Skip to content

roedyrustam/vibes-plug

v4.5.0MIT

Universal AI Agent Swarm & MCP plugin with 176 specialized skills and 8 autonomous subagents for Cursor, Antigravity, Claude Code, Windsurf, and OpenAI Codex.

GitHubOpen Website

Changelog / Catatan Perubahan

All notable changes to this project will be documented in this file. Semua perubahan penting pada proyek ini akan didokumentasikan dalam berkas ini.

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

[4.5.0] - 2026-10-25

Added / Ditambahkan

  • Universal Skill-as-a-Tool (SaaT) Execution Engine (scripts/skill-dispatcher.mjs):
    • Transformed all 176 declarative skills into deterministic executable tools via vibes skill exec <skill> <action> (blueprint, slice, extract-code, audit, eval, verify, generate, fuzz, optimize, health, budget) and MCP tool vibes_skill_dispatch.
    • Built 8 specialized domain adapters (scripts/adapters/*.mjs for Database, Frontend, API, Security, TDD, AI/MCP, Cloud/DevOps, and Mobile/Spatial) aligned 1:1 with the 8 native specialist subagents.
    • Added Autonomous Skill Composition Chains (vibes skill chain <recipe>): saas-launch-gate, ai-agent-hardening, fintech-settlement-gate, cloud-release-gate, and frontend-ui-polish.
    • Added Pre-Compiled Micro-Slice Compiler (vibes skill compile-slices & vibes skill matrix) generating .skill-slices.json across all 176 skills and 1,000 executable code blueprints with 93% subagent Fan-Out token compression.
  • Reverse Engineering & Binary Analysis (REA) 5-Skill Suite (morluto/rea):
    • Integrated reverse-engineer-anything, rea-tool-design, rea-analysis-optimization, rea-ci-tests-audit, and rea-changelog-update with full bilingual orchestration matrices, TypeScript Evidence/Cache/Audit contracts, and registration in brainstorming and zero-to-prod-orchestrator (Phases 1, 6, and 8).
    • Synchronized CATALOG.md, CATALOG.json, .cache-prefix.json, and .skill-slices.json across all 176 verified skills with 100/100 Ecosystem Health (PRISTINE) and 2,024 directed graph edges.

[4.4.3] - 2026-10-24

Security / Keamanan

  • Critical Security Remediation (GHSA-f63r-xxp2-3m23 / CWE-306):
    • Loopback Host Binding: Enforced strict 127.0.0.1 binding on Studio Cockpit (scripts/studio-server.mjs) and Remote MCP SSE transport (bin/mcp-server.mjs), closing unauthenticated network listener exposure on 0.0.0.0.
    • Untrusted Cross-Origin Gate: Blocked DNS rebinding and cross-site request attacks by rejecting incoming cross-origin requests from untrusted origins with 403 Forbidden.
    • Test Runner Command Allowlist & Metacharacter Sanitization: Implemented validateTestCommand() in scripts/self-healer.mjs with ALLOWED_TEST_RUNNERS whitelist. Completely prohibited shell metacharacters (; & | \ $ > < \ !) and newlines in /api/heal/diagnoseand MCPvibes_self_heal`.
    • Automated Regression Tests: Added end-to-end regression test suite verifying rejection of arbitrary command execution, unauthorized binaries, and untrusted cross-origin requests.
    • Security Policy Update: Hardened execution boundary guarantees in SECURITY.md.
    • Credit: Discovered and responsibly reported by Artur Chlebicki (@Artur12555).

[4.4.2] - 2026-10-23

Added / Ditambahkan

  • 171-Skill & 8-Subagent Frontier Expansion:
    • Added 10 new 2026–2027 frontier domain skills: double-entry-ledger-fintech-expert, fga-rebac-authorization-expert, a2a-agent-protocol-expert, durable-agent-execution-expert, passkeys-webauthn-identity-expert, reactive-sync-engine-expert, olap-columnar-analytics-expert, kubernetes-gitops-platform-expert, sbom-supply-chain-security-expert, and computer-use-os-agent-expert.
    • Added 3 new native specialist subagents in agents/: cloud-devops-sre, data-rag-architect, and mobile-spatial-craftsman (expanding native swarm subagents from 5 to 8).
    • Deepened 14 core foundation skills with decision matrices, production TypeScript/SQL/Rust/YAML implementations, and verification checklists.
    • Upgraded scripts/audit-graph.mjs and connected all 171 skills into a 100% strongly-connected graph (1,986 directed edges, 0 isolated skills, 0 broken references).
    • Expanded scripts/swarm-runner.mjs planSwarm() from 4 basic checks to 8 specialized swarm domains.

Changed / Diubah

  • 171 / 171 Universal Skill Harmonization & Upgrade to v4.4.2:
    • Full ecosystem audit across all 171 skills in skills/ and .cursor/skills/ upgraded to frontmatter version: "4.4.2".
    • Harmonized orchestration architecture: categorized markdown cross-references standardized directly under ## English (### Orchestration & Integration) and ## Bahasa Indonesia (### Integrasi Orkestrasi).
    • Eradicated all residual trailing duplicate blocks (### Integration with Other Skills, ### Referenced By Orchestrators, and Indonesian counterparts) across all skills.
    • Strict Anti-Slop verification: 100% compliant with 0 errors and 0 warnings across all 579 files under node scripts/check-anti-slop.js --strict.
  • Catalogs & Registries Synchronization:
    • CATALOG.md and CATALOG.json completely re-indexed and synchronized with all 171 skills reflecting version 4.4.2.
    • Template templates/SKILL_TEMPLATE.md upgraded to 4.4.2 with standardized bilingual orchestration layout.
    • Rule files (AGENTS.md, rules/AGENTS.md, CLAUDE.md, CODEX.md, .cursorrules, .traerules, .windsurfrules) and CONTRIBUTING.md aligned to version 4.4.2.
  • Ecosystem Consistency Pass:
    • Catalog-count consistency pass across live surfaces: Cursor/Claude/Codex core rules, Cursor plugin manifest, VS Code status tooltip, CONTRIBUTING, README share link, and Studio/showcase version labels now all read 171 skills.
    • Plugin manifests bumped to 4.4.2 (package.json, plugin.json, gemini-extension.json, .cursor-plugin, .codex-plugin, VS Code companion).
    • scripts/sync-all-skill-counts.mjs now covers the missed phrases (159-skill registry, against the 155 skills, .cursor-plugin description, VS Code 160-Skill tooltip).

[4.4.1] - 2026-10-20

Added / Ditambahkan

  • 160-Skill Milestone & Enterprise Gap Closure:
    • github-actions-expert: Enterprise guide for GitHub Actions CI/CD workflows, reusable workflows (workflow_call), composite actions, multi-platform test matrices, zero-secret OIDC cloud role authentication for AWS/GCP/Azure, immutable 40-character commit SHA action pinning, and least-privilege GITHUB_TOKEN permissions.
    • docker-container-expert: Production-grade containerization with Docker 27+ and Docker Compose v2, BuildKit persistent cache mounts (--mount=type=cache), rootless execution (UID 65532 non-root) on Google Distroless and Scratch, internal network isolation (internal: true), container health checks, and compose dependency sequencing (condition: service_healthy).
    • Full registration of both skills in skills/brainstorming/SKILL.md (Domain Matrix in EN & ID) and skills/zero-to-prod-orchestrator/SKILL.md (Phase 7 DevOps & Hardening in EN & ID).
  • Public Web Showcase & Live Vercel Deployment (showcase/):
    • Built zero-dependency, ultra-fast vanilla HTML/CSS/JS Web Showcase with live interactive 160-skill explorer, category tabs, real-time search, terminal CLI simulator, benchmark audit viewer, and IDE quick-copy snippets.
    • Deployed publicly to Vercel at https://vibes-plug.vercel.app/ with automated continuous deployment on git push.
    • Automated GitHub Actions deployment pipeline (.github/workflows/deploy-showcase.yml) for GitHub Pages fallback.
  • Master Ideation Protocol (brainstorming) v4.4.1 Deep Architectural Upgrade:
    • Tri-Track Adaptive Velocity: Introduced Fast-Track (immediate Understanding Lock synthesis for mature prompts), Socratic Interview (focused turn-by-turn dialogue), and Deep Enterprise (mission-critical auditability & idempotency).
    • Native UI Modal Protocol: Enforced interactive UI forms (ask_question) with selectable multiple-choice cards over plain chat text.
    • Parallel Discovery Swarm (Fan-Out Audit): Subagent A (Repo & Living AST Audit via app-analyzer-optimizer + living-codebase-ast-graph), Subagent B (Market & Competitive Tech Discovery via deep-research-analyst + web-scraper), and Subagent C (Security & Compliance Bounds via compliance-gdpr-privacy-expert + zero-trust-secret-vault), combined with episodic memory recall via session-memory-manager / xerj_memory_recall.
    • Pre-Mortem Failure Simulation & FinOps Cost Gate: Automated catastrophic failure stress-testing (autonomous-red-teamer + error-resilience-expert) and monthly token/cloud cost projections (llm-finops-router).
    • MADR-Standard Decision Logs & Subagent Contract Slices: Architectural Decision Records in MADR format and isolated CONTEXT_MAP.md type slices to eradicate downstream context window pollution.
  • CI/CD Quality Gates & Test Suite Expansion:
    • Expanded test suite in scripts/test-cli.mjs to 42 passing tests (covering all 161 skills, doctor diagnostics, studio endpoints, and showcase assets).
    • All 161 skills verified with 0 errors across frontmatter, author, version, bilingual requirement, and code block formatting.

Changed / Diubah

  • Updated all catalogs, documentation, Studio Cockpit, and README across English, Indonesian, Mandarin, German, and Japanese to reflect 161 verified skills.
  • Resolved single-quoted string literal syntax in scripts/skill-drift-monitor.mjs restoring pristine status to all Studio drift APIs and CLI monitor dashboards.

