agnidia/corpflow
Staged worktask pipeline with a durable state ledger, cross-plugin delegation, and dependency-DAG orchestration.
Use when coordinating agent handoffs, debugging multi-stage execution, writing audit.jsonl rows, or managing parallel agent workflows. Patterns for multi-agent coordination, handoffs, parallel execution, and error escalation.
Use when the user invokes $appstore or asks for the canonical Corpflow appstore workflow.
Use when the user invokes $arch-debt or asks for the canonical Corpflow arch-debt workflow.
Use when the user invokes $arch-decision or asks for the canonical Corpflow arch-decision workflow.
Use when the user invokes $arch-review or asks for the canonical Corpflow arch-review workflow.
Use when the user invokes $cc-update or asks for the canonical Corpflow cc-update workflow.
Use when evaluating ethical implications, applying constitutional principles, or reviewing harm potential. Constitutional principles, ethics, and behavioral guidelines for AI agents.
Use when writing or editing source comments or DocC (/// or //) in any context — worktask stages, direct edits, reviews, remediation. Compact source comment standard — comment the non-obvious WHY and contract, never the WHAT, history, or provenance.
Use when writing a stage handoff, sizing an artifact against its stage budget, or recovering a worktask after compaction. Per-stage token budgets, handoff frontmatter and template, and per-content compression techniques.
Use for budget management, model selection, and cost tracking and optimization of AI agent worktasks. Model and effort right-sizing, prompt-cache TTL, search batching, cost estimation formula, efficiency analysis.
Use when the user invokes $create-agent or asks for the canonical Corpflow create-agent workflow.
Use when delegating a worktask stage to an external plugin agent (apple-, system-, android-, frontend-, backend-developer, ai-engineer) or integrating a new plugin. Covers dispatch injection, handoff schema, AR consultation and the plugin contract.
Use when generating CSV export files for estimates, budgets, timelines, or reports. 13-category CSV export structure for Google Sheets import.
Use when the user invokes $design-accessibility or asks for the canonical Corpflow design-accessibility workflow.
Use when the user invokes $design-review or asks for the canonical Corpflow design-review workflow.
Use when the user invokes $design-specs or asks for the canonical Corpflow design-specs workflow.
Use when the user invokes $docs-audit or asks for the canonical Corpflow docs-audit workflow.
Use when the user invokes $docs-readme or asks for the canonical Corpflow docs-readme workflow.
Use when the user invokes $docs-release-notes or asks for the canonical Corpflow docs-release-notes workflow.
Use when DV is about to complete and metadata.requires_screenshots is true (the default) — the completion gate blocks otherwise. Captures DV screenshots (or an annotated diff for meta-work) as visual evidence for QA, DR and the PR.
Use when the user invokes $estimate or asks for the canonical Corpflow estimate workflow.
Use when estimating task complexity, effort, hours, or budget, or determining worktask tier and PL0 stage set. Complexity scoring (0-50 scale), T-shirt sizing, and mid-run stage escalation for project estimation.
Use when the user invokes $ethics-review or asks for the canonical Corpflow ethics-review workflow.
Use when the PL stage publishes a plan, a repeat /worktask runs in an existing .context/, or before any gh issue create / publish-pl-issue.sh call. Keeps one GitHub issue per .context/; later runs comment on it instead of duplicating.
Use when the user invokes $improve-yourself or asks for the canonical Corpflow improve-yourself workflow.
Use when handling production incidents or emergency hotfixes. Incident classification, hotfix worktask, rollback procedures, and post-mortem templates for IR stage.
Use when an agent pipes build, test, simulator, Monitor, or incident output via background Bash or tee. Route runtime log capture to .context/logs/; covers filename grammar, errors/<agent>.md vs logs/ split, and cleanup.
Use for /megatask, multi-issue batches, or any dependency-ordered fan-out of worktasks. Meta-orchestration of many worktasks across a GitHub milestone or explicit issue array — dependency/blocker DAG, priority ordering, isolated per-issue worktrees.
Use when the user invokes $milestone or asks for the canonical Corpflow milestone workflow.
Use when the user invokes $optimize-agent or asks for the canonical Corpflow optimize-agent workflow.
Use when the user invokes $optimize-command or asks for the canonical Corpflow optimize-command workflow.
Use when generating design mockups, creating .pen files, or integrating design tokens. Design mockup generation worktask using Pencil MCP tools for the Designer agent.
Use when the dv-screenshot-capture apple-canvas adapter runs, before swift run SnapshotHost. Detects SwiftUI View types with SwiftSyntax and appends a minimal #Preview block to modified View files that have no #Preview or PreviewProvider.
Use when the user invokes $product-requirements or asks for the canonical Corpflow product-requirements workflow.
Use when the user invokes $prompt-audit or asks for the canonical Corpflow prompt-audit workflow.
Use when preparing releases, generating changelogs, or checking deployment readiness. Semantic versioning from conventional commits, Keep-a-Changelog generation, and rollback plans for the RE stage.
Use when the user asks for a plan, an approach, a breakdown, "how would you tackle this", or scoping — even without the word "plan". Turn a free-form request into a context-aware plan (goal, scope, phases, effort, risks) and recommend /worktask.
Use when the user invokes $roadmap or asks for the canonical Corpflow roadmap workflow.
Use when conducting a security review, threat-modeling a diff, scanning for secrets, or triaging dependency and supply-chain risk. OWASP Top 10 checklist, STRIDE threat model, secrets scanner, and secure coding patterns for the SR stage.
Use when the ST stage runs or post-delivery user edits need classifying. Capture user edits at ST stage, classify them, propose scoped updates to agents/skills/commands that participated in the worktask; human-in-the-loop, never auto-applies.
Use when the user invokes $sprint or asks for the canonical Corpflow sprint workflow.
Use when creating, naming, or resolving worktask artifact paths. The .context/ folder layout — run-indexed stage artifacts, design and image assets, per-agent error files, runtime logs.
Use when the user invokes $tech-code-review or asks for the canonical Corpflow tech-code-review workflow.
Use when the user invokes $test-coverage or asks for the canonical Corpflow test-coverage workflow.
Use when the user invokes $test-plan or asks for the canonical Corpflow test-plan workflow.
Use when executing multi-stage worktasks, initializing tasks, or managing worktask state. Holds dynamic sizing, the orchestrator loop, and stage handoff rules.
Use when planning test strategy in the worktask PL (Planning) or AR (Architecture) stage, before implementation. Plan and architecture test templates, required tests per feature type, UI selectors, multi-substate UITests.
Use when writing chat replies, handoffs, reports, reviews, docs, PR bodies or plans — any prose a person reads. ASD-STE100 rules at ~80% — single-topic sentences, active voice, imperative steps, articles kept, one term per meaning, tables over prose.