{
  "markdown": "# Agentic Development Workflow\n\nA structured, 5-phase framework for building software with AI coding agents. Gives LLMs clear specifications, atomic tasks, and role-matched expertise — so you get predictable, high-quality output instead of hoping the AI figures it out.\n\n> **Note:** I created this project to better understand agentic software development methodology. There may be a bunch of stupid stuff here. For real projects, check out [Superpowers](https://github.com/obra/superpowers), [GSD](https://github.com/gsd-build/get-shit-done), [Spec Kit](https://github.com/github/spec-kit), [EPCC](https://github.com/aws-solutions-library-samples/guidance-for-claude-code-with-amazon-bedrock/blob/main/assets/claude-code-plugins/README.md), [oh-my-claudecode](https://github.com/Yeachan-Heo/oh-my-claudecode), [Compound Engineering](https://github.com/EveryInc/compound-engineering-plugin), [Agent Skills](https://github.com/addyosmani/agent-skills) and incorporate whatever works for you.\n\n## The Problem\n\nMost people use AI coding tools by throwing a vague prompt at an LLM and hoping for the best. This works for small tasks but falls apart for anything real:\n\n- The AI makes assumptions instead of asking\n- Specifications live in your head, not in documents the AI can read\n- Tasks are too big, so context degrades and quality drops\n- No verification — you merge code you haven't meaningfully reviewed\n- Knowledge is lost between sessions\n\n## How This Helps\n\nThis workflow breaks AI-assisted development into 5 phases with human gates between each:\n\n```\nPhase 1        Phase 2          Phase 3           Phase 4         Phase 5\nResearch  -->  Specification  -->  Task Breakdown  -->  Execution  -->  Verification\n          [GATE]             [GATE]               [GATE]          [GATE]\n```\n\n**Phase 1: Research** — Human gathers materials, AI performs deep research to fill gaps and cross-reference findings.\n\n**Phase 2: Specification** — AI generates architecture specs (SPEC.md) with milestones, schemas, API contracts, and diagrams. No code. Optional multi-role review (security, domain, product).\n\n**Phase 3: Task Breakdown** — Spec is decomposed into atomic tasks grouped into parallel waves, with role assignments and test separation.\n\n**Phase 4: Execution** — One task per session, role-matched agents, automated verification, human review via task review files, PR on approval.\n\n**Phase 5: Verification** — Per-milestone end-to-end testing and retrospective.\n\nEach phase produces persistent documents (SPEC.md, PLAN.md, PROGRESS.md, decision records, task reviews) that survive across sessions and prevent context rot.\n\n## What's Included\n\n| Component | Count | Description |\n|-----------|-------|-------------|\n| **Skills** | 8 | `init-project`, `research`, `spec`, `plan`, `execute`, `review`, `verify`, `auto` |\n| **Agents** | 13 | Software Architect, Backend Engineer, QA Engineer, Security Reviewer, and more |\n\n## Installation\n\n### Claude Code Plugin (recommended)\n\n```bash\n# Add the marketplace\n/plugin marketplace add wilsonkichoi/agentic_development_workflow\n\n# Install the plugin — select \"Install for you, in this repo only (local scope)\"\n/plugin install agentic-dev@wilsonkichoi-agentic-dev\n```\n\nWhen prompted for install scope, choose **\"Install for you, in this repo only\"** (local scope). This keeps the plugin scoped to your project rather than applying globally, which is recommended while the workflow is still evolving.\n\nAfter installation:\n- Skills are available as `/agentic-dev:research`, `/agentic-dev:spec`, etc.