[4.4.0] - 2026-10-18

Added / Ditambahkan

  • Pillar 1: Universal Cross-IDE State Bridge (scripts/cross-ide-bridge.mjs):
    • Full bidirectional state bridge across Google Antigravity (.agents/), Anthropic Claude Code (.claude/), Cursor IDE (.cursor/), and OpenAI Codex (.codex/).
    • Seamlessly links core rules (AGENTS.md, CLAUDE.md, .cursorrules, CODEX.md), synchronizes the 155 specialized skills via relative symlinks with graceful copy fallback, and generates unified markdown context bundles (CONTEXT_BUNDLE.md).
    • Integrated into CLI vibes bridge [status|sync|bundle], Studio API /api/bridge/*, and MCP tools vibes_bridge_status & vibes_bridge_sync.
  • Pillar 2: Autonomous Skill Synthesizer (scripts/skill-synthesizer.mjs):
    • Autonomous engine that synthesizes complete, gold-standard bilingual skills from raw project codebases or repository patterns (vibes learn <skill-name>).
    • Automatically analyzes source code symbols, generates complete frontmatter, bilingual docs (English & Indonesian), domain integration matrices, and auto-registers the synthesized skill into both brainstorming/SKILL.md and zero-to-prod-orchestrator/SKILL.md.
    • Integrated into CLI vibes learn <skill-name>, Studio endpoint /api/skill/learn, and MCP tool vibes_learn_skill.
  • Pillar 3: Pre-Disk Verification Gate (scripts/wasm-sandbox.mjs):
    • Atomic in-memory pre-commit gate that validates candidate code before writing to disk (vibes verify-gate / vibes sandbox --verify).
    • Evaluates AST syntax parsing, checks compliance against the 7 Anti-Slop Pillars (detecting empty catches, placeholders, and syntax narration), validates against dangerous API calls (process.exit, child_process, eval), and runs in-memory micro-sandbox execution in <2ms.
    • Integrated into CLI vibes verify-gate, Studio endpoint /api/sandbox/verify, and MCP tool vibes_verify_gate.
  • Pillar 4: Real-Time Swarm Concurrency Bus (scripts/swarm-bus.mjs):
    • High-performance in-memory event bus and Server-Sent Events (SSE) telemetry transport for autonomous agent swarms (vibes swarm:bus).
    • Captures agent lifecycle events (agent:spawn, agent:complete, agent:fail), tool calls (tool:call), memory writes (memory:write), and inter-agent broadcasts (agent:broadcast).
    • Wired into scripts/swarm-dispatcher.mjs for live observability. Integrated into CLI vibes swarm:bus [status|events|broadcast], Studio SSE endpoints /api/swarm/stream and /sse, query /api/swarm/events, broadcast /api/swarm/broadcast, and MCP tool vibes_swarm_bus.
  • Pillar 5: Local Hybrid GraphRAG Search (scripts/graphrag-indexer.mjs):
    • High-precision hybrid search combining Lexical BM25 (Lucene formula) + TF-IDF Vector cosine similarity + AST symbol connectivity ranking.
    • Features camelCase and snake_case code tokenization and multi-field weighted indexing across 48 codebase files and 249 indexed symbols with sub-millisecond query latency.
    • Integrated into CLI vibes graph:search <query> [--top-k 5], Studio endpoint /api/graph/search, and MCP tool vibes_graph_search.
  • Pillar 6: Zero-Dependency VSIX Extension Packager (scripts/package-extension.mjs):
    • Native Open Packaging Conventions (OPC) .vsix builder using built-in node:zlib (deflateRawSync and crc32) in <50ms without pulling external dependencies like @vscode/vsce.
    • Packages the companion extension for VS Code, Cursor, and Windsurf containing 11 native commands.
    • Integrated into CLI vibes package:vscode, npm script npm run package:vscode, Studio endpoint /api/extension/package, and MCP tool vibes_package_extension.
  • Studio Cockpit Interactive UI Expansion (scripts/studio-server.mjs):
    • Added 4 brand new interactive cockpit panels in the web UI:
      • Cross-IDE Bridge Panel: Visual status cards for all 4 AI platforms (Antigravity, Claude Code, Cursor, Codex), 1-click universal sync, and context bundle download.
      • Hybrid GraphRAG Search Studio: Live search bar with instant presets, top-K selection, and visual breakdown bars for BM25 score, Vector cosine similarity, and AST boost (+20%).
      • Skill Synthesizer & Pre-Disk Gate Studio: Scaffolding UI for synthesizing custom skills and interactive code sandbox for validating candidate code against the Pre-Disk Gate.
      • Live Swarm Stream Panel: Real-time SSE streaming log terminal connected to /api/swarm/stream with event filtering chips, broadcast input, and status indicator.
    • Updated MCP tool select dropdown with 7 newly added MCP tools (total 27 tools supported).

Changed / Diubah

  • Version bumped from 4.3.0 to 4.4.0 across all package manifests, extensions, and documentation.
  • Test suite expanded to 34 passing tests covering all 6 strategic pillars, CLI commands, and Studio server endpoints.

[4.3.0] - 2026-10-17

Added / Ditambahkan

  • Autonomous Self-Healing Test Runner & Failure Diagnoser (scripts/self-healer.mjs):
    • Mesin diagnosa kegagalan pengujian otonom yang mengeksekusi test suite atau perintah kode, menyadap stack trace, mengkorelasikan dan mengisolasi simbol AST yang rusak via Codebase Symbol Graph, memverifikasi perbaikan di Ephemeral Micro-Sandbox, dan menghasilkan proposal patch diff terverifikasi. Terintegrasi ke CLI vibes heal [cmd|file] [--apply], API Studio /api/heal/diagnose, dan Remote MCP tool vibes_self_heal.
  • Decentralized Community Skill Registry Hub (scripts/registry-manager.mjs):
    • Sistem registry terdesentralisasi untuk penemuan, pencarian berbobot relevansi, instalasi lokal ke .agents/skills/, dan pemaketan audit skill kustom terhadap 155 Gold Standard (frontmatter, bilingual, matrix orkestrasi, zero slop) dengan verifikasi hash SHA-256. Terkoneksi ke CLI vibes registry [list|search|install|pack], API Studio /api/registry/*, dan Remote MCP tool vibes_registry_search & vibes_registry_install.
  • Codebase AST Symbol Knowledge Graph Indexer (scripts/graphrag-indexer.mjs):
    • Pengindeksan simbol AST kode lokal multi-bahasa (.js, .mjs, .ts, .tsx, .jsx, .py, .go) super cepat (<60ms) yang mengekstrak fungsi, kelas, antarmuka, dan tipe ke .agents/graph/symbol_graph.json. Terkoneksi ke CLI vibes graph:build, vibes graph:query <symbol>, endpoint API /api/graph/query, dan Remote MCP tool vibes_graph_query tanpa membanjiri context window AI.
  • Autonomous GitHub PR Review Critic Bot (scripts/github-pr-bot.mjs):
    • Bot peninjau Pull Request cerdas yang mengaudit git diff atau berkas lokal terhadap Doktrin Sovereign Anti-Slop (Pilar 1–6) dan Standar Visual UI 2026 (Pilar 7). Menghasilkan laporan scorecard Markdown GitHub siap tempel dengan skor numerik dan putusan APPROVED / CHANGES_REQUESTED. Terintegrasi ke CLI vibes pr:review [diff|file], API /api/pr/review, dan MCP vibes_pr_review.
  • Self-Healing Diagnostics (vibes doctor --fix):
    • Opsi perbaikan otomatis pada vibes doctor yang secara mandiri memulihkan direktori .agents/ yang hilang (skills, graph, telemetry), mengompilasi ulang .cache-prefix.json, dan memasang git pre-commit hook untuk mencegah regresi slop ke repositori.
  • Studio Cockpit Visual & Syntax Polish (scripts/studio-server.mjs):
    • Perbaikan tuntas string template literal newline escaping pada trace logs, eliminasi text overlap pada input pencarian dengan SVG search icon 15x15 crisp, serta penambahan dropdown tool MCP baru vibes_graph_query dan vibes_pr_review.
  • Interactive Scaffolding Wizard (vibes wizard):
    • Perintah baru vibes wizard berbasis @inquirer/prompts untuk scaffolding proyek dan pemilihan stack kustom interaktif (Frontend, Backend, Database/ORM, Auth, Payment) serta injeksi otomatis skill bundel terkurasi ke .agents/skills/.
  • Automated Documentation Code Block Audit (scripts/validate-skills.mjs):
    • Validasi sintaksis dan integritas 847 blok kode dokumentasi lintas 155 skill (JSON schema strict parsing dan deteksi placeholder slop).
  • Ephemeral Micro-Sandbox Engine (vibes sandbox, scripts/wasm-sandbox.mjs):
    • Mesin eksekusi kode terisolasi berbasis node:vm berkecepatan ultra-tinggi (<2ms) dengan pengamanan batas waktu (timeout kill 200ms default), isolasi memori, penyadapan log konsol, dan penolakan statis identifier berbahaya (process, child_process, fs, require, __proto__). Terkoneksi ke CLI vibes sandbox <code|file> dan tool Remote MCP vibes_eval_sandbox.
  • Visual DAG Workflow Canvas di Studio Cockpit (scripts/studio-server.mjs):
    • Visualisasi alur kerja topologis DAG interaktif dengan konektor kurva Bézier SVG, pelacakan trace log simulasi real-time, dan 3 preset orkestrasi bawaan: zero-to-prod (8 fase), critic-validator (gate 5 langkah), dan agentic-commerce (alur settlement M2M).
  • Headless Concurrent Swarm Dispatcher (vibes swarm:dispatch, scripts/swarm-dispatcher.mjs):
    • Orkestrasi multi-agen asinkron non-blocking dengan pembatasan konkurensi (concurrency throttling), auto-binding instruksi domain dari 155 skill, verifikasi sandbox in-memory, audit Critic-in-the-Loop, dan sintesis kontrak lintas domain terpadu. Terintegrasi ke CLI (vibes swarm:dispatch) dan tool Remote MCP (vibes_swarm_dispatch).
  • Local FinOps & Token Savings Tracker (vibes finops, scripts/finops-tracker.mjs):
    • Pelacak telemetri konsumsi token lokal dan kalkulator penghematan biaya prompt caching (KV-cache 75–90% discount) untuk model Gemini 4 Pro, Claude 5.1, dan GPT-5. Terkoneksi ke CLI (vibes finops [status|record|reset]), API Studio Cockpit (/api/finops), dan tool Remote MCP (vibes_finops_report).
  • Native VSCode & Cursor Extension Companion (extensions/vscode/):
    • Paket ekstensi siap pakai untuk VSCode, Cursor, dan Windsurf: indikator status bar 155 Skills Active, 1-klik buka Studio Cockpit, Command Palette runner untuk Scaffolding Wizard, Sovereign Guard, Active Document Critic, dan FinOps report.
  • Official Reusable GitHub Action (.github/actions/vibes-guard/action.yml):
    • Composite GitHub Action siap pakai untuk menjalankan Sovereign Anti-Slop Gate dan Critic-in-the-Loop Validator di repositori CI/CD mana pun.
  • 5 Skill Frontier Baru (Milestone 155 Skills!):
    • thinking-budget-orchestrator: Pengaturan dinamis budget token penalaran (thinking budget) lintas model frontier (Gemini 4 Pro, Claude 5.1 Fable/Mythos, GPT-5/o3) dengan 4 profil teruji (speed, balanced, deep, exhaustive).
    • structured-output-contracts: Standardisasi skema output terstruktur multi-bahasa (Pydantic v2, Zod v3, Rust Serde, Go Struct) dengan garansi zero-markdown-fencing dan validasi skema runtime ketat.
    • mcp-sampling-critic: Integrasi Model Context Protocol (MCP v1.x) sampling/createMessage untuk review gate Critic-in-the-Loop otonom sebelum kode atau PRD diserahkan.
    • visual-ui-auditor: Audit kepatuhan visual antarmuka (Pillar 7 Anti-Slop) berbasis Multimodal Vision LLM dengan rubrik 20 poin (kontras, hierarki, tabular numerals, 5-state contract).
    • parallel-tool-orchestrator: Eksekusi batch paralel multi-tool otonom (vibes_batch) dengan dependensi topologis DAG dan circuit breaker terisolasi.
  • CLI Commands Baru:
    • vibes critic <file>: Audit file kode / PRD melalui arsitektur Critic-in-the-Loop.
    • vibes visual-audit <file|code>: Audit kepatuhan visual UI kode / tangkapan layar terhadap Pilar 7.
  • Skill Baru: spec-driven-development-expert:
    • Panduan ahli untuk Spec-Driven Development (SDD): TypeSpec, OpenAPI 3.1 contract-first design, Behavior-Driven Development (Cucumber/Gherkin BDD), Consumer-Driven Contract Testing (Pact), dan mock virtualization (Prism).
  • Skill Baru: agentic-m2m-settlement:
    • Protokol kliring dan settlement otonom Machine-to-Machine (M2M) dengan Dual-Rail (Fiat DOKU SNAP BI QRIS/VA & Crypto ERC-4337 Base/USDC), FinOps velocity caps, dan AP2AP invoicing.
  • Skill Baru: browser-extension-mv3-pro:
    • Panduan ahli rekayasa ekstensi peramban modern berbasis Manifest V3 (MV3): Chrome SidePanel API, Background Service Worker lifecycle, Offscreen Document WASM/Audio sandbox, dan declarativeNetRequest.
  • Test Suite Otomatis CLI (scripts/test-cli.mjs):
    • Unit test suite berbasis node:test untuk memverifikasi fungsionalitas CLI: --help, version, doctor, search, recipe list, dan skill info.
  • GitHub Actions CI & Anti-Slop Gate (.github/workflows/anti-slop-gate.yml):
    • Workflow CI otomatis untuk memvalidasi kepatuhan 150 skill, menjalankan CLI test suite, dan audit Sovereign Anti-Slop ketat pada setiap push/PR ke branch main.
  • Paket Resep Baru CLI (bin/vibes.mjs):
    • agentic-commerce: Bundel 7 skill untuk ekosistem agen niaga & settlement otonom.
    • frontier-ai: Bundel 6 skill untuk rekayasa frontier model, KV-cache locking, dan MCP.
    • spec-driven-api: Bundel 6 skill untuk arsitektur API contract-first, TypeSpec, BDD, dan Pact testing.
  • Peningkatan TUI Interaktif (vibes ui):
    • Menu selektor 3-mode: instalasi instan berdasarkan Paket Resep (Curated Recipes), Eksplorasi per Kategori Domain (Frontend, Backend, AI, SaaS, Keamanan, Data), atau kustomisasi penuh dari 150 skill.
  • Perintah Generator Katalog Dokumentasi (vibes docs):
    • Perintah baru vibes docs [markdown|json] [outputFile] untuk menghasilkan tabel markdown atau JSON catalog dari seluruh 150 skill secara otomatis (CATALOG.md & CATALOG.json).
  • Vibes-Plug Studio Cockpit & Visual Recipe Builder (vibes studio, scripts/studio-server.mjs):
    • Dashboard web lokal interaktif dengan arsitektur single-page application (SPA) tanpa dependensi eksternal (menggunakan native node:http).
    • Fitur lengkap: Visualisasi 150 skill dengan filter 6 domain dan pencarian instan, topologi visual 5 subagent spesialis, status live engine P1–P5 (KV-cache, AST, in-memory DB, PRM, SMT), drawer detail skill, dan penyuntikan skill 1-klik (/api/inject) ke .agents/skills/.
    • Visual Recipe Builder & Batch Injection: Seleksi multi-skill dengan checkbox, floating drawer tray, ekspor custom recipe JSON, generator CLI command, dan batch injection (POST /api/recipes/inject).
    • Remote Model Context Protocol (MCP v1.x) SSE Transport: Integrasi streaming Server-Sent Events (/sse & /messages) langsung di Vibes Studio dan CLI (vibes mcp --transport sse --port <n>), memungkinkan Claude Desktop, Cursor, dan remote AI agents terhubung langsung ke 150 skill dan P1–P5 tools.
    • Konsol Interaktif MCP: Tab konsol live di Studio untuk menguji eksekusi tool MCP secara langsung dengan pengukur latensi real-time dan konfigurasi 1-klik copy untuk IDE.
  • Perintah CLI Baru (vibes guard & vibes mcp):
    • vibes guard: Menjalankan validasi Sovereign Anti-Slop (281 berkas) dan seluruh rangkaian pengujian CLI (13 test suite) dalam satu perintah terpadu.
    • vibes mcp: Menjalankan server Model Context Protocol bawaan baik melalui mode stdio maupun mode remote streaming HTTP/SSE (--transport sse --port 3333).
  • Peningkatan Template Scaffolding (vibes bootstrap & vibes init):
    • Menambahkan template ke-6: spec-api untuk inisialisasi Contract-First API berbasis Hono/Node.js, TypeSpec, dan OpenAPI 3.1.
    • Memperkaya dokumen bawaan (PRD.md, BLUEPRINT.md, PROGRESS.md) dengan spesifikasi teknis konkret (PostgreSQL DDL schema, API JSON contract, NFR latency & accessibility budgets) sesuai Sovereign Anti-Slop Directive.
  • Dokumentasi Instalasi Universal via GitHub:
    • Menambahkan panduan instalasi lengkap via GitHub di README.md dalam 5 bahasa (English, Bahasa Indonesia, Mandarin, Deutsch, Japanese).
    • Mendukung eksekusi langsung via NPX tanpa kloning (npx github:roedyrustam/vibes-plug --all / npm i -g git+...).
    • Menambahkan panduan Git Clone untuk installer multi-platform (npm run install-all) dan kloning langsung ke direktori plugin Antigravity (~/.gemini/config/plugins/vibes-plug).
    • Menambahkan panduan injeksi rules terarah per-proyek untuk Claude Code, Cursor IDE, Codex, Windsurf, Trae, dan VSCode (Cline).

