Skip to content

indhra/agent-hunter

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

18 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

agent-hunter

Skill package manager for Claude Code. Discovers relevant skills and MCP servers for your projects.

Version License Claude Code Privacy


What It Does

Given your repository, agent-hunter identifies the top 3 relevant skills or MCPs, explains why they fit, and identifies unsafe ones to avoid.

$ agent-hunter hunt .

Found: FastAPI, PostgreSQL, pytest, Docker

Top 3 recommendations:

1. fastapi-backend-testing
 Status: SAFE
 Reason: Your repo uses FastAPI, pytest, and Docker. This skill provides
 test and deployment workflow helpers.

2. postgres-mcp-server
 Status: SAFE
 Reason: PostgreSQL detected. Provides database query and schema tools.

3. api-security-scanner
 Status: REVIEW
 Reason: REST API security audits. Flagged for filesystem access.
 Review SKILL.md before installing.

Blocked (security): 2 results

Installation

Requirements: Claude Code, Git, Python 3.10+

git clone --depth 1 https://github.com/indhra/agent-hunter.git ~/.claude/skills/agent-hunter
cd ~/.claude/skills/agent-hunter
./setup

The setup script:

  • Verifies Python 3.10+
  • Creates isolated virtual environment
  • Registers agent-hunter in global ~/.claude/CLAUDE.md
  • Makes /agent-hunter available in all Claude Code sessions

Usage

Basic Hunt

/agent-hunter

Or:

agent-hunter hunt /path/to/project

Proactive Mode (Optional)

Enable automatic hunting when opening new projects:

export AGENT_HUNTER_AUTO=1

Add to ~/.zshrc or ~/.bash_profile for permanence.

See INSTALL.md for full proactive setup options.


First Run

Open any repo and type:

/agent-hunter

Or ask Claude:

What skills or tools should I install for this project?

agent-hunter will:

  1. Read your project's tech stack (framework names only - no code, no paths)
  2. Hunt curated skills + GitHub results
  3. Security-scan every result
  4. Show top 3 recommendations with clear explanations
  5. Guide you through install if useful

Discovery Tiers

Tier 1: Curated Index (Always Active)

Source: ~100 verified skills (offline) Speed: Instant Trust: Highest (human-reviewed) Setup: None required

Manually vetted skills. Cryptographically signed.

Tier 2: GitHub API Search (Optional)

Source: GitHub Code Search - 5,000+ repositories with SKILL.md Speed: 2-5 seconds Trust: Medium (automated filtering) Setup: Requires GITHUB_TOKEN

export GITHUB_TOKEN=your_token

Generate at https://github.com/settings/tokens (no scopes required).

Tier 3: Web Search (On-Demand)

Source: Web search across GitHub, marketplaces, documentation Speed: 5-15 seconds Trust: Variable (security-scanned before showing) Activation: User request or auto-trigger when < 3 results


Commands

# Main command - hunt for skills/MCPs relevant to your project
agent-hunter hunt .

# Health check - audit all installed skills
agent-hunter audit

# Restore registry to last known good state
agent-hunter rollback

Run from Claude Code: /agent-hunter



Update / Uninstall

Update:

cd ~/.claude/skills/agent-hunter && git pull && ./setup

Uninstall:

rm -rf ~/.claude/skills/agent-hunter

Then remove the ## agent-hunter section from ~/.claude/CLAUDE.md.


Privacy

Extracts tech keywords only (framework names, library names). No code, paths, or project-specific strings leave your machine.

See CONTRIBUTING.md for details.


Contributing

See CONTRIBUTING.md for details.


License

MIT © 2026 Indhra Kiranu N A

About

GitHub skill discovery for Claude Code — scans repos for SKILL.md and MCP servers, security-scans every result, and surfaces ranked recommendations

Topics

Resources

License

Contributing

Security policy

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors