Skill Seeker

Overview

Skill Seeker is an end-to-end toolchain to transform documentation (websites, repos, PDFs) into production-ready AI "skills". Instead of manually reading and summarizing docs, Skill Seeker scrapes, analyzes, enhances, and packages content into formats ready to upload or install into AI agents. It targets workflows where you want a model (e.g., Claude or other LLMs) to have a structured, example-rich knowledge bundle derived from source docs and code.

Key Capabilities

• Documentation scraping: universal scraper with llms.txt support, smart page selection and categorization.
• GitHub analysis: deep AST parsing for multiple languages (Python, JS/TS, Java, C++, Go), API extraction (functions/classes/signatures), issues & changelog extraction.
• Conflict detection: automated comparison between documented APIs and actual code implementation (signature mismatches, missing-in-code, missing-in-docs, missing-in-docs, description differences) with side-by-side reports.
• Multi-source merging: combine docs + repo + PDFs into a single unified skill, with rule-based or AI merging strategies and transparent warnings.
• AI enhancement: generate comprehensive SKILL.md with high-quality examples, troubleshooting, and guides using LLMs (supports API-based and local enhancement modes).
• Multi-LLM packaging: export optimized packages for Claude (ZIP + YAML), Google Gemini (tar.gz), OpenAI ChatGPT (ZIP + vector store) or generic Markdown.
• MCP integration: provides MCP server/client tools so you can call Skill Seeker directly from AI editors/agents (Claude Code, Cursor, Windsurf, VS Code/Cline, IntelliJ) using a set of MCP tools.
• Rate-limit & resume management: multi-token profiles, intelligent rate-limit strategies, job checkpointing and resume capability for long scrapes.

Typical Workflow

1. Configure or choose a preset (React, Godot, Django, etc.).
2. Run scraper to fetch docs/repo/PDF and build a structured reference set.
3. (Optional) Run AI enhancement to generate high-quality SKILL.md and examples.
4. Package the skill for the target LLM platform, then upload or install into your agent.

Example commands:

# Install from PyPI
pip install skill-seekers
 
# Scrape docs or repo
skill-seekers scrape --config configs/react.json
skill-seekers github --repo facebook/react
 
# Enhance and package
skill-seekers enhance output/react/
skill-seekers package output/react/ --target claude

Where it fits

Skill Seeker is most useful for developer teams, documentation maintainers, and learners who want a polished, queryable knowledge package derived directly from authoritative sources. It's also suited for organizations that need to convert internal docs and repositories into agent-installable skills while detecting and reporting documentation drift.

Integration & Extensibility

• MCP tools (18+) let AI agents list configs, generate/validate configs, scrape, package, upload and more.
• Plugin-like optional dependencies: choose only Gemini/OpenAI/Claude integrations as needed.
• Config-driven scrapers allow customizing selectors, categories, rate limits and split strategies for very large documentation sets.

Limitations & Notes

• Although optimized for Claude skills, it supports multiple target formats; uploading automatically to a hosted LLM may require API keys for that provider.
• AI enhancement can be run via API (requires provider key) or locally via supported local agent runtimes when available.

Repository & Contact

Source: GitHub repository at the provided URL. Primary author/maintainer: the GitHub owner (yusufkaraaslan).