Changed / Diubah

  • Sinkronisasi versi paket ke 4.2.2 di package.json, bin/vibes.mjs, .cursorrules, dan seluruh direktif IDE.
  • Peningkatan katalog ekosistem dari 147 skill menjadi 150 skill tersinkronisasi penuh (Milestone 150 Skills).

[4.2.1] - 2026-10-06

Fixed / Diperbaiki

  • NPM Package Publishing Hotfix:
    • Penyelarasan registry staging & bump versi 4.2.1 untuk rilis resmi publik npm.

[4.2.0] - 2026-10-06

Added / Ditambahkan

  • OpenAI Codex CLI Native Integration:
    • Menambahkan file konfigurasi utama CODEX.md di root repositori khusus untuk OpenAI Codex CLI (codex) dan OpenAI coding agents.
    • Menambahkan .codex/rules/vibes-plug-core.md untuk modular rules engine platform Codex.
    • Memperbarui installer CLI scripts/install.js dengan flag --codex dan deteksi platform ~/.codex/ (global maupun per-project).
    • Mengintegrasikan verifikasi otomatis CODEX.md dan .codex/rules/ pada bin/vibes.mjs (vibes sync).
  • Superintelligence Engineering Substrate (P1 - P5 Engines):
    • P1: Canonical Prompt-Cache Compiler (scripts/compile-cache.mjs, vibes compile-cache): Mengompilasi prefix instruksi deterministik 147 skill ke dalam .cache-prefix.json (5.469 bytes / ~1.440 tokens), mengunci efisiensi prompt caching >1024 token untuk Anthropic Claude, OpenAI Codex/o3, Google Gemini, dan Cursor.
    • P2: In-Memory DB & Migration Sandbox (scripts/eval-migration.mjs, vibes db:eval): Menjalankan simulasi DDL/SQL di memori untuk menguji keamanan RLS, mencegah penghapusan destruktif (DROP TABLE/COLUMN), dan memverifikasi indeks Foreign Key sebelum dieksekusi ke DB produksi.
    • P3: AST Subgraph Query Engine (scripts/ast-query.mjs, vibes ast): Ekstraksi bedah simbol kode (symbols), isolasi fungsi (inspect), dan kalkulasi radius dampak dependensi (blast-radius) tanpa membanjiri context window Transformer.
    • P4: Speculative Multi-Draft & PRM Arbiter (scripts/speculative-arbiter.mjs, vibes arbitrate): Evaluator Process Reward Model (PRM) heuristik untuk menilai dan mengarbitrasi draf kode kompetitif berdasarkan Type Safety, Anti-Slop, Ketahanan Error, dan Kompleksitas Algoritma.
    • P5: Neuro-Symbolic Invariant Verifier (scripts/verify-invariants.mjs, vibes verify-invariants): Solver verifikasi formal transisi state machine (SaaS billing, auth, order flow), deteksi deadlock dan unreachable states, serta generasi spesifikasi SMT-LIB 2.0 untuk Z3/CVC5 theorem provers.
  • Native Model Context Protocol (MCP v1.x) Server (bin/mcp-server.mjs, vibes mcp, npm run mcp):
    • Menyediakan server MCP mandiri berbasis stdio JSON-RPC 2.0 yang mengekspos 8 tools AI frontier:
      1. vibes_ast_query: Query simbol AST bedah dan kalkulasi blast radius.
      2. vibes_eval_sql: Evaluasi skema DDL & audit kebijakan RLS in-memory.
      3. vibes_arbitrate_code: Arbitrase PRM multi-draf kode kompetitif.
      4. vibes_verify_invariants: Verifikasi formal invarian transisi state machine & generasi SMT-LIB 2.0.
      5. vibes_get_skill: Akses instan ke 147 skill spesialis langsung dari agen AI.
      6. vibes_compile_cache: Kompilasi dan verifikasi prefix KV-cache deterministik (>1024 token).
      7. vibes_memory: Manajemen memori episodik arsitektur dan pembuatan checkpoint ultra-kompak.
      8. vibes_swarm: Perencanaan dan eksekusi swarm multi-agen otonom.
  • Autonomous Swarm Runner (scripts/swarm-runner.mjs, vibes swarm, npm run swarm):
    • Engine orkestrasi swarm otonom dengan 3 topologi eksekusi 2026: Fan-Out/Fan-In (paralel multi-domain), Pipeline Saga (eksekusi fase dependen berurutan), dan Critic-Validator Loop (pintu gerbang kualitas anti-slop).
  • Local Episodic Memory Store & Daemon (scripts/memory-daemon.mjs, vibes memory, npm run memory):
    • Penyimpanan keputusan arsitektur lintas-sesi dalam .agents/memory/episodic_graph.json dan generator checkpoint ultra-kompak CHECKPOINT.md (<150 token) untuk mengatasi amnesia konteks pada obrolan berulang.
  • Unified Sovereign Guard (vibes guard, npm run guard):
    • Menggabungkan validasi seluruh ekosistem (147 skill), pemindaian Anti-Slop ketat (256 file), kompilasi KV-Cache (P1), simulasi database in-memory (P2), verifikasi formal SMT (P5), Swarm runner self-test, dan pembaharuan memory checkpoint dalam 1 pipeline CI otomatis.

Changed / Diubah

  • Universal Frontier AI Models Harmonization (2026 Edition):
    • Sinkronisasi dan harmonisasi menyeluruh pada seluruh 147 skill spesialis untuk model frontier terbaru:
      • Google Gemini: Gemini 4 Pro (2M+ context window reasoning), Gemini 4 Flash (low-latency execution), Project Astra (real-time multimodal video/audio streaming), dan Multimodal Live API.
      • Anthropic Claude: Claude 5.1 Fable / Mythos, Claude 5 Sonnet, dan Claude Code Agentic Swarm Patterns (Fan-Out/Fan-In, Pipeline Saga, Critic-Validator Loop).
      • OpenAI: GPT-5.6, GPT-5, o3, o3-mini (structured reasoning), dan GPT-4.5.
      • DeepSeek: DeepSeek-R1 (open reasoning), DeepSeek-Pro.
  • Sovereign Token Optimization & KV-Cache Protocol (Skema Hemat Token):
    • Menerapkan protokol hemat token berdaulat tanpa mengorbankan kualitas atau kelengkapan kode (Non-Degradation Invariant):
      • KV-Cache Determinism (kv-cache-prefix-optimizer): Penguncian prefix deterministik untuk menghemat biaya token 75-90% dan mempercepat TTFT hingga 10x.
      • Selective Context Slicing: Pangkas muatan konteks subagent saat Fan-Out hanya pada irisan CONTEXT_MAP.md dan kontrak tipe relevan, mencegah ledakan konteks kuadratik.
      • Adaptive Model Cascading (adaptive-model-cascade): Alokasi cerdas model tier rendah (flash/haiku) untuk riset/linter, dan model frontier (pro/opus) untuk arsitektur/keamanan.
      • Ultra-Compact Checkpointing (session-memory-manager): Checkpoint ringkas (<300 token) di setiap batas sesi.
  • Pillar 7 Anti-Slop (UI Sanitation) & 3Dviz Spatial Reasoning Integration:
    • Menggabungkan prinsip antislop-ui ke dalam skills/anti-slop/SKILL.md (Pilar 7: Sanitasi UI & Visual).
    • Mengintegrasikan prinsip 3dviz-pro-max ke dalam skills/web-3d-graphics-expert/SKILL.md dan skills/data-visualization-expert/SKILL.md (kerajinan 3 skala, kontinuitas join, amplop tabrakan, review 16 poin).
  • Ecosystem-Wide Version Bump to v4.2.0:
    • Seluruh 147 skill spesialis di skills/*/SKILL.md dinaikkan versinya secara resmi ke 4.2.0.
    • package.json, plugin.json, .cursorrules, AGENTS.md, CLAUDE.md, dan CLI bin/vibes.mjs diperbarui ke 4.2.0.
    • Skrip validasi scripts/validate-skills.mjs diperbarui untuk menegakkan standar verifikasi ketat version: "4.2.0" pada semua skill.

