModel()
  • #326 (P1-A): Memory context tier limits expanded (1200→3000 chars memory, 800→1200 chars profile)
  • #327 (P1-B): CC minimum version updated to >= 2.1.33 across CLAUDE.md, README, hooks README, marketplace
  • MCP configuration docs updated with Opus 4.6 sequential-thinking deprecation note
  • CI workflow renames for clarity and pipeline parallelism
  • Skill count: 197 → 200 (added upgrade-assessment, platform-upgrade-knowledge, audit-full)
  • Hook count: 117 → 119 (91 global + 22 agent + 6 skill-scoped)
  • Opus 4.6 callouts added to top 5 user-invocable skills (verify, review-pr, fix-issue, implement, explore)
  • Agent memory frontmatter expanded from 22 to all 36 agents

Removed

  • Deprecated sequential-thinking-auto pretool hook (Opus 4.6 native adaptive thinking replaces MCP sequential-thinking)
  • #362: 6 coordination hooks removed as redundant with CC native Agent Teams (team-formation-hook, team-coordinator, teammate-monitor, team-cleanup, team-health-check, team-context-share)

6.0.1 - 2026-02-05

Changed

  • orkl manifest: Removed 12 language-specific skills to keep orkl truly universal (107 → from 119 skills)
    • Removed: run-tests, background-jobs, connection-pooling, caching-strategies, rate-limiting, api-versioning, error-handling-rfc9457, input-validation, property-based-testing, i18n-date-patterns, image-optimization, type-safety-validation

Added

  • LangGraph skills: Added Quick Start sections, 6 cross-links (up from 3), and evaluation test cases
  • New skills: langgraph-streaming, langgraph-subgraphs, langgraph-tools
  • Progressive disclosure: Refactored mcp-server-building and database-versioning into main SKILL.md + references/

Fixed

  • Documentation count sync: README.md, CLAUDE.md, CONTRIBUTING.md now show correct counts (197 skills, 36 agents, 117 hooks)

6.0.0 - 2026-02-04

Changed

  • Breaking: Reorganized plugins from 26 granular plugins to 2-tier architecture:
    • orkl (119 skills) — Universal toolkit, language-agnostic, all workflows work out of the box
    • ork (195 skills) — Full specialized toolkit with Python, React, LLM/RAG patterns
  • Renamed ork-lite to orkl for shorter prefix
  • All 36 agents and 117 hooks included in BOTH plugins
  • Deleted 25 domain-specific plugin manifests (ork-accessibility, ork-backend-patterns, ork-memory-*, etc.)
  • Memory plugins (graph, mem0, fabric) now included directly in orkl
  • No more dependency issues — every workflow works without additional installs

Added

  • web-research-analyst agent for browser automation and competitive intelligence
  • research category for agent indexes
  • competitive-monitoring skill (22 user-invocable skills total)

Fixed

  • Bash 3.2 compatibility in generate-indexes.sh (macOS default shell)
  • Agent index generation now includes all categories correctly
  • Playgrounds updated for two-tier plugin system (setup-wizard, data.js, index.html)

5.7.5 - 2026-02-04

Fixed

  • CI report generation using tsx for ESM compatibility

5.7.3 - 2026-02-03

Changed

  • Removed redundant skill-resolver hook (Claude Code natively auto-injects skills from agent frontmatter)
  • Hook count: 117 total (89 global + 28 agent/skill-scoped)
  • Eval framework focuses on agent routing correctness via CLAUDE.md agent-index (removed skill metrics)

Fixed

  • Eval scripts: Fixed dry-run mode to create all required output directories
  • Eval scripts: Fixed path prefixes in golden test commands
  • Eval scripts: Added explicit exit 0 to ensure proper exit status
  • CI workflow: Exclude eval scripts from artifact to prevent overwriting

5.7.2 - 2026-02-03

Added

  • Three-tier auto-suggest UX: Confidence-based messaging for skill injection

    • SILENT (≥90%): No user notification, just inject
    • NOTIFY (80-89%): Brief "💡 Loaded: X" notification
    • SUGGEST (70-79%): Suggest skills with context
    • HINT (50-69%): Hint at possible matches
  • AGENTS.md cross-tool compatibility: Generate AGENTS.md alongside CLAUDE.md

    • Enables compatibility with Cursor, Codex, Amp, Zed
    • AGENTS.md follows the open standard at https://agents.md/
  • CI-based index effectiveness evaluation: A/B comparison framework

    • 12 golden tests covering 8 different agents
    • Compares with-index vs without-index performance
    • Metrics: build success, lint compliance, test pass, agent routing
  • Dynamic test discovery via scripts/ci/run-tests.sh (discovers 154 tests vs ~100 hardcoded)

  • Reusable test-runner workflow (.github/workflows/reusable-test-runner.yml)

  • Composite action for test execution (.github/actions/run-tests/action.yml)

  • Expanded npm caching strategy (node_modules, vitest cache)

  • Multi-tier skill scenario test assertion (notify-tier generates systemMessage)

