Skill Seekers: Turn Any Documentation Into Claude Skills in Minutes
Building AI agents is 20% architecture and 80% getting the right knowledge into the right format at the right time. Skill Seekers is the tool that handles the 80%.
It converts documentation websites, GitHub repositories, PDFs, videos, Jupyter notebooks, wikis, and 10+ other source types into structured knowledge assets — ready to power Claude AI skills, RAG pipelines, or AI coding assistants in minutes.
The Problem It Solves
If you’ve built Claude skills or set up a RAG pipeline, you know the friction: you have a documentation site, a GitHub repo, maybe a PDF spec — and you need to turn all of it into something an AI can reliably retrieve from. Doing that manually means writing scrapers, chunking strategies, deduplication logic, and conflict resolution — before you’ve written a single line of the actual agent.
Skill Seekers is the preprocessing layer that sits between raw documentation and every AI system that consumes it.
What It Does
pip install skill-seekers
# Turn a docs site into a Claude skill
skill-seekers fetch --url https://docs.anthropic.com --output ./claude-skill
# Turn a GitHub repo into cursor rules
skill-seekers fetch --repo owner/repo --format cursorrules
# Turn a PDF into a LangChain knowledge base
skill-seekers fetch --pdf spec.pdf --format langchain10+ source types: documentation websites, GitHub repos, PDFs, YouTube videos, Jupyter notebooks, wikis, OpenAPI specs, and more.
Output formats: Claude AI skills, Cursor .cursorrules, Windsurf rules, Cline rules, LangChain/LlamaIndex/Pinecone-ready chunks.
Automatic conflict detection: if two sources define the same concept differently, Skill Seekers flags it rather than silently merging contradictory information — a detail most similar tools miss entirely.
MCP server built in: expose your knowledge assets as an MCP server directly, so any Claude-compatible client can query them.
The Full Ecosystem
Skill Seekers is structured as a multi-repo project:
| Repo | Purpose |
|---|---|
Skill_Seekers | Core CLI + MCP server |
skillseekersweb | Web UI + 24+ preset configs |
skill-seekers-configs | Community config repository |
skill-seekers-action | GitHub Action for CI/CD |
skill-seekers-plugin | Claude Code plugin |
The preset configs browser is particularly useful — community-contributed configurations for common documentation sources, so you’re not starting from scratch.
Why This Matters for OpenClaw/Agent Builders
If you’re building OpenClaw skills or any Claude-based agent that needs to reason over external documentation, Skill Seekers eliminates the most tedious part of the process. Instead of:
- Write a scraper for the docs site
- Figure out chunking strategy
- Handle deduplication
- Write the skill markdown manually
You do:
skill-seekers fetch --url https://docs.example.com --output ./my-skillAnd get a properly structured skill file ready to drop into your agent.
The conflict detection is the underrated feature. When you’re pulling from multiple sources — a README, an API reference, a changelog — they often contradict each other on edge cases. Most preprocessing tools merge everything and let the LLM sort it out. Skill Seekers surfaces the conflicts explicitly so you can resolve them intentionally.
Install
pip install skill-seekers
# or
brew install skill-seekers # macOS via Homebrew tap