[4.1.0] - 2026-10-03

Added / Ditambahkan

  • 147-Skill Swarm Architecture Expansion (2026 Frontier Edition):
    • 15 skill baru ditambahkan ke ekosistem vibes-plug, meningkatkan total registry dari 132 menjadi 147 skill spesialis:
      1. context-window-engineer: Rekayasa context window ultra-besar (2M+ token), sliding window, dan token budgeting untuk Gemini 4 Pro, Claude 5.5, dan GPT Astra 6.
      2. screenshot-to-code-expert: Konversi tangkapan layar antarmuka, mockup, frame Figma, dan sketsa whiteboard menjadi kode frontend siap produksi via vision frontier models.
      3. adaptive-model-cascade: Kaskade dan routing model cerdas berbasis skor kompleksitas tugas (dari Flash/Haiku ke Sonnet/Opus/Astra) untuk efisiensi biaya token 40-60%.
      4. prompt-injection-firewall: Pertahanan berlapis terhadap injeksi prompt langsung, indirect injection, jailbreak, dan data exfiltration via tool calls.
      5. ai-code-review-autonomous: Review kode mandiri multi-pass otonom oleh AI tanpa tooling eksternal sebelum kode dipresentasikan ke pengguna.
      6. llm-observability-expert: Observabilitas LLM di produksi (Langfuse, Helicone, Lunary, OpenTelemetry GenAI) dengan pelacakan latensi, pemantauan halusinasi, dan analitik biaya token.
      7. property-mutation-testing-expert: Property-based testing (fast-check, Hypothesis) dan mutation testing (Stryker) untuk pembuktian invarian dan kualitas test suite.
      8. speculative-multi-draft-synthesizer: Penyusunan draf multi-agen spekulatif, hipotesis paralel, dan rekonsiliasi kode sadar-AST.
      9. kv-cache-prefix-optimizer: Rekayasa prefix prompt deterministik, KV-cache locking untuk mencapai hit rate 90%+.
      10. test-time-compute-optimizer: Penskalaan test-time compute, alokasi reasoning token dinamis (MCTS/PRM).
      11. living-codebase-ast-graph: Graf pengetahuan AST kode in-memory, pemetaan dependensi simbol real-time, dan pembersihan kode mati.
      12. autonomous-api-drift-healer: Deteksi drift skema API otonom, monitoring breaking change, dan pembuatan adapter SDK self-healing.
      13. ephemeral-wasm-sandbox-executor: Sandbox WebAssembly efemeral dan micro-runtime terisolasi untuk eksekusi kode AI yang aman.
      14. formal-spec-z3-verifier: Verifikasi formal matematis, SMT solver constraints (Z3, Dafny, TLA+), dan pembuktian invarian FinTech.
      15. multimodal-spatial-video-cloner: Reverse engineering UI/UX dari rekaman layar dan video spasial ke Tailwind CSS v4 & Framer Motion.
    • Master Orchestrator Synchronization:
      • skills/brainstorming/SKILL.md: Matriks orkestrator diperbarui penuh dalam Bahasa Inggris dan Bahasa Indonesia untuk seluruh domain terkait.
      • skills/zero-to-prod-orchestrator/SKILL.md: Alur kerja 8-Fase terintegrasi dengan skill baru pada Fase 1, Fase 4, Fase 5, Fase 6, dan Fase 7 (Bilingual: EN & ID).
  • DOKU Payment Gateway Architecture Upgrade (v5.1.0):
    • DOKU Checkout Integration:
      • Dukungan penuh Hosted Checkout dan Modal Popup (jokul-checkout-1.0.0.js) melalui /checkout/v1/payment.
      • Skema otentikasi Non-SNAP HMAC-SHA256 dengan Base64 Digest body SHA-256 dan API Pengecekan Status Order /orders/v1/status/{invoice}.
      • Panduan pemilihan solusi: Rekomendasi DOKU Checkout untuk 90% use case SaaS dan web application.
    • Direct API: Standar SNAP BI (Mandat Bank Indonesia):
      • Autentikasi dua lapis: OAuth B2B Access Token via Asymmetric RSA-SHA256 PKCS#1 v1.5 (/authorization/v1/access-token/b2b).
      • API Transaksional dengan Symmetric HMAC-SHA512 berdasar formula HTTPMethod:EndpointUrl:AccessToken:LowercaseHexBodyHash:Timestamp dan header CHANNEL-ID: H2H.
      • Cakupan lengkap Virtual Account SNAP (FIX_BILL, NO_BILL, BILL_VARIABLE_AMOUNT, PARTIAL_AMOUNT), reusable VA (DIPC/MGPC), serta callback inquiry dan notification.
      • Cakupan lengkap QRIS SNAP (/qr-mpm-generate, /qr-mpm-query, /qr-mpm-decode, /qr-mpm-cancel, dan /qr-mpm-refund).
    • Direct API: Non-SNAP:
      • Form Credit Card Payment Page dengan 3D Secure (3DS).
      • OVO Push Payment (/ovo-emoney/v1/payment) dengan parameter keamanan SHA-256 Checksum dan protokol polling/webhook timeout 70 detik.
    • Production-Ready TypeScript Service:
      • Service class DokuService lengkap, menangani token B2B caching, pembuatan invoice checkout, pembuatan VA & QRIS SNAP, dan verifikasi webhook berbasis raw-body.
      • Matriks Anti-Pattern Integrasi DOKU dan penegakan atomic idempotency transaksi.
  • DOKU MCP Server (v5.0.0):
    • Tool MCP baru create_doku_checkout_session untuk inisiasi transaksi hosted/modal.
    • Tool MCP baru check_doku_order_status untuk verifikasi status transaksi tingkat invoice.
    • Helper fungsi pembuat header autentikasi SNAP BI (generateSnapSignature) dan Non-SNAP (generateNonSnapSignature).
  • Payment & SaaS Billing Cross-Skill Synchronization:
    • payment-gateway-expert: Standardisasi pemisahan algoritma signature antara DOKU Checkout (HMAC-SHA256 + Digest) dan Direct API SNAP BI (HMAC-SHA512).
    • saas-billing: Integrasi eksplisit DOKU Checkout & SNAP BI ke dalam lanskap payment gateway SaaS Indonesia dengan panduan raw-body parsing dan proteksi race condition database.
    • brainstorming: Pembaruan matriks orkestrator domain SaaS, e-commerce, dan fintech untuk merekomendasikan doku-payment-gateway dan doku-mcp-server.

Changed / Diubah

  • Core Package & Universal Rules Alignment:
    • package.json, plugin.json, dan CLI bin/vibes.mjs dinaikkan versinya secara resmi ke 4.1.0 dengan total 147 skill registry.
    • AGENTS.md, CLAUDE.md, dan README.md diperbarui mencerminkan Swarm Director 147 spesialis.
    • scripts/install.js diperbarui untuk konfirmasi sinkronisasi 147 skill lintas Antigravity, Claude, dan Cursor.
  • Quality & Anti-Slop Audit:
    • Lolos 100% pemindaian ketat npm test (147 skill valid, 243 file AST anti-slop lolos dengan 0 error dan 0 warning).

[4.0.0] - 2026-10-01

Added / Ditambahkan

  • Sovereign Anti-Slop Directive (v4.0.0 Edition):
    • Penegakan doktrin anti-AI slop tanpa toleransi terhadap placeholder malas (// TODO, // ...), basa-basi percakapan, narasi sintaksis, dan kode halusinasi di seluruh agen, subagen, dan skill.
    • Injeksi Sovereign Anti-Slop Directive ke seluruh konfigurasi IDE (AGENTS.md, CLAUDE.md, .cursorrules, .windsurfrules, .traerules, .clinerules, .kimirules, .kirorules, .cursor/rules/vibes-plug-core.mdc, .claude/rules/vibes-plug-core.md).
    • Pemindai & Validator AST mandiri tanpa dependensi eksternal: scripts/check-anti-slop.js, check-anti-slop.mjs, dan skills/anti-slop/scripts/check-anti-slop.js dengan opsi --strict, --fix, dan --json.
    • Integrasi perintah terminal baru: vibes anti-slop, npm run anti-slop, npm run anti-slop:strict, dan npm run anti-slop:fix.
    • Pembaruan pipeline npm test yang mewajibkan validasi ketat ekosistem skill dan pemindaian anti-slop tanpa kompromi (zero slop errors).
  • UI/UX Anti-Slop Architecture Upgrade:
    • ui-ux-pro-max: 6 Pilar Anti-Slop UI/UX (pelarangan gradien neon AI ungu/sian generik, kontrak wajib 5-status komponen interaktif, struktur DOM semantik bebas div-soup, tata letak adaptif bebas horizontal clipping, mikro-animasi berfaedah yang menghormati prefers-reduced-motion, dan konten mock autentik).
    • design-system-architect: Kontrak Desain Komponen Anti-Slop dengan token kontras rasio WCAG 2.2 Level AA dan token @theme terstandarisasi.
    • hig: Poin pemeriksaan ke-6 Kebersihan Visual Anti-Slop pada protokol audit HIG.
    • tailwind-expert: Bagian 6 Arahan Anti-Slop Tailwind v4 (larangan hack nilai arbitrer, kewajiban cincin fokus aksesibel, eliminasi pembungkus redundant).
    • form-validation-expert: Arahan Anti-Slop Formulir (larangan pesan error samar, proteksi status pending double-submission, pengikatan error ARIA, hukum perlindungan input pengguna, dan paritas kontrak skema server-side).
    • accessibility-testing-expert: Gerbang Verifikasi Aksesibilitas Anti-Slop (larangan outline fokus terlucuti, elemen interaktif semantik, target sentuh minimum).
    • svg-animation-motion-expert: Arahan Anti-Slop Gerakan & Animasi (kepatuhan mutlak prefers-reduced-motion, animasi khusus compositor, eliminasi bouncy slop).
    • brainstorming: Mendaftarkan anti-slop pada domain UI/UX & Design Systems.

Changed / Diubah

  • Universal Ecosystem Version Bump to v4.0.0: Seluruh 132 skill di skills/*/SKILL.md dinaikkan versinya secara resmi ke version: "4.0.0".
  • Validation CI Enforcement: scripts/validate-skills.mjs diperbarui untuk menegakkan standar minimum versi 4.0.0 (major >= 4).
  • Core Package Alignment: package.json, plugin.json, dan README.md diperbarui serentak ke versi 4.0.0.

[3.10.0] - 2026-09-29

Added / Ditambahkan

  • Skill app-promo-media-expert: Panduan ahli untuk memproduksi media promosi aplikasi visual (gambar) dan video animasi gerak (App Store & Google Play screenshots, social cards Product Hunt/Twitter/LinkedIn, banner dinamis Open Graph via @vercel/og, video teaser terprogram dengan Remotion React, formula storyboard 30 detik berkonversi tinggi, dan otomasi headless screenshot via Playwright).
  • 132-Skill Ecosystem Expansion: Sinkronisasi 132 skill di brainstorming/SKILL.md (Domain File & Media), zero-to-prod-orchestrator/SKILL.md (Fase 5 & Fase 8), package.json, plugin.json, dan README.md.

[3.9.0] - 2026-09-29

Added / Ditambahkan

  • Universal Multi-Agent Swarm Integration: Dukungan konfigurasi aturan Swarm lintas IDE (AGENTS.md untuk Antigravity, CLAUDE.md untuk Claude Code, .cursorrules untuk Cursor, Windsurf, Trae, Cline, Kimi, dan Kiro).
  • 131-Skill Strict Validation: Seluruh 131 skill diverifikasi dengan pipeline CI ketat (scripts/validate-skills.mjs), memastikan kepatuhan frontmatter YAML v3.0+, bebas referensi usang, dan struktur dwibahasa (Inggris & Indonesia).
  • Zero Tech Debt Auditor Enhancements: Standardisasi penuh spesifikasi zero-tech-debt-auditor dengan panduan bilingual dan protokol eliminasi dead code (Knip Protocol).

Fixed / Diperbaiki

  • Frontmatter Deduplication: Membersihkan duplikasi metadata header pada 114 skill untuk menjaga integritas pembacaan skill oleh runner model.
  • Skill Count Synchronization: Sinkronisasi total 131 skill di seluruh dokumentasi package.json, plugin.json, README.md, dan CLI.

