Repo Researcher
Find and research GitHub repositories that expand AI assistant capabilities.
When to Use
- •User wants to discover new tools/MCP servers/CLIs
- •User asks "what's trending" or "find me repos"
- •User provides a specific repo to research
- •Looking for integrations with specific services (Twilio, Linear, Slack, etc.)
Workflow
If User Provided a Repo (owner/repo format):
- •Use deepwiki skill to research it:
bash
node /Users/kokayi/.openclaw/workspace/skills/deepwiki/scripts/deepwiki.js ask <owner/repo> "what does this project do, what are its key features, and how might it be useful for expanding an ai assistant's capabilities?"
- •Format output as:
- •What it is — plain english summary
- •Key capabilities — tools/features exposed
- •Integration potential — how it plugs into LibreChat/OpenClaw/opencode
- •Quickstart — how to actually use it
- •Link — https://github.com/<owner/repo>
If No Repo Provided (Discovery Mode):
- •Search for trending repos:
bash
python3 /Users/kokayi/.openclaw/workspace/skills/repo-researcher/scripts/find_trending.py
- •
Filter results by relevance to user's stack:
- •MCP servers (high priority — expands tool ecosystem)
- •CLI tools that could become OpenClaw skills
- •Browser automation tools
- •Integrations with services they use (Twilio, Linear, Slack, etc.)
- •AI/LLM utilities
- •
Check what skills already exist to avoid duplicates:
bash
ls /Users/kokayi/.openclaw/workspace/skills/
- •
Present 3-4 suggestions with:
- •Repo name (linked: https://github.com/owner/repo)
- •Why it matters for their setup
- •Brief description
- •
Ask which to deep-dive, then use deepwiki skill.
User Context (for relevance filtering)
- •Uses LibreChat as frontend with multiple endpoints
- •Has opencode for coding tasks
- •Runs OpenClaw locally with skills (check which exist)
- •Main project: Consuelo (Twilio-based AI sales coaching)
- •Always wants to expand AI capabilities
Output Format
Always include full GitHub links in suggestions:
- •
[owner/repo](https://github.com/owner/repo)
Discovery mode table columns:
| repo | why it matters |
|---|---|
| owner/repo | description |
Examples
Good suggestion:
| browser-use/browser-use | AI browser automation — lets me actually interact with sites (click, type) vs just reading them |
Bad suggestion (missing link):
browser-use/browser-use — AI browser automation tool
Discovery Priority
- •MCP servers (modelcontextprotocol/*, awesome-mcp-servers list)
- •CLI tools that wrap APIs (twilio-cli, linear-cli, etc.)
- •Browser automation (playwright-mcp, browser-use, etc.)
- •AI agent frameworks
- •Productivity integrations (notion, slack, github mcp servers)