Skill Seekers

Skill Seekers turns documentation sites, GitHub repositories, PDFs, videos, notebooks, wikis and other material into structured knowledge assets for AI systems. The project calls itself a data layer for AI and handles eighteen source types, preparing content once so it can be exported to twenty-two targets, such as skills for Claude, Gemini and OpenAI, RAG pipelines on LangChain, LlamaIndex and Pinecone, and AI coding assistants like Cursor, Windsurf and Cline.

A quickstart builds a packaged skill archive from a source such as Django documentation. An AI-driven project scan can read a project's manifests, README, Dockerfile, CI files and sampled imports, then emit a configuration for each detected framework plus a codebase description, generating a fresh config when no preset exists and optionally publishing it to a community registry. Topics mention conflict detection, OCR, an AST parser and an MCP server.

It is written in Python and MIT licensed, with a website, sponsors and translations of its README into many languages. It runs locally on your machine. It suits developers who want AI assistants grounded in up-to-date documentation and codebases.