[3.8.0] - 2026-09-28

Added / Ditambahkan

  • Gemini 4 Pro (Next-Gen Readiness): Mengintegrasikan prinsip arsitektur masa depan ke orkestrator inti (gemini-agent-booster, multi-agent-orchestration, mcp-server-architect, rainstorming).
  • Determinisme Prefix KV-Caching: Mengunci struktur prompt statis (static prefix) pada orkestrator rainstorming/SKILL.md untuk memangkas latensi TTFT (Time-to-First-Token) hingga 90% pada konteks raksasa.
  • Agentic MoE & Swarm Paralel Masif: Mendorong eksekusi subagent massal serentak (batch tooling) dan penggunaan state memori monolitik daripada RAG terfragmentasi per agen.
  • High-Throughput MCP Server: Panduan penanganan konkurensi (pemanggilan serentak via Promise.all atau syncio.gather) di mcp-server-architect untuk mengimbangi kecepatan model-model frontier gen-4.
  • Native Any-to-Any Multimodal Tokens: Standarisasi integrasi API tanpa melalui perantara STT/TTS pada arsitektur agen real-time.

[3.7.0] - 2026-09-24

Added / Ditambahkan

  • Standar SNAP BI 1.0 (Bank Indonesia): Implementasi penuh standar SNAP BI pada doku-payment-gateway dan payment-gateway-expert. Menggunakan enkripsi Asimetrik RSA-SHA256 untuk B2B Access Token (/api/v1.0/access-token/b2b) dan Simetrik HMAC-SHA512 untuk endpoint transaksional (Virtual Account, QRIS, E-Wallet).
  • Pencegahan Race Condition & Atomic Idempotency:
    • Pola Atomic Update (UPDATE ... WHERE status = 'PENDING') dan penguncian tabel unik pada saas-billing dan payment-gateway-expert. Mencegah saldo/langganan bertambah ganda saat webhook DOKU/Stripe terpanggil serentak.
    • Kewajiban penggunaan Raw Body Parser (req.text() atau express.raw({ type: 'application/json' })) untuk menjamin signature validasi tidak rusak akibat spasi atau serialisasi ulang JSON.
  • Swarm Execution Topologies (2026 Master Protocol):
    • Protokol pendelegasian subagent di zero-to-prod-orchestrator dan brainstorming: Fan-Out / Fan-In (paralel lintas UI, Backend, DB), Pipeline Saga (eksekusi 8-Fase berurutan), dan Critic-Validator Gate (audit mandiri sebelum serah terima).
    • Integrasi loop pengujian otomatis perbaikan mandiri (autonomous-tdd-debugger) di senior-frontend dan js-backend-expert.

Changed / Diubah

  • DOKU Gateway Integration: Migrasi penuh dari legacy Jokul API v2 (HMAC-SHA256) ke standar SNAP BI (HMAC-SHA512).
  • Frontend & Backend Standards: Penegasan Tailwind CSS v4 CSS-first (@theme), Next.js 15 Partial Prerendering (PPR), React 19 Compiler tanpa useMemo/useCallback, dan Node.js 24 LTS.
  • Version Bump: Versi aplikasi dan plugin dinaikkan dari 3.6.0 ke 3.7.0 di package.json, plugin.json, bin/vibes.mjs, dan CHANGELOG.md.

Fixed / Diperbaiki

  • Typo Skill References: Memperbaiki salah ketik referensi skill ( utonomous-red-teamer → autonomous-red-teamer) di brainstorming/SKILL.md dan zero-to-prod-orchestrator/SKILL.md.
  • Duplicate ORM Reference: Membersihkan penyebutan ganda database-orm-expert pada matriks orkestrasi js-backend-expert.

[3.6.0] - 2026-09-14

Added / Ditambahkan

  • CLI: vibes search <query>: Pencarian semantik berbasis frekuensi kata kunci di seluruh 125 skill. Hasil diurutkan berdasarkan relevansi (score). Contoh: vibes search payment → 18 skill relevan, dipimpin oleh doku-mcp-server dan payment-gateway-expert.
  • CLI: vibes recipe list: Tampilkan 6 built-in skill bundles yang dikurasi dengan deskripsi dan daftar skill di tiap resep.
  • CLI: vibes recipe apply <name>: Install semua skill dari sebuah resep ke .agents/skills/ sekaligus. Skip skill yang sudah terinstall. 6 resep tersedia:
    • ai-stack — AI/LLM + RAG + Vector DB + Multi-Agent
    • fullstack-pro — Next.js 15 + Auth + DB + Payments + Testing
    • security-hardened — Zero-Trust + Rate-Limit + GDPR + Red Team
    • realtime-app — WebSockets + SSE + CRDT + State
    • mobile-first — Expo + PWA Offline + Push Notifications
    • data-platform — ETL + Visualization + Telemetry + Search

[3.5.0] - 2026-09-14

Added / Ditambahkan

  • CLI: vibes skill info <name>: Tampilkan metadata lengkap sebuah skill (name, version, description, author, orchestration links) langsung dari terminal. Juga menunjukkan apakah skill sudah terinstall di project lokal.
  • CLI: vibes hooks install: Install Git pre-commit hook yang menjalankan Anti-Slop Audit secara otomatis sebelum setiap git commit. Melindungi codebase dari AI placeholder code.
  • CLI: vibes hooks remove: Hapus Git pre-commit hook dengan aman.
  • Bootstrap: 3 template baru — mobile (Expo SDK), api (Hono/Node.js), fullstack (T3 Stack). Total kini 5 template: saas, ecommerce, mobile, api, fullstack. Setiap template dilengkapi dengan skill set yang sesuai domain.

[3.4.0] - 2026-09-14

Added / Ditambahkan

  • CLI: vibes version current: Tampilkan versi saat ini dari package.json.
  • CLI: vibes version bump <major|minor|patch>: Bump versi secara terpusat dan sinkron di package.json, plugin.json, vibes.mjs, dan CHANGELOG.md sekaligus — eliminasi human error saat rilis.
  • CLI: vibes doctor: Health diagnostics — memeriksa Node.js version, npx, global skills directory, @inquirer/prompts, dan local .agents/skills/ sekaligus dalam satu perintah.

Changed / Diubah

  • CJS → ESM Migration: Konversi scripts/check-anti-slop.js → scripts/check-anti-slop.mjs dan scripts/update_skills.js → scripts/update_skills.mjs menggunakan import/export syntax. Seluruh codebase kini 100% ESM.
  • package.json scripts: Ditambahkan "audit" dan "update-skills" npm script agar bisa dijalankan via npm run audit dan npm run update-skills.
  • runAudit: Diperbarui untuk menggunakan check-anti-slop.mjs (ESM) menggantikan versi CJS lama.

[3.3.0] - 2026-09-14

Added / Ditambahkan

  • CLI: vibes list [filter]: New command to list all 125+ skills from the global registry. Shows which skills are already installed locally (✅ marker). Supports optional keyword filter (e.g., vibes list saas → shows 3 saas-related skills).
  • CLI: vibes remove <skill-name>: New command to cleanly uninstall a skill from the local project's .agents/skills/ directory. Simetris dengan vibes add.
  • CLI: vibes bootstrap: Super-scaffold a full Next.js 15 app with auto-injected AI skills. Templates: saas and ecommerce.
  • CLI: vibes ui: Interactive TUI using @inquirer/prompts for visual multiselect skill installation.

Fixed / Diperbaiki

  • Hardcoded skill count di scripts/validate-skills.mjs — sebelumnya hardcode "All 124 skills", sekarang dinamis menggunakan variabel totalSkills.
  • Missing version tag di frontmatter frontier-ai-models-expert/SKILL.md — menyebabkan validation gagal. Sekarang version: "3.0.0" ditambahkan.

[3.2.0] - 2026-09-13

Added / Ditambahkan

  • CLI Enhancements:
    • Added vibes create-skill <skill-name> command to easily scaffold new skills using a standard template.
    • Added vibes audit command to manually run the Anti-AI Slop quality gate check.
  • New Frontier Skill: frontier-ai-models-expert:
    • Added a dedicated skill for late-2026 frontier AI models, including Claude 5.1 (Fable/Mythos) and Project Astra (Gemini 3.1).
    • Covers Enterprise Frontier Safeguards (EFS), zero data retention, and multimodal capabilities.
    • Orchestrated in brainstorming and zero-to-prod-orchestrator.

Changed / Diubah

  • Synchronized orchestrator matrices in brainstorming/SKILL.md and zero-to-prod-orchestrator/SKILL.md to include frontier-ai-models-expert.
  • Bumped versions in package.json and plugin.json to 3.2.0.

[2.13.0] - 2026-09-11

Added / Ditambahkan

  • New Frontier Skill: anti-slop (145+ Skills Total):
    • Added dedicated zero-tolerance Anti-AI Slop quality gate skill (skills/anti-slop/SKILL.md).
    • Covers the 5 Pillars of AI Slop: elimination of conversational sycophancy/fluff, strict ban on lazy code truncation (// TODO, // ... rest of code unchanged, mock arrays in production), speculative over-engineering avoidance, elimination of syntax-narrating decorative comments, and enforcement of high technical density documentation.
    • Added automated pre-commit/CI validator script scripts/check-anti-slop.js to catch AI slop and placeholder omissions deterministically.
  • Bilingual Gold Standard Compliance:
    • Restructured and upgraded graph-rag-knowledge-expert, pwa-offline-first-expert, and voice-ai-realtime-agent to provide 100% compliant English and Bahasa Indonesia sections with standard bilingual anchor links and ## Integrasi Orkestrasi.

Changed / Diubah

  • Orchestrators & Swarm Integration:
    • Registered anti-slop in skills/brainstorming/SKILL.md across Testing & Security and Execution Handoff matrices.
    • Integrated anti-slop into skills/zero-to-prod-orchestrator/SKILL.md (Phase 1, Phase 6 Code Quality Audit, and general token efficiency guidelines in both English & Indonesian).
    • Cross-referenced anti-slop in production-ready-hardener, vibe-code-gardener, and token-saver.
    • Updated all ecosystem documentation and counts to 145+ skills (package.json, plugin.json, AGENTS.md, CLAUDE.md, BLUEPRINT.md, README.md).

[2.12.0] - 2026-09-10

Added / Ditambahkan

  • Ecosystem Expansion to 144+ Skills: Introduced 4 frontier 2026 AI & Agentic engineering skills:
    • vercel-ai-sdk-expert: Complete production guide for Vercel AI SDK (Core, UI, RSC) in Next.js 15 & React 19. Covers streamText, generateObject, multi-provider fallback switching (Anthropic, OpenAI, Google, Groq), tool-calling agent loops, and Server Actions streaming. (Panduan ahli Vercel AI SDK, streaming data terstruktur, dan integrasi AI).
    • deep-research-analyst: Autonomous deep research pipeline with recursive query decomposition, multi-source crawling (Crawl4AI / Firecrawl), credibility scoring, fact triangulation (>= 2 independent sources), and evidence graph synthesis with markdown citations. (Panduan ahli riset mendalam otonom, pencarian web iteratif, dan mitigasi halusinasi).
    • pydantic-ai-expert: Type-safe Python AI agent framework with Pydantic AI. Features runtime dependency injection (RunContext), model-agnostic routing, structured Pydantic schema validation, and graph workflows. (Panduan ahli pengembangan agen AI Python type-safe dengan Pydantic AI).
    • synthetic-data-finetuning-expert: End-to-end synthetic dataset generation pipeline with LLM-as-a-judge quality filtering, QLoRA fine-tuning with Unsloth, DPO alignment, and GGUF/Ollama model export for local SLMs. (Panduan ahli generasi data sintetis, fine-tuning QLoRA, DPO, dan ekspor GGUF/Ollama).

Changed / Diubah

  • Universal Multi-Platform & Swarm Orchestration:
    • Updated all core orchestrators (brainstorming, zero-to-prod-orchestrator, AGENTS.md, CLAUDE.md, .cursorrules, .cursor/rules/vibes-plug-core.mdc) to orchestrate the 144+ skills ecosystem.
    • Integrated deep-research-analyst into Phase 1 (Discovery & AI PRD Planning), vercel-ai-sdk-expert, pydantic-ai-expert, and synthetic-data-finetuning-expert into Phase 4 (Backend & AI Agents) and Phase 5 (Frontend).
    • Synchronized versions and documentation across package.json, plugin.json, README.md, BLUEPRINT.md, index.js, and GITHUB_PROMO_KIT.md.