\n- Agents appear in `/agents` (e.g., `agentic-dev:backend-engineer`)\n- Auto-updates when the repo is updated\n\n### Copilot CLI\n\n```bash\ncopilot plugin install wilsonkichoi/agentic_development_workflow\n```\n\nAfter installation:\n- Skills are available as `/init-project`, `/research`, `/spec`, `/plan`, `/execute`, `/review`, `/verify`, `/auto`\n- Agents appear in `/agent` (e.g., `backend-engineer`, `qa-engineer`)\n- Plugins install globally to `~/.copilot/state/installed-plugins/`\n\n> **Note:** Copilot CLI does not support per-project plugin scoping. Installed plugins are always global — there is no `--scope` flag or disable/enable per workspace (unlike Gemini CLI). If you need project-level control, use Claude Code's local scope install or Gemini CLI's workspace scope instead.\n\n### Gemini CLI\n\nTo avoid polluting the global context, install the extension globally but keep it disabled by default. Enable it strictly on a per-workspace basis:\n\n```bash\n# 1. Install the extension from GitHub\ngemini extensions install https://github.com/wilsonkichoi/agentic_development_workflow\n\n# 2. Disable the extension globally (user scope) so it doesn't pollute all your projects\ngemini extensions disable agentic-development-workflow --scope user\n\n# 3. Navigate into specific projects and enable it locally (workspace scope)\ncd /path/to/your-project\ngemini extensions enable agentic-development-workflow --scope workspace\n```\n\nAfter installation:\n- Skills are available directly via the CLI (e.g., `gemini plan`).\n- Subagents are available via the `@` mention or interactively (e.g., `gemini --agent=backend-engineer`).\n\n### Local Plugin Testing\n\n```bash\nclaude --plugin-dir /path/to/agentic_development_workflow\n```\n\n## Updating\n\n### Claude Code Plugin\n\n```bash\n# Update marketplace listings\n/plugin marketplace update wilsonkichoi-agentic-dev\n\n# Or enable auto-updates in /plugin → Marketplaces → Enable auto-update\n```\n\n**Important:** After updating, start a new session (`/clear` or new terminal). Skill file paths are cached per session and `/reload-plugins` may not fully refresh them.\n\n### Copilot CLI\n\n```bash\n# Update the plugin\ncopilot plugin update agentic-dev\n\n# View installed plugins\ncopilot plugin list\n\n# Uninstall\ncopilot plugin uninstall agentic-dev\n```\n\n### Gemini CLI\n\n```bash\n# Update the extension\ngemini extensions update agentic-development-workflow\n\n# View installed extensions\ngemini extensions list\n\n# Uninstall\ngemini extensions uninstall agentic-development-workflow\n```\n\n## Usage\n\n### Starting a New Project\n\n```bash\n# From parent directory\n/agentic-dev:init-project ./my-project\n\n# Already inside the project directory\n/agentic-dev:init-project .\n```\n\nThis creates:\n\n```\nmy-project/\n├── CLAUDE.md                 # Project instructions (edit first)\n├── workflow/\n│   ├── research/manual/      # Place human research here\n│   ├── research/final/       # AI research output\n│   ├── spec/                 # SPEC.md, HANDOFF.md\n│   ├── plan/                 # PLAN.md, PROGRESS.md\n│   │   └── reviews/          # Per-task review files\n│   ├── decisions/            # Architecture Decision Records\n│   └── retro/                # Retrospectives\n├── src/\n└── tests/\n```\n\n### Running Phases\n\n| Phase | Command | What It Does |\n|-------|---------|-------------|\n| 1. Research | `/agentic-dev:research` | AI deep research on materials in `workflow/research/manual/` |\n| 2. Specification | `/agentic-dev:spec` | Generate SPEC.md with milestones, schemas, contracts |\n| 3. Task Breakdown | `/agentic-dev:plan` | Decompose into atomic tasks with wave grouping |\n| 4. Execution | `/agentic-dev:execute` | Implement one task with role-matched agent |\n| 4b. Review | `/agentic-dev:review` | Code review, fix plan validation, fix verification |\n| 5. Verification | `/agentic-dev:verify` | End-to-end testing + retrospective |\n| Auto | `/agentic-dev:auto` | Full pipeline (execute+review+fix+verify) per wave or milestone |\n\n### Using Agents (Phase 4)\n\nDuring execution, tasks are matched to specialized