Agenthood — Setup Guide¶
Source: fworks-tech/agenthood · GitHub
Category: Multi-Agent Architecture / Development Workflows
npm: agenthood (v3.0.0) · License: MIT
Agenthood packages a complete AI engineering team as 14 individual SKILL.md files. Each agent — from The Reviewer to The Auditor — is a self-contained role with defined responsibilities, communication standards, and quality gates. They work with any agent runtime including Hermes Agent, Claude Code, and Copilot.
The Team¶
| # | Agent | Role | Best For |
|---|---|---|---|
| 1 | The Architect | System design, ADRs | Planning features, technical decisions |
| 2 | The Auditor | Security scanning | Vulnerability assessment, dependency audit |
| 3 | The Debugger | Error triage | Root cause analysis, stack trace investigation |
| 4 | The Doorman | Validation gates | Branch protection, CI health checks |
| 5 | The Envoy | Cross-provider translation | Convention validation across runtimes |
| 6 | The Herald | Release management | Versioning, changelog generation |
| 7 | The Librarian | Documentation | API references, knowledge management |
| 8 | The Oracle | Institutional knowledge | Authoring templates, best practices |
| 9 | The Reviewer | Code review | Standards enforcement, PR quality |
| 10 | The Scribe | Commits & PRs | Commit messages, PR descriptions |
| 11 | The Sentinel | Integrity monitoring | Cross-member contradiction detection |
| 12 | The Steward | Context economy | Provider cache strategies, token optimization |
| 13 | The Tester | Test coverage | TDD, edge case generation |
| 14 | The Warden | Code health | Complexity enforcement, technical debt tracking |
Installation¶
Option A: Via skills.sh (Recommended for Hermes)¶
# Install the full agent team
npx skills add fworks-tech/agenthood
# Or install specific agents
npx skills add fworks-tech/agenthood --skill the-reviewer
npx skills add fworks-tech/agenthood --skill the-auditor
npx skills add fworks-tech/agenthood --skill the-architect
npx skills add fworks-tech/agenthood --skill the-tester
Option B: Via npm (Standalone CLI)¶
npm install --save-dev agenthood
npx agenthood init # Interactive setup (~5 minutes)
npx agenthood check # Verify installation
Option C: Direct GitHub Clone¶
git clone https://github.com/fworks-tech/agenthood.git
cp -r agenthood/skills/* ~/.hermes/profiles/corpusiq/skills/
Usage with Hermes Agent¶
Load Individual Agents as Skills¶
After installing via skills.sh, each agent is available as a skill in your Hermes profile:
# Verify installed skills
hermes skills list --marketplace | grep agenthood
Invoke agents by their role in natural language:
"Run the-reviewer on the current PR diff"
"Have the-auditor scan our authentication flow"
"Ask the-architect to review the proposed database schema"
Compose Agent Teams¶
For complex workflows, load multiple agents and let Hermes route tasks:
"Run the full agenthood team on this PR: the-scribe for commit messages,
the-reviewer for code quality, the-tester for coverage gaps, and
the-warden for complexity checks"
Autonomous Execution¶
Agenthood includes a TypeScript CLI for running agents independently:
# Set LLM provider (Groq free tier, Anthropic, or OpenAI)
export GROQ_API_KEY="gsk_..."
# Or use Ollama for fully offline execution
# List available agents
npx agenthood list
# Run an agent autonomously
npx agenthood run the-scribe "write a commit message for the current diff"
npx agenthood run the-reviewer "review the changes in the last commit"
npx agenthood run the-architect "plan the implementation for issue #42"
Key Workflows for Hermes¶
PR Review Pipeline¶
# 1. Review code quality
hermes --skill the-reviewer "Review the PR at HEAD"
# 2. Check security
hermes --skill the-auditor "Audit dependencies and auth flow"
# 3. Verify tests
hermes --skill the-tester "Check test coverage for changed files"
# 4. Generate changelog
hermes --skill the-scribe "Generate changelog entry for this PR"
Architecture Decision Records¶
hermes --skill the-architect "Create an ADR for our caching strategy:
- Problem: Repeated expensive queries
- Options: Redis vs in-memory vs CDN edge caching
- Constraints: <50ms p95 latency, multi-region"
Security Audit Pipeline¶
hermes --skill the-auditor "Run full security audit:
1. Dependency vulnerability scan
2. Auth flow review
3. Secret detection in codebase
4. API endpoint exposure check"
Provider Configuration¶
| Variable | Provider | Free Tier |
|---|---|---|
GROQ_API_KEY |
Groq (default) | console.groq.com |
ANTHROPIC_API_KEY |
Anthropic | — |
OPENAI_API_KEY |
OpenAI | — |
Or use Ollama for fully offline execution (no API key required):
ollama pull llama3.3
npx agenthood run the-reviewer "review this code" --provider ollama
Why This Matters¶
Agenthood solves a critical gap in the agent ecosystem: standardized, composable agent roles. Instead of every team reinventing code review agents, security auditors, and PR validators, Agenthood provides battle-tested implementations that work across runtimes.
For Hermes specifically: - Drop-in quality gates — the-doorman validates branches, the-reviewer enforces standards - Multi-agent blueprints — reference architecture for Hermes agent teams - Runtime-agnostic — skills work with Hermes, Claude Code, Copilot, and the standalone CLI - Opinionated standards — agents have real standards (they will block merges with bad commit messages)
Verification¶
# Confirm installation
npx skills list | grep agenthood
hermes skills list --marketplace | grep agenthood
# Test an agent
npx agenthood run the-scribe "hello"
# Check skill files are accessible
ls ~/.hermes/profiles/corpusiq/skills/the-reviewer/