Changed

  • Removed redundant skill-resolver hook (Claude Code natively auto-injects skills from agent frontmatter)
  • Hook count: 117 total (89 global + 28 agent/skill-scoped)
  • CI workflows now use SHA-pinned GitHub Actions
  • Test execution uses dynamic discovery instead of hardcoded invocations
  • Eval framework focuses on agent routing correctness via CLAUDE.md agent-index

Security

  • Pin googleapis/release-please-action@v4 to SHA commit (SEC-001)
  • Fix shell injection risk in version-check.yml by using env variable pattern for github.head_ref (SEC-002)
  • Quote inputs.ci-setup-flags in setup action to prevent command injection (SEC-003)
  • All 8 external GitHub Actions now pinned to SHA commits for supply chain security
  • Add SHA documentation in .github/action-versions.md
  • Replaced arbitrary eval with allowlist-based command execution
  • Pinned yq to v4.40.5 with SHA256 checksum verification
  • Added cleanup traps for temp directories on EXIT/ERR

5.7.1 - 2026-02-02

Added

  • Dynamic test discovery via scripts/ci/run-tests.sh (discovers 154 tests vs ~100 hardcoded)
  • Reusable test-runner workflow (.github/workflows/reusable-test-runner.yml)
  • Composite action for test execution (.github/actions/run-tests/action.yml)
  • Expanded npm caching strategy (node_modules, vitest cache)

Changed

  • CI workflows now use SHA-pinned GitHub Actions
  • Test execution uses dynamic discovery instead of hardcoded invocations

Security

  • Replaced arbitrary eval with allowlist-based command execution
  • Pinned yq to v4.40.5 with SHA256 checksum verification
  • Added cleanup traps for temp directories on EXIT/ERR

5.7.0 - 2026-02-02

Changed

  • Model Updates (February 2026):

    • OpenAI: gpt-4o → gpt-5.2, gpt-4o-mini → gpt-5.2-mini
    • Anthropic: Claude model dates updated to 20251101 (Opus 4.5, Sonnet 4.5, Haiku 4.5)
    • Meta: llama3.2 → llama3.3
    • Google: Gemini 2.5 → Gemini 3
    • xAI: Grok 3 → Grok 4
  • Framework Updates:

    • Next.js 16: Cache Components with "use cache" directive, async params/searchParams
    • Tailwind CSS v4: CSS-first configuration with @theme {}, no tailwind.config.js needed
    • LangGraph 1.0: create_react_agent → create_agent, prompt= → system_prompt=
    • Redis 8.4: Built-in Search/JSON modules, FT.HYBRID with RRF fusion
    • Vite 8: Rolldown bundler with advancedChunks API
    • CrewAI 1.8.x: Flows architecture with @start(), @listen(), @router() decorators
    • Playwright 1.58: Agents workflow with init-agents, planner/generator/healer pattern

Added

  • New reference files:

    • cache-components.md — Next.js 16 Cache Components comprehensive guide
    • nextjs-16-upgrade.md — Breaking changes and migration path
    • gpt-5-2-codex.md — GPT-5.2-Codex agentic coding model documentation
    • Redis 8 FT.HYBRID comparison added to pgvector-search skill
    • CrewAI 1.8.x Flows patterns added to crewai-patterns.md
  • Updated skills (193 skills updated with current best practices):

    • react-server-components-framework — v1.4.0 with Cache Components
    • vite-advanced — Rolldown bundler patterns
    • e2e-testing — Playwright Agents workflow
    • semantic-caching — Redis 8 built-in modules
    • rag-retrieval — LangGraph 1.0 API updates
    • alternative-agent-frameworks — OpenAI Agents SDK 0.7.0, CrewAI 1.8.x

Fixed

  • memory/SKILL.md — Renamed "Quick Start" to "Usage" for test validation
  • implement/SKILL.md — Reduced from 527 to 474 lines (under 500 limit)
  • data.js — Regenerated with correct version 5.7.0
  • session-tracking.test.ts — Fixed TEST_PROJECT_DIR initialization

5.6.2 - 2026-02-02

Fixed

  • Critical: Remove "deps" key from marketplace.json that caused "Invalid schema: Unrecognized key: deps" errors on all 26 plugins when installing via /plugin
  • Remove dependency propagation from build script (deps stay in manifests for internal use only)
  • Upgrade marketplace schema test from denylist to allowlist validation to catch any future unrecognized keys

5.6.1 - 2026-02-02

Fixed

  • Build script syncs all marketplace.json plugin versions to the project version (eliminates version drift)
  • Build script propagates dependencies from manifests into marketplace.json deps fields
  • Version bump script now updates all 26 manifests (prevents build from overwriting bumped versions)

5.6.0 - 2026-02-01

Added

  • Memory health check library (memory-health.ts) — Validates JSONL integrity, tier status, queue depths, and file analysis for /ork:doctor
  • Memory metrics collector (memory-metrics.ts) — Counts decisions by category/type, queue depths, completed flows; appends timestamped snapshots to memory-metrics.jsonl
  • Memory metrics lifecycle hook (memory-metrics-collector.ts) — SessionStart hook that collects and persists memory metrics on every session