A full AI engineering team as plain Markdown files.
14 specialized AI agents — architect, reviewer, security expert, DevOps engineer, and more — each a single Markdown skill file any agent runtime can load into any project. No lock-in. No configuration. Just drop them in.
They have opinions about your PR descriptions. They will not merge your branch if your commit says fix stuff. They are not polite about it.
I commit with intention. I branch with purpose. I review with honesty. I ship with confidence. I never push to main.
Each agent is a single .md file that describes a role, its responsibilities, standards, and how it communicates. Load one or all of them into Claude Code, Cursor, or any runtime that supports skill files.
- Clone the repo — get all 14 agent skill files in one command
- Load into your runtime — point Claude Code, Cursor, or your agent framework at the skills directory
- Invoke any agent — ask the Reviewer to check your PR, ask Security to audit your auth flow. They know their role. They have standards.
Every role a real software team needs — available as a skill file with impeccable standards.
| Agent | Role | What it does |
|---|---|---|
| The Scribe | Commits, PRs, Changelogs | Conventional Commits enforcement, PR quality checks, changelog generation |
| The Architect | System Design, ADRs | Spec-first development, task decomposition, architecture decision records |
| The Reviewer | Code Review | Five-axis review (Correctness, Readability, Architecture, Security, Performance) |
| The Tester | TDD, Coverage | Red-Green-Refactor cycle, test pyramid balancing, regression-first testing |
| The Debugger | Error Triage | Read → Reproduce → Hypothesize → Test → Fix protocol, CI diagnosis |
| The Auditor | Security | OWASP Top 10 review, dependency audit, secret scanning |
| The Herald | Releases | Semantic versioning, changelog translation, GitHub Releases |
| The Librarian | Documentation | README framework, ADR lifecycle, stale doc detection |
| The Doorman | Validation | Commit/PR validation, health checks, branch protection |
| The Oracle | Institutional Knowledge | Member authoring templates, naming validation, convention rationale |
| The Envoy | Cross-Provider Translation | Provider detection (7 runtimes), skill translation, convention validation |
| The Sentinel | Integrity | Cross-member contradiction detection, structural drift detection |
| The Warden | Code Health | Code smell detection, cyclomatic complexity enforcement, dead code audit |
| The Steward | Context Economy | Context budget management, member routing, provider cache strategies |
npm install --save-dev agenthood
npx agenthood init # interactive setup (~5 minutes)
npx agenthood check # verify everything is in placeMembers are loaded as context by your existing AI assistant. Works with Claude Code, GitHub Copilot, Gemini CLI, OpenAI Codex CLI, and CodeBuddy.
Execute members as real LLM agents that reason, act, and remember across sessions. Powered by a TypeScript runtime with Groq as the default free LLM provider.
npm run build # build the runtime (once)
agenthood list # see available agents
agenthood run the-scribe "write a commit message for the current diff"
agenthood run the-reviewer "review the changes in the last commit"
agenthood run the-architect "plan the implementation for issue #42"Set GROQ_API_KEY in your environment (free at console.groq.com), or use Ollama for fully offline execution.
Agenthood is published as open Agent Skills that any compatible runtime can discover and load.
npx skills add fworks-tech/agenthoodThis installs all 14 members from the Skills.sh registry. You can also browse the indexed collection on SkillsMP.
| Version | Milestone | Status |
|---|---|---|
| v1.0.0 | Operationally Ready — consistent APIs, test harness, health checks | ✅ Shipped |
| v1.1.0 | VS Code modernization, Python runtime bootstrap | ✅ Shipped |
| v1.2.0 | VS Code workspace observer event bus | ✅ Shipped |
| v1.4.0 | The Living Editor — Doorman SCM validation, Auditor on-save scanning, Reviewer Diagnostics | ✅ Shipped |
| v1.5.0 | Open Standard — SKILL.md migration for all 14 members | ✅ Shipped |
| v1.6.x | The Academy — Level 1 articles, GitHub Pages, npm via OIDC | ✅ Shipped |
| v2.0.0 | Foundation — TypeScript runtime: ILLMProvider, LLMRouter, ReActLoop, BaseAgent | 📋 Planned |
| v2.1.0 | Intelligence — 5-tier memory, LanceDB, AgenticRAG, Tree-sitter | 📋 Planned |
| v2.2.0 | The Full Society — OracleAgent, DiffImpactAnalyzer, IProtocol | 📋 Planned |
The framework runs on five core principles adapted from production AI agent systems. Read the architecture docs to understand how agents coordinate, prioritize, fail over, and stay safe.