agents:\n\n| Task Type | Agent |\n|-----------|-------|\n| API endpoints, services | `agentic-dev:backend-engineer` |\n| UI components | `agentic-dev:frontend-developer` |\n| Database migrations | `agentic-dev:data-engineer` |\n| Test implementation | `agentic-dev:qa-engineer` |\n| Auth, encryption | `agentic-dev:security-engineer` |\n| Infrastructure, CI/CD | `agentic-dev:devops-engineer` |\n\nTest tasks run in **separate sessions** from implementation tasks — the QA agent sees only the spec and public interface, not the implementation. This prevents the AI from writing tests that match its code rather than testing the contract.\n\n## Key Concepts\n\n**Human gates** — Every phase produces reviewable artifacts. Nothing moves forward without human approval.\n\n**One task per session** — Fresh context for each task. No context rot.\n\n**Decision records** — Significant decisions are captured in `workflow/decisions/DR-NNN-*.md` with options, rationale, and threaded human/AI discussion.\n\n**Task review files** — Every task gets `workflow/plan/reviews/task-X.Y.md` with a work summary and space for `*FEEDBACK:*` / `*AI:*` discussion before PR creation.\n\n**Review loop** — After execution, `/agentic-dev:review` produces durable review files (`wave-mM-N.md` or appends to `task-X.Y.md`). Any AI session in any tool can read the review file and continue the loop: review issues, validate fix plans, verify fixes. No copy-paste between sessions — the review file is the connective tissue. Review and fixes operate on the wave's feature branch; the feature branch merges to `main` only after all issues are resolved.\n\n**Parallel waves** — Independent tasks run concurrently when they belong to different components, have frozen API contracts, and share no mutable state.\n\n**Test separation** — Tests are written by a QA agent in a separate session from the implementation agent, testing the spec contract rather than the implementation.\n\n## Contributing\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) for development setup, testing checklist, and PR guidelines.\n\n## Documentation\n\n- [WORKFLOW.md](WORKFLOW.md) — Full workflow reference with all phases, principles, cost optimization, and tool-specific guides\n- [CI_CD.md](CI_CD.md) — CI/CD pipeline design notes\n\n\n## References\n\n- [Superpowers](https://github.com/obra/superpowers) — composable skills, TDD-first, subagent-driven execution\n- [GSD (Get Shit Done)](https://github.com/gsd-build/get-shit-done) — meta-prompting, wave-based parallel execution, context rot mitigation\n- [Spec Kit](https://github.com/github/spec-kit) — spec-driven development, multi-agent support, extension ecosystem\n- [EPCC Workflow](https://github.com/aws-solutions-library-samples/guidance-for-claude-code-with-amazon-bedrock/blob/main/assets/claude-code-plugins/README.md) — explore-plan-code-commit, session continuity, feature tracking\n- [oh-my-claudecode](https://github.com/Yeachan-Heo/oh-my-claudecode) — reference for skill implementation and architecture\n- [Anthropic Skills](https://github.com/anthropics/skills) — official skills reference\n- [Agency Agents](https://github.com/msitarzewski/agency-agents) — role-based agent definitions (adapted into `agents/`)\n- [PUA](https://github.com/tanweai/pua) — debugging persistence framework (adapted into Phase 4)\n- [Compound Engineering](https://github.com/EveryInc/compound-engineering-plugin) — Agent-Driven Development using the power of LLM.\n- [Agent Skills](https://github.com/addyosmani/agent-skills) — Production-grade engineering skills for AI coding agents.\n\n## License\n\nMIT\n",
  "bytes": 11112,
  "sha": "2cea46d9e18139132b9a16829deb48467c085c57f59d2e8b544963587803c805",
  "repo_slug": "wilsonkichoi/agentic_development_workflow",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_wilsonkichoi_agentic_development_workflo_a709bf6d/readme"
}