[2.11.0] - 2026-09-08

Added / Ditambahkan

  • Ecosystem Expansion to 140+ Skills: Introduced 3 high-impact frontier 2026 skills:
    • voice-ai-realtime-agent: Ultra-low-latency (<300ms) bi-directional conversational voice AI with WebRTC, OpenAI Realtime API, Gemini Multimodal Live Audio (PCM 24kHz), Semantic VAD, and LiveKit Agents. (Panduan ahli AI suara percakapan real-time berlatensi ultra-rendah).
    • graph-rag-knowledge-expert: Knowledge Graphs and GraphRAG architectures for multi-hop relational retrieval, entity-relationship extraction, Microsoft GraphRAG hierarchical community detection, and Neo4j Text2Cypher. (Panduan ahli Knowledge Graph, GraphRAG, dan inferensi multi-hop).
    • pwa-offline-first-expert: Enterprise Local-First and Progressive Web App architectures with zero-latency local operations (OPFS SQLite, RxDB), conflict-free multi-device sync (ElectricSQL, PowerSync), and PWABuilder store packaging. (Panduan ahli arsitektur Offline-First & Local-First PWA).

Changed / Diubah

  • Skill Orchestrator Matrix Synchronization:
    • brainstorming: Mapped voice-ai-realtime-agent to AI & Communication domains, graph-rag-knowledge-expert to Database & AI domains, and pwa-offline-first-expert to Frontend & Mobile domains (English & Indonesian).
    • zero-to-prod-orchestrator: Linked new skills into Phase 3 (Database), Phase 4 (Backend APIs & AI Agents), and Phase 5 (Frontend & Mobile) (English & Indonesian).

[2.10.0] - 2026-09-08

Added / Ditambahkan

  • Ecosystem Expansion to 137+ Skills: Introduced 3 frontier 2026 specialized skills:
    • ai-evals-benchmark-expert: Automated evaluation framework for LLMs & AI agents using Promptfoo, DeepEval, Ragas, deterministic assertions, and regression benchmarking. (Panduan ahli evaluasi otomatis LLM & Agen AI).
    • local-slm-edge-ai-expert: Edge AI and Small Language Models (SLMs) execution directly in the browser and edge environments via WebLLM, Transformers.js v3, ONNX Runtime Web, and WebGPU with zero cloud latency. (Panduan ahli eksekusi SLM lokal dan AI edge di browser).
    • modern-css-native-expert: Cutting-edge 2026 native CSS guide covering CSS Anchor Positioning, @starting-style entry animations, View Transitions Level 2, Container Queries, and :has() relational selector. (Panduan ahli fitur CSS native modern 2026).

Changed / Diubah

  • Resilience & Security Hardening:
    • background-jobs-queue-expert: Deepened with complete production-grade BullMQ v5 recipes, Redis idempotency locks, Dead Letter Queue (DLQ) retry strategies with exponential backoff and jitter, and priority job processing.
    • autonomous-red-teamer: Deepened with automated adversarial test suites, prompt injection detection pipelines, SSRF filter bypass methodologies, and automated security patch remediation templates.
    • brainstorming & zero-to-prod-orchestrator: Synchronized domain matrices and Phase 4, Phase 5, and Phase 6 orchestration chains to natively coordinate all 137 skills.

[2.9.0] - 2026-09-08

Changed / Diubah

  • Production Code Recipes & 2026 Frontier Hardening: Upgraded 4 key AI & distributed execution skills with robust production-ready code recipes and 2026 standards:
    • browser-automation-expert: Updated vision models to Gemini 3.8 Flash and Claude 3.7 Sonnet (Computer Use); added concrete production recipes for Stagehand v0.4+ and Browser-Use with anti-bot evasion and Playwright stealth.
    • mcp-server-architect: Added full production FastMCP (Python) server template and @modelcontextprotocol/sdk (TypeScript) McpServer implementation featuring Streamable HTTP / SSE transport, typed tools, dynamic resources, and security guardrails.
    • vector-db-rag-expert: Upgraded to 2026 Deep RAG standards with Reciprocal Rank Fusion (RRF) hybrid search (combining BM25 and pgvector 0.8+ HNSW), Cross-Encoder Re-ranking via FlashRank, and Late Chunking architecture.
    • async-queue-temporal-expert: Added concrete Temporal.io TypeScript SDK implementation featuring the distributed Saga pattern with compensating rollbacks, and Trigger.dev v3 durable tasks with retry jitter. (Peningkatan Resep Kode Produksi & Kesiapan 2026: Meng-upgrade 4 skill kunci AI & eksekusi terdistribusi dengan resep kode siap pakai di lingkungan produksi dan standar 2026: Browser Automation, MCP Server Architect, Deep RAG & Vector DB, serta Temporal Async Queue.)

[2.8.0] - 2026-09-08

Changed / Diubah

  • Core AI Skills Upgrade (2026 Edition): Upgraded 3 foundational AI engineering skills to current 2026 frontier standards:
    • gemini-agent-booster: Upgraded to Gemini 3.x ecosystem (Gemini 3.8 Flash, Gemini 3.5/3.1 Pro/Flash) with 1M–2M token context, native cachedContent Context Caching, dynamic thinkingBudget reasoning configuration, and Gemini Multimodal Live API bidirectional WebSocket streaming.
    • ai-llm-integration-expert: Added Anthropic Claude 3.7 Sonnet (Hybrid/Extended Thinking), OpenAI o1/o3/o3-mini & GPT-4.5/4o, DeepSeek-R1 (MoE reasoning), Model Context Protocol (MCP) Streamable HTTP transport, MCP Sampling, and Vercel AI SDK 5.x/6.x reasoning token streaming.
    • multi-agent-orchestration: Formalized 5 Anthropic 2026 Core Agentic Design Patterns (Prompt Chaining, Routing, Parallelization, Orchestrator-Workers, Evaluator-Optimizer Loop), LangGraph v0.3+ persistent checkpointers, and OpenAI Agents SDK handoffs & guardrails. (Pembaruan Skill Inti AI 2026: Meng-upgrade 3 skill AI utama mencakup Gemini 3.x, Claude 3.7 Hybrid Thinking, DeepSeek-R1, MCP Streamable HTTP & Sampling, dan 5 pola desain agentik modern Anthropic.)

[2.7.1] - 2026-09-04

Changed / Diubah

  • Complete 134-Skill Synchronization: Fully audited all 134 skills in skills/ and ensured 100% representation across README.md, BLUEPRINT.md, brainstorming/SKILL.md, and zero-to-prod-orchestrator/SKILL.md. (Sinkronisasi Lengkap 134 Skill: Mengaudit seluruh 134 skill dan memastikan representasi 100% pada semua file dokumentasi dan orkestrator.)
  • Skill Orchestrator Matrix Update: Registered asisten-ramah and skill-baru in both English and Indonesian matrices in brainstorming/SKILL.md (Execution Handoff) and zero-to-prod-orchestrator/SKILL.md (Phase 1 / Fase 1) per mandatory plugin protocol. (Pembaruan Matriks Orkestrator Skill: Mendaftarkan asisten-ramah dan skill-baru pada orkestrator utama dalam bahasa Inggris dan Indonesia.)
  • README & Blueprint Alignment: Added detailed descriptions for all 47 previously uncataloged skills across the 9 core engineering domains in README.md, and regenerated BLUEPRINT.md with full 134-skill bilingual manifests. (Penyelarasan README & Blueprint: Menambahkan deskripsi mendalam untuk seluruh 47 skill yang sebelumnya belum terdaftar di README, dan memperbarui BLUEPRINT.md.)
  • Installer Version Fix: Corrected banner version display in scripts/install.js to match current release.

[2.7.0] - 2026-08-18

Added / Ditambahkan

  • Enhanced Swarm Topologies: Introduced 4 standardized Swarm Topologies (Hierarchical Star, Pipeline Saga, Peer-to-Peer Mesh, Critic-Validator Quality Gates) in AGENTS.md and multi-agent-orchestration/SKILL.md. (Topologi Swarm Tingkat Lanjut: Menambahkan 4 topologi swarm standar untuk eksekusi paralel dan sekuensial.)
  • Swarm Circuit Breakers & Dynamic Routing: Automated failure detection, retry thresholds, and intelligent fallback delegation across specialized skills. (Circuit Breaker Swarm & Perutean Dinamis: Deteksi kegagalan otomatis dan delegasi fallback cerdas.)
  • Multi-Platform Swarm Protocols: Unified execution standards across Antigravity (AGY) subagent spawning, Claude Code background tasks, and Cursor IDE Composer rules. (Protokol Swarm Multi-Platform: Standarisasi eksekusi terpadu di seluruh platform.)
  • Synchronized Swarm Architecture Diagram: Updated Mermaid architecture in README.md reflecting multi-platform entry points and the 9-domain swarm engine (134+ skills). (Diagram Arsitektur Swarm Tersinkronisasi: Diagram Mermaid diperbarui untuk mencakup semua platform dan 9 domain skill.)

[2.6.0] - 2026-08-18

Added / Ditambahkan

  • Universal Multi-Platform Support: Added full compatibility for Claude Code/Desktop (CLAUDE.md, .claude/rules/vibes-plug-core.md) and Cursor IDE (.cursorrules, .cursor/rules/vibes-plug-core.mdc). (Dukungan multi-platform universal: Menambahkan kompatibilitas penuh untuk Claude Code dan Cursor IDE.)
  • Cross-Platform CLI Installer: Created scripts/install.js for automated zero-friction environment deployment across Antigravity, Claude, and Cursor. (Script installer cross-platform: Membuat scripts/install.js untuk deployment otomatis.)
  • 20 New Specialized Skills (Expanding total skills from 114 to 134):
    1. ai-media-generation-expert: AI Image/Video/Voice synthesis (Flux, DALL-E, Sora, ElevenLabs, Whisper).
    2. headless-cms-expert: Sanity v3, Payload CMS 3.x, Strapi 5, Contentful, Storyblok.
    3. astro-framework-expert: Astro 5+ Content Collections, Islands Architecture, View Transitions.
    4. data-visualization-expert: Recharts, Tremor, D3.js v7, Chart.js 4, Apache ECharts.
    5. rich-text-editor-expert: Tiptap v2, Lexical, ProseMirror, Yjs collaborative editing.
    6. svelte-sveltekit-expert: Svelte 5 Runes, SvelteKit 2, server actions.
    7. chatbot-messaging-expert: WhatsApp Business Cloud API, Telegram Bot, Discord.js, Slack Bolt.
    8. ecommerce-expert: Shopify Storefront API, Medusa.js v2, Saleor.
    9. documentation-site-expert: Mintlify, Docusaurus 3, Storybook 8, Astro Starlight.
    10. biome-linter-formatter-expert: Biome v2 (Rust-based linter/formatter replacing ESLint/Prettier).
    11. solidjs-expert: SolidJS 2, SolidStart, fine-grained reactivity.
    12. geospatial-maps-expert: Mapbox GL JS v3, Leaflet, Google Maps, PostGIS.
    13. pdf-document-generation-expert: React PDF, Puppeteer PDF, jsPDF, pdf-lib.
    14. search-engine-expert: Typesense, Meilisearch, Elasticsearch.
    15. n8n-automation-expert: n8n self-hosted workflows, custom nodes, AI chains.
    16. angular-expert: Angular 19+ Standalone, Signals, NgRx SignalStore, SSR Hydration.
    17. wordpress-headless-expert: WPGraphQL, ACF Pro, Faust.js, Next.js/Astro.
    18. desktop-electron-expert: Electron 33+ Context Isolation, typed IPC, auto-updater.
    19. blockchain-web3-expert: viem, wagmi v2, RainbowKit, EVM smart contracts.
    20. accessibility-testing-expert: axe-core, Pa11y, WCAG 2.2 Level AA/AAA automated testing.