| Principle | Document |
|---|---|
| Multi-agent orchestration & agent roles | agent-system.md |
| Priority queues & concurrency slots | concurrency-and-queues.md |
| Agent mode vs Ask mode | operating-modes.md |
| Multi-LLM support & automatic failover | provider-failover.md |
| Tool registry, scoping & safety caps | built-in-tools.md |
Architecture Decision Records in docs/adr/:
| ADR | Decision |
|---|---|
| ADR-001 | Markdown-based skills over code-based agents |
| ADR-002 | Conventional Commits as the commit standard |
| ADR-003 | Dual enforcement via bash hooks and commitlint |
| ADR-004 | 14 specialized agents over a general-purpose agent |
| ADR-005 | Orchestrator pattern over peer-to-peer agent communication |
| ADR-008 | TypeScript-native runtime over Python + DeepAgents |
| ADR-009 | Groq as the default free LLM provider |
| ADR-010 | LanceDB for vector storage |
Agenthood is agent-agnostic. Agents work with:
- Claude Code — via
.claude/skills/ - GitHub Copilot — via
.github/agents/ - Gemini CLI — via
GEMINI.md+ skills - OpenAI Codex CLI — via
AGENTS.md+ skills - CodeBuddy — via
.codebuddy/skills/
The TypeScript runtime (agenthood run) works alongside any of these — it calls agents as real LLM agents independently of which IDE you use.
npm install && npm run build
make setup # activates git hooks and commit template
npm test # run all tests
npm run typecheck # strict TypeScript check
npm run lint # ESLintagenthood/
├── README.md ← You are here
├── AGENTS.md ← Agent registry (agent-agnostic)
├── oath.md ← The oath
│
├── conventions/ ← Rules every agent follows
│ ├── .gitmessage
│ ├── commitlint.config.cjs
│ └── COMMIT_CONVENTION.md
│
├── members/ ← 14 agent skill files
│ ├── the-scribe/
│ ├── the-architect/
│ ├── the-reviewer/
│ ├── the-tester/
│ ├── the-debugger/
│ ├── the-auditor/
│ ├── the-herald/
│ ├── the-librarian/
│ ├── the-doorman/
│ ├── the-oracle/
│ ├── the-envoy/
│ ├── the-sentinel/
│ ├── the-warden/
│ └── the-steward/
│
├── rituals/ ← Scheduled automations
│ ├── morning-briefing.md
│ ├── the-inspection.md
│ ├── the-watchman.md
│ └── evening-report.md
│
├── portals/ ← External system connectors
│ ├── github.md
│ ├── linear.md
│ ├── jira.md
│ ├── slack.md
│ └── sentry.md
│
├── agentic-workflows/ ← Multi-step workflow templates
│ ├── README.md
│ ├── triage-issues.agent.md
│ ├── review-pr.agent.md
│ ├── diagnose-ci-failure.agent.md
│ └── sync-docs.agent.md
│
├── architecture/ ← Agent system design docs
│ ├── agent-system.md
│ ├── built-in-tools.md
│ ├── concurrency-and-queues.md
│ ├── operating-modes.md
│ └── provider-failover.md
│
├── docs/adr/ ← Architecture Decision Records (ADR-001–ADR-010)
│
├── src/ ← Node.js CLI + TypeScript runtime
│ ├── cli.ts ← Entry point
│ ├── commands/ ← CLI commands (init, check, run, list, …)
│ ├── agents/ ← BaseAgent + 4 specialized agents
│ ├── llm/ ← ILLMProvider, LLMRouter, 4 providers
│ ├── skills/ ← ISkill, SkillRegistry
│ ├── memory/ ← ShortTerm, LongTerm, Episodic, Project, Residual
│ ├── rag/ ← Embedder, Retriever, Indexer, AgenticRAG
│ ├── reasoning/ ← ReActLoop, ThinkingBudget, ChainOfThought
│ ├── workflows/ ← WorkflowEngine, GoalChain
│ ├── evals/ ← EvalRunner, EpisodeLearner
│ └── observability/ ← Tracer, TokenCounter, EventBus
│
├── .github/workflows/ ← CI enforcement
│ ├── auto-assign.yml ← Assigns owner to all new issues and PRs
│ ├── auditor.yml ← Secret scanning on push
│ ├── commitlint.yml ← Validates commit messages on PRs
│ ├── docs.yml ← Deploys Academy to GitHub Pages
│ ├── labeler.yml ← Labels PRs by changed file paths
│ ├── librarian.yml ← Checks docs stay in sync with code
│ ├── semantic-release.yml ← Automated release + npm publish via OIDC
│ ├── sentinel.yml ← Agent integrity checks
│ ├── tester.yml ← Runs the full test suite
│ ├── vscode-extension.yml ← Builds the VS Code extension
│ └── warden.yml ← File size and code health checks
│
├── .githooks/ ← Local git hook enforcement
│ ├── commit-msg ← Validates commit message format
│ ├── pre-commit ← Blocks main commits, scans secrets
│ ├── pre-push ← Enforces ticket-first branch naming
│ └── prepare-commit-msg ← Injects commit message template
│
└── vscode-extension/ ← VS Code extension
├── src/
│ ├── doormanService.ts ← Real-time SCM input validation
│ ├── auditorService.ts ← On-save secret and dep scanning
│ ├── reviewerService.ts ← [blocking]/[suggestion] diagnostics
│ ├── librarianService.ts ← Stale documentation nudge
│ └── memberWatchProvider.ts ← 14-agent live sidebar TreeView
└── package.json
- Conventional Commits
- addyosmani/agent-skills
- groq-sdk — default free LLM provider
- LanceDB — TypeScript-native vector storage
- semantic-release
- commitlint
Open source. No sign-up. Works with any agent runtime. Membership is free. Standards are not.