{
  "markdown": "# Workflow Skills\n\nA complete feature development workflow for AI coding agents. Structured SDLC phases, scope-based routing, and requirement traceability from PRD through to shipped code.\n\nWorks with **Claude Code**, **Cursor**, **Codex**, **OpenCode**, and **Gemini CLI**.\n\n## Quick Start\n\n### Claude Code / Cursor\n\n```bash\n/plugin install workflow@dirkkok-skills\n```\n\n### Codex\n\n```bash\ngit clone https://github.com/dirkkok101/skills.git ~/.codex/workflow-skills\nln -s ~/.codex/workflow-skills/skills ~/.agents/skills/workflow\n```\n\n### Gemini CLI\n\n```bash\ngemini extensions install https://github.com/dirkkok101/skills\n```\n\n### OpenCode\n\nSee [`.opencode/plugins/workflow.js`](.opencode/plugins/workflow.js).\n\n### Initialize Your Project\n\n```\n/workflow:init\n```\n\n## Skills\n\n| Command | When to Use |\n|---------|-------------|\n| `/workflow:init` | Starting a new project, setting up docs structure |\n| `/workflow:research` | Need competitive analysis, landscape review, prior art |\n| `/workflow:brainstorm` | New feature idea, exploring approaches |\n| `/workflow:discovery` | Complex feature requiring deep domain requirements |\n| `/workflow:prd` | Writing formal product requirements |\n| `/workflow:technical-design` | Architecture, API specs, data models |\n| `/workflow:plan` | Breaking design into implementation tasks |\n| `/workflow:beads` | Creating work packages from approved plan |\n| `/workflow:execute` | Implementing from approved beads |\n| `/workflow:review` | Multi-agent code review after implementation |\n| `/workflow:review-prd` | Quality gate between PRD and design |\n| `/workflow:review-design` | Quality gate between design and plan |\n| `/workflow:review-plan` | Quality gate between plan and beads |\n| `/workflow:review-beads` | Quality gate between beads and execute |\n| `/workflow:review-execute` | Post-execution bead satisfaction verification |\n| `/workflow:qa` | Browser-based QA testing |\n| `/workflow:benchmark` | Performance benchmarking |\n| `/workflow:security-audit` | OWASP + STRIDE security audit |\n| `/workflow:ship` | Create PR with changelog and traceability |\n| `/workflow:compound` | Capture learnings for future sessions |\n| `/workflow:diagnose` | Bug investigation and root cause analysis |\n| `/workflow:autoresearch` | Automated document quality convergence |\n\n## Pipeline\n\n```\nresearch ─> brainstorm ─> discovery ─> prd ─> technical-design ─> plan ─> beads ─> execute\n(optional)                 (COMP only)\n\n  execute ─> qa ─> benchmark ─> review ─> review-execute ─> security-audit ─> ship ─> compound\n             (opt)   (opt)                                      (opt)\n```\n\nQuality gates (`review-*`) are optional between pipeline stages.\n\n### Scope-Based Routing\n\nBrainstorm classifies features and routes to the right pipeline depth:\n\n| Scope | Pipeline |\n|-------|----------|\n| **BRIEF** (0-2 pts) | brainstorm → plan → beads → execute → review → ship |\n| **STANDARD** (3-4 pts) | brainstorm → prd → technical-design → plan → beads → execute → review → ship |\n| **COMPREHENSIVE** (5+ pts) | brainstorm → discovery → prd → technical-design → plan → beads → execute → review → ship |\n\n### Other Entry Points\n\n| Scenario | Path |\n|----------|------|\n| Known requirements | prd → technical-design → plan → beads → execute → review → ship |\n| Technical improvement | brainstorm → technical-design → plan → beads → execute → review → ship |\n| Bug fix | diagnose → fix / beads / brainstorm |\n| Pre-release check | qa → benchmark → review → security-audit → ship |\n\n## Key Concepts\n\n**Traceability Chain** — Requirements trace from PRD → Design → Plan → Beads → Execute → Review. Each phase references upstream artifacts by stable FR IDs.\n\n**Intent Over Implementation** — Beads carry objectives and acceptance criteria, not code. The executing agent writes code from codebase patterns.\n\n**Surgical Context** — Each bead specifies exactly which files to read. Context resets between beads to prevent drift.\n\n**Self-Regulation** — Execute and QA track cumulative risk scores. High scores trigger PAUSE or STOP to prevent doing more harm than good.\n\n**CONVERGE Mode** — Review skills auto-fix mechanical findings and re-review until 0 FAILs or convergence. Judgment calls are escalated to the user.\n\n**Stage Gates** — Every phase pauses for user approval via `AskUserQuestion` before proceeding. Five interaction patterns: Decision Gate, Comparison, Batch Review, Guided Review, Combined.\n\n## Project Structure After Init\n\n```\nyour-project/\n├── CLAUDE.md\n├── .beads/beads.db\n└── docs/\n    ├── prd/           # Requirements\n    ├── designs/       # Technical design\n    ├── plans/         # Implementation plans\n    ├── execution/     # Execute manifests\n    ├── reviews/       # Review reports\n    ├── learnings/     # Compound learnings\n    ├── adr/           # Architecture decisions\n    ├── patterns/      # Coding patterns\n    └── architecture/  # System architecture\n```\n\n## Prerequisites\n\n**Required:** `br` (beads-rust) — local issue tracking for beads\n\n```bash\ncargo install beads-rust\n```\n\n**Optional:** `agent-browser` for QA skill, project-level tools configured in your CLAUDE.md.\n\n## Running Tests\n\n```bash\nbash tests/run-all.sh\n```\n\nTests verify CSO compliance (descriptions are trigger-only), structural integrity (frontmatter, versions, no project-specific refs), and reference link validity.\n\n## Philosophy\n\n- **Scope-Routed** — Weighted signals route features to the right pipeline depth\n- **Traceable** — FR IDs chain through every phase from PRD to PR\n- **Documentation-First** — All phases produce permanent docs\n- **Intent Over Implementation** — Beads contain objectives, not source code\n- **Stack-Agnostic** — Skills work with any tech stack\n- **Continuous Learning** — `/compound` captures learnings for future sessions\n\n## License\n\nMIT\n",
  "bytes": 5800,
  "sha": "f9567462b866d240198a559f0c5007b46be78111fea92dc000bdad9daf000b2f",
  "repo_slug": "dirkkok101/skills",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_dirkkok101_skills_810ebc8c/readme"
}