Changed / Diubah

  • Updated orchestrator matrices in brainstorming/SKILL.md and zero-to-prod-orchestrator/SKILL.md (both EN and ID) to integrate all 20 new skills.
  • Synchronized skill counts to 134+ across README.md, CLAUDE.md, .cursorrules, package.json, and plugin.json.

Added / Ditambahkan

  • Registered 6 previously undocumented skill modules in BLUEPRINT.md and README.md: autonomous-chaos-monkey, autonomous-red-teamer, hyper-context-synthesizer, llm-cost-arbitrage-router, post-quantum-crypto-migrator, self-healing-cloud-orchestrator. (Mendaftarkan 6 modul skill yang sebelumnya tidak terdokumentasi di BLUEPRINT.md dan README.md: autonomous-chaos-monkey, autonomous-red-teamer, hyper-context-synthesizer, llm-cost-arbitrage-router, post-quantum-crypto-migrator, self-healing-cloud-orchestrator.)

Changed / Diubah

  • Naming Standardization: Renamed 3 skill directories from underscore to kebab-case convention: asisten_ramah → asisten-ramah, skill_baru → skill-baru, ui_ux_expert → ui-ux-expert. (Standardisasi penamaan: Mengganti nama 3 direktori skill dari underscore ke konvensi kebab-case.)
  • Updated all cross-references to renamed skills across README.md, CONTRIBUTING.md, bootstrap-to-modern/SKILL.md, and skill-baru/SKILL.md. (Memperbarui seluruh referensi silang ke skill yang diganti nama di README.md, CONTRIBUTING.md, bootstrap-to-modern/SKILL.md, dan skill-baru/SKILL.md.)
  • Synchronized skill count from "103" to "109" across README.md (badge, mermaid, footer), BLUEPRINT.md (EN & ID headers), and all documentation. (Menyelaraskan jumlah skill dari "103" menjadi "109" di README.md (badge, mermaid, footer), BLUEPRINT.md (header EN & ID), dan seluruh dokumentasi.)
  • Synchronized BLUEPRINT.md version from v2.4.0 to v2.5.0 to match plugin.json. (Menyelaraskan versi BLUEPRINT.md dari v2.4.0 ke v2.5.0 agar sesuai dengan plugin.json.)

[2.4.0] - 2026-08-11

Added / Ditambahkan

  • Updated saas-transformer master orchestrator: Reconstructed 9-phase ASCII architecture map and enriched skill maps across all 9 transformation phases in both English and Bahasa Indonesia sections. (Pembaruan saas-transformer: Merestrukturisasi peta arsitektur ASCII 9-fase dan memperkaya pemetaan skill di seluruh 9 fase transformasi.)
  • Updated production-ready-hardener master orchestrator: Reconstructed 7-phase ASCII architecture map and enriched skill maps across all 7 pre-production hardening phases. (Pembaruan production-ready-hardener: Merestrukturisasi peta arsitektur ASCII 7-fase dan memperkaya pemetaan skill di seluruh 7 fase pengerasan pra-produksi.)
  • Added Windows standard output UTF-8 stream reconfiguration (sys.stdout.reconfigure(encoding='utf-8')) in scanner scripts (saas_transformation_scanner.py and production_readiness_scanner.py) to prevent terminal encoding errors. (Menambahkan rekonfigurasi encoding stream UTF-8 pada script scanner Python untuk mencegah error encoding di terminal Windows.)
  • Synchronized BLUEPRINT.md to list all 87 skill modules across both English and Bahasa Indonesia documentation sections. (Menyelaraskan BLUEPRINT.md untuk mencakup seluruh 87 modul skill pada dokumentasi Bahasa Inggris dan Bahasa Indonesia.)

Changed / Diubah

  • Restructured rules/ directory into a single AGENTS.md at the plugin root to ensure full compatibility and automatic rule loading in Antigravity IDE and Desktop. (Merestrukturisasi direktori rules/ menjadi satu file AGENTS.md di root plugin untuk memastikan kompatibilitas penuh dan pemuatan aturan otomatis di Antigravity IDE dan Desktop.)
  • Fixed UTF-8 character encoding corruptions (mojibake) across skill guidelines (SKILL.md) and project documentation (BLUEPRINT.md). (Memperbaiki kerusakan encoding karakter UTF-8 (mojibake) pada petunjuk skill dan dokumentasi proyek.)
  • Bumped plugin, package, and blueprint versions to v2.4.0. (Meningkatkan versi plugin, package, dan blueprint ke v2.4.0.)

[1.8.0] - 2026-08-09

Added / Ditambahkan

  • Added website-design-cloner skill module: Reverse-engineer website designs & templates directly from any live target URL into 1:1 code (Tailwind CSS v4 @theme, React 19, Next.js 15 App Router). (Menambahkan modul skill website-design-cloner: Mempelajari dan merekayasa balik desain situs web & template secara langsung dari URL target ke kode presisi 1:1.)
  • Integrated URL design cloning triggers and handoff protocols into web-scraper, ui-ux-pro-max, and brainstorming. (Mengintegrasikan pemicu duplikasi desain URL dan protokol handoff ke dalam web-scraper, ui-ux-pro-max, dan brainstorming.)

Changed / Diubah

  • Bumped plugin and package versions to v1.8.0. (Meningkatkan versi plugin dan package ke v1.8.0.)

[1.7.0] - 2026-08-09

Added / Ditambahkan

  • Automated PRD & Roadmap creation protocol: prd-architect and zero-to-prod-orchestrator now enforce generating both PRD.md and ROADMAP.md before any code implementation begins on new projects. (Protokol otomatisasi PRD & Roadmap: prd-architect dan zero-to-prod-orchestrator sekarang mewajibkan pembuatan PRD.md dan ROADMAP.md sebelum penulisan kode dimulai.)
  • Explicit Skill Orchestration & Handoff matrices added across domain expert skills (senior-frontend, js-backend-expert, ai-llm-integration-expert, saas-multi-tenant, auto-doc-updater). (Menambahkan matriks Skill Orchestration & Handoff eksplisit pada seluruh skill domain spesialis.)

Changed / Diubah

  • Upgraded master orchestrator brainstorming and zero-to-prod-orchestrator to seamlessly cross-reference and delegate to all 77 specialized skills in vibes-plug. (Memperbarui orchestrator master brainstorming dan zero-to-prod-orchestrator untuk mendelegasikan tugas secara dinamis ke seluruh 77 skill di vibes-plug.)
  • Bumped plugin and package versions to v1.7.0. (Meningkatkan versi plugin dan package ke v1.7.0.)

[1.5.0] - 2026-07-26

Added / Ditambahkan

  • Added session-handoff-resume skill module, providing zero-token loss session continuation across account switches and chat resets via ultra-compact STATE_HANDOFF.md checkpoints. (Menambahkan modul skill session-handoff-resume untuk menyimpan checkpoint STATE_HANDOFF.md super hemat token dan melanjutkan proyek secara instan saat ganti akun/sesi.)
  • Added multi-agent-orchestration skill module, providing expert-level guidelines for multi-agent systems, LangGraph, CrewAI, AutoGen, supervisor routing, state graphs, and Human-in-the-Loop guardrails. (Menambahkan modul skill multi-agent-orchestration yang menyediakan pedoman tingkat ahli untuk sistem multi-agen, LangGraph, CrewAI, AutoGen, perutean supervisor, dan gerbang persetujuan manusia.)
  • Added design-system-architect skill module, providing expert-level guidelines for enterprise UI design systems, Design Tokens, Radix UI/Base UI headless primitives, Tailwind CSS v4 @theme, CVA variants, and WCAG 2.2 AAA accessibility. (Menambahkan modul skill design-system-architect yang menyediakan pedoman tingkat ahli untuk design system UI, Design Tokens, headless primitives Radix UI/Base UI, Tailwind v4 @theme, CVA, dan aksesibilitas WCAG 2.2 AAA.)
  • Added mcp-server-architect skill module, providing expert-level guidelines for designing, building, and securing Model Context Protocol (MCP) servers across TypeScript, Python, and Go (stdio/SSE transports, Zod/Pydantic validation, security guardrails). (Menambahkan modul skill mcp-server-architect yang menyediakan pedoman tingkat ahli untuk merancang, membangun, dan mengamankan server Model Context Protocol (MCP) pada TypeScript, Python, dan Go.)
  • Added go-programming-expert skill module, providing expert-level guidelines for Go 1.23/1.24+ backend APIs, microservices, concurrency patterns, sqlc, net/http, Gin/Echo/Fiber, gRPC, and table-driven testing. (Menambahkan modul skill go-programming-expert yang menyediakan pedoman tingkat ahli untuk Go 1.23/1.24+ backend API, microservices, pola konkurensi, sqlc, net/http, Gin/Echo/Fiber, gRPC, dan testing.)
  • Added js-backend-expert skill module, providing expert-level guidelines for Node.js 22 LTS, Bun 1.2+, Deno 2.x, Fastify 5, Hono, Express 5, NestJS, Prisma 6, Drizzle ORM, WebSockets, and BullMQ background jobs. (Menambahkan modul skill js-backend-expert yang menyediakan pedoman tingkat ahli untuk Node.js 22 LTS, Bun 1.2+, Deno 2.x, Fastify 5, Hono, Express 5, NestJS, Prisma 6, Drizzle ORM, WebSocket, dan pemrosesan background job BullMQ.)

Changed / Diubah

  • Upgraded all master orchestrator skills (zero-to-prod-orchestrator, app-analyzer-optimizer, production-ready-hardener, saas-transformer) to seamlessly integrate and delegate tasks across all newly added 2026 skills (session-handoff-resume, mcp-server-architect, multi-agent-orchestration, design-system-architect, go-programming-expert, js-backend-expert, seo-geo). (Memperbarui seluruh skill master orkestrator (zero-to-prod-orchestrator, app-analyzer-optimizer, production-ready-hardener, saas-transformer) agar secara otomatis mengintegrasikan dan mendelegasikan tugas ke seluruh skill baru 2026.)
  • Standardized and updated all 48 skills within vibes-plug to match current 2026 technical relevance and industry standards. (Memperbarui dan menyelaraskan seluruh 48 skill di vibes-plug agar sesuai dengan relevansi teknis dan standar industri terkini tahun 2026.)
  • Updated AI/LLM skills (ai-llm-integration-expert, brainstorming) with Model Context Protocol (MCP), Vercel AI SDK 4.x/5.x, reasoning models (DeepSeek-R1/V3, Gemini 3.5/3.6, Claude 3.7), and HNSW vector search. (Memperbarui skill AI/LLM dengan Model Context Protocol (MCP), Vercel AI SDK 4.x/5.x, model penalaran, dan pencarian vektor HNSW.)
  • Updated Frontend & Mobile skills (senior-frontend, tailwind-expert, mobile-expo-expert, tauri-expert) with Next.js 15+, React 19, Tailwind CSS v4 @theme, Expo SDK 52+, React Native 0.76+ New Architecture, and Tauri v2.0+ stable. (Memperbarui skill Frontend & Mobile dengan React 19, Next.js 15+, Tailwind v4, Expo SDK 52+, React Native 0.76+ New Architecture, dan Tauri v2.0+.)
  • Updated Language & Runtime skills (python-programming-expert, rust-programming-expert, bun-runtime-expert) to Python 3.12/3.13+ (PEP 695 generics, uv, Ruff), Rust 2024 (v1.85+), and Bun 1.2+. (Memperbarui skill Bahasa & Runtime ke Python 3.12/3.13+, Rust 2024, dan Bun 1.2+.)
  • Updated Search & Scraping skills (seo-geo, web-scraper) with Generative Engine Optimization (GEO for AI Overviews, Perplexity, ChatGPT Search) and Crawl4AI / Playwright extraction engines. (Memperbarui skill Search & Scraping dengan Generative Engine Optimization (GEO) dan engine ekstraksi Crawl4AI / Playwright.)

[1.4.3] - 2026-07-13

Added / Ditambahkan

  • Added mvc-expert skill module, providing expert-level guidelines to refactor legacy/obsolete PHP spaghetti codebases into modern, secure, and scalable MVC architectures using modern PHP 8.2+ OOP features and PSR standards. (Menambahkan modul skill mvc-expert yang menyediakan pedoman tingkat ahli untuk merefaktor codebase PHP spageti lama/usang menjadi arsitektur MVC yang modern, aman, dan skalabel menggunakan fitur OOP PHP 8.2+ modern dan standar PSR.)

Changed / Diubah

  • Bumped project and plugin versions to v1.4.3. (Meningkatkan versi proyek dan plugin ke v1.4.3.)

[1.4.2] - 2026-07-13

Changed / Diubah

  • Improved technology version modernization script update_skills.js to support matching non-breaking space variants (like &nbsp; and \u0026nbsp;). (Meningkatkan skrip pencocokan standardisasi teknologi update_skills.js agar mendukung pencocokan variasi spasi non-breaking seperti &nbsp; dan \u0026nbsp;.)
  • Updated ui-ux-pro-max dataset to target Next.js 15 instead of Next.js 14 in non-breaking spaces documentation. (Memperbarui dataset ui-ux-pro-max untuk menargetkan Next.js 15 daripada Next.js 14 pada bagian dokumentasi spasi non-breaking.)
  • Documented missing auto-doc-updater and skill-baru skills in README.md and BLUEPRINT.md. (Mendokumentasikan skill auto-doc-updater dan skill-baru yang sebelumnya belum tercantum di README.md dan BLUEPRINT.md.)
  • Bumped project and plugin versions to v1.4.2. (Meningkatkan versi proyek dan plugin ke v1.4.2.)

[1.4.1] - 2026-06-30

Changed / Diubah

  • Updated ui-ux-expert and ui-ux-pro-max skill modules to add comprehensive design and implementation guidelines for modern, professional, and standard dashboard architectures. (Memperbarui modul skill ui-ux-expert dan ui-ux-pro-max untuk menambahkan panduan desain dan implementasi komprehensif bagi arsitektur dashboard yang modern, profesional, dan standar.)
  • Bumped project and plugin versions to v1.4.1. (Meningkatkan versi proyek dan plugin ke v1.4.1.)

[1.4.0] - 2026-06-25

Added / Ditambahkan

  • Added fullstack-expert skill module, providing expert-level guidelines for multi-language (TypeScript, Python, Go, Rust), multi-framework (Next.js, FastAPI, Gin, Axum) web development, API design patterns, database architectures, DevOps, and observability. (Menambahkan modul skill fullstack-expert yang menyediakan pedoman tingkat ahli untuk pengembangan web multi-bahasa (TypeScript, Python, Go, Rust), multi-framework (Next.js, FastAPI, Gin, Axum), pola desain API, arsitektur database, DevOps, dan observability.)
  • Added saas-transformer master orchestrator skill module to guide systematic 8-phase transformation of standard applications into production-grade multi-tenant SaaS platforms (covering database isolation, billing/Stripe, teams, and feature gating). (Menambahkan modul skill master orkestrator saas-transformer untuk memandu transformasi sistematis 8-fase dari aplikasi standar menjadi platform SaaS multi-tenant tingkat produksi (mencakup isolasi database, billing/Stripe, tim, dan feature gating).)
  • Added production-ready-hardener master orchestrator skill module to conduct automated 7-phase application pre-launch audits across security, performance, accessibility, testing, and deployment, including a Python-based diagnostic scanner tool. (Menambahkan modul skill master orkestrator production-ready-hardener untuk melakukan audit pra-peluncuran aplikasi 7-fase secara otomatis pada aspek keamanan, performa, aksesibilitas, testing, dan deployment, termasuk alat scanner diagnostik berbasis Python.)
  • Added/documented missing core skill modules in the plugin index, including tanstack-query-expert (asynchronous state management & caching), web-scraper (multi-strategy data extraction workflow), supabase-migration (database schema migration tracking), ui-ux-expert (responsive web design & layout optimization), and asisten-ramah (friendly conversational styling). (Menambahkan/mendokumentasikan modul skill inti yang sebelumnya belum tercantum di indeks plugin, termasuk tanstack-query-expert (manajemen state asinkron & caching), web-scraper (alur kerja ekstraksi data multi-strategi), supabase-migration (pelacakan migrasi skema database), ui-ux-expert (desain web responsif & optimasi tata letak), dan asisten-ramah (gaya percakapan ramah).)

[1.3.9] - 2026-06-19

Changed / Diubah

  • Executed audit and validation across all 29 skill modules in the repository to ensure standardization of frontmatter, modern tech stacks (React 19, Next.js 15, Tailwind CSS v4, TanStack Query v5, Bun v1.1+), trigger conditions formatting, and clean emoji encodings. (Melakukan audit dan validasi di seluruh 29 modul skill dalam repositori untuk memastikan standardisasi frontmatter, stack teknologi modern (React 19, Next.js 15, Tailwind CSS v4, TanStack Query v5, Bun v1.1+), format kondisi pemicu, dan pengodean emoji yang bersih.)

[1.3.8] - 2026-06-15

Added / Ditambahkan

  • Added secure-fuzz-testing skill module detailing coverage-guided fuzzing target creation (Atheris for Python, cargo-fuzz for Rust, native Go fuzzing), sanitizers configuration (ASan, MSan, UBSan), diagnostic analysis, and automated DevSecOps CI/CD pipelines integration. (Menambahkan modul skill secure-fuzz-testing yang mendetailkan pembuatan target fuzzing berbasis cakupan (Atheris untuk Python, cargo-fuzz untuk Rust, native Go fuzzing), konfigurasi sanitizer (ASan, MSan, UBSan), analisis diagnostik, dan integrasi pipa DevSecOps CI/CD otomatis.)

[1.3.7] - 2026-06-15

Added / Ditambahkan

  • Added python-programming-expert skill module detailing Python 3.12+ features, type parameters, generic validation schemas using Pydantic v2, structured concurrency with asyncio TaskGroups, database interactions with SQLAlchemy 2.0 and SQLModel, package management via uv/Poetry, linting with Ruff, and testing with pytest. (Menambahkan modul skill python-programming-expert yang mendetailkan fitur Python 3.12+, parameter tipe, skema validasi generik menggunakan Pydantic v2, konkurensi terstruktur dengan asyncio TaskGroups, interaksi database dengan SQLAlchemy 2.0 dan SQLModel, manajemen paket melalui uv/Poetry, linting dengan Ruff, dan pengujian dengan pytest.)

[1.3.6] - 2026-06-14

Changed / Diubah

  • Cleaned up obsolete metadata fields (github:, risk:, source:, and date_added:) from all skill definition files in the repository. (Membersihkan bidang metadata usang (github:, risk:, source:, dan date_added:) dari seluruh berkas definisi skill di repositori.)

[1.3.5] - 2026-06-14

Changed / Diubah

  • Added header banner image banner.png to the top of README.md. (Menambahkan gambar banner header banner.png di bagian atas README.md.)

[1.3.4] - 2026-06-14

Added / Ditambahkan

  • Added scalability-clean-code skill module outlining clean coding guidelines (SOLID, readability, DRY, KISS) and scalability architecture principles (Clean Architecture, decoupling, caching, database scale). (Menambahkan modul skill scalability-clean-code yang menguraikan pedoman kode bersih (SOLID, readability, DRY, KISS) dan prinsip arsitektur skalabilitas (Clean Architecture, decoupling, caching, database scale).)

[1.3.3] - 2026-06-14

Added / Ditambahkan

  • Added tailwind-expert skill module detailing Tailwind CSS v4 CSS-first configuration, OKLCH theme variables, responsive design rules, state modifiers, custom utilities, bundle optimization, and class merging. (Menambahkan modul skill tailwind-expert yang merinci konfigurasi CSS-first Tailwind CSS v4, variabel tema OKLCH, aturan desain responsif, modifikator status, utilitas kustom, optimasi bundle, dan penggabungan kelas.)

[1.3.2] - 2026-06-14

Added / Ditambahkan

  • Added firebase-security-expert skill module for Firebase security checks, including Firestore/Storage/Realtime Database rules, Service Account safety, GCP API key restrictions, and App Check. (Menambahkan modul skill firebase-security-expert untuk pemeriksaan keamanan Firebase, termasuk aturan Firestore/Storage/Realtime Database, keamanan Service Account, pembatasan API key GCP, dan App Check.)

[1.3.1] - 2026-06-14

Added / Ditambahkan

  • Added CONTRIBUTING.md containing detailed contribution guidelines for fork developers to create and submit new skills. (Menambahkan CONTRIBUTING.md yang berisi panduan kontribusi terperinci bagi pengembang fork untuk membuat dan mengirimkan skill baru.)

Changed / Diubah

  • Synced the metadata version in plugin.json to match the project version of 1.3.1. (Menyelaraskan versi metadata di plugin.json agar sesuai dengan versi proyek 1.3.1.)

[1.3.0] - 2026-06-12

Added / Ditambahkan

  • Added token-saver skill to enforce concise AI responses and minimal codebase rewrites. (Menambahkan skill token-saver untuk memaksakan respons AI yang ringkas dan meminimalkan penulisan ulang kode.)
  • Added tauri-expert skill outlining Tauri v2 best practices, IPC communication, and security capabilities. (Menambahkan skill tauri-expert yang menguraikan praktik terbaik Tauri v2, komunikasi IPC, dan kemampuan keamanan.)
  • Added prd-architect skill serving as a mandatory guardrail to generate and validate Product Requirements Documents (PRD) before generating code for new projects. (Menambahkan skill prd-architect yang berfungsi sebagai guardrail wajib untuk membuat dan memvalidasi Product Requirements Document (PRD) sebelum menghasilkan kode untuk proyek baru.)

Changed / Diubah

  • Standardized all 23 skill metadata (frontmatter) formats across the plugin. (Menstandardisasi semua 23 format metadata (frontmatter) skill di seluruh plugin.)
  • Standardized the trigger header to ## Kondisi Pemicu in all skills. (Menstandardisasi header pemicu menjadi ## Kondisi Pemicu di semua skill.)
  • Broadly updated technical relevance in existing skills (bumped to React 19, Next.js 15, Tailwind v4, TanStack Query v5, Bun v1.1+). (Memperbarui relevansi teknis secara luas pada skill yang ada (ditingkatkan ke React 19, Next.js 15, Tailwind v4, TanStack Query v5, Bun v1.1+).)
  • Fixed markdown script paths in ui-ux-pro-max and emoji encoding issues in ui_ux_expert. (Memperbaiki path skrip markdown di ui-ux-pro-max dan masalah pengkodean emoji di ui_ux_expert.)

[1.2.6] - 2026-05-24

Added / Ditambahkan

  • Created a new rust-programming-expert skill module for Rust programming (Rust 2024 / v1.85+). (Membuat modul skill baru rust-programming-expert untuk pemrograman Rust (Rust 2024 / v1.85+).)

[1.2.5] - 2026-05-24

Added / Ditambahkan

  • Created a new bun-runtime-expert skill module for Bun runtime (v1.3+). (Membuat modul skill baru bun-runtime-expert untuk runtime Bun (v1.3+).)

[1.2.4] - 2026-05-24

Changed / Diubah

  • Updated the brainstorming skill with 2026 modern web architecture guidance. (Memperbarui skill brainstorming dengan panduan arsitektur web modern 2026.)

[1.2.3] - 2026-05-24

Changed / Diubah

  • Updated the senior-frontend skill to target React 19 / Next.js 15 / Tailwind CSS v4. (Memperbarui skill senior-frontend untuk menargetkan React 19 / Next.js 15 / Tailwind CSS v4.)

[1.2.2] - 2026-05-24

Changed / Diubah

  • Comprehensively updated the senior-fullstack skill with professional, production-grade architectural guidance and code. (Memperbarui secara komprehensif skill senior-fullstack dengan panduan arsitektur dan kode tingkat produksi profesional.)

[1.2.1] - 2026-05-24

Changed / Diubah

  • Updated the saas-mvp-launcher skill file with state-of-the-art 2026 patterns. (Memperbarui file skill saas-mvp-launcher dengan pola mutakhir 2026.)