🔥 Hot Repo: 17 Sources to 12 AI Platforms, Zero Boilerplate

The One Command That Builds a Claude Skill From Anything Getting documentation into a Claude AI skill normally means hours of copying, reformatting, and chunking text by hand. Skill Seekers eliminates that entirely. Three shell commands take any URL, GitHub repository, PDF, video, or Jupyter notebook and output a deployment-ready Claude skill ZIP file. pip install skill-seekers skill-seekers create https://docs.django.com/ skill-seekers package output/django --target claude The result is output/django-claude.zip, ready to drop into Claude Code or the Claude API. The Python project has reached 12,262 stars and 1,232 forks since its October 2025 launch. Its latest stable release is v3.4.0 March 25, 2026. 17 Source Types, 12 Export Targets v3.4.0 expanded the platform from 5 to 12 LLM export targets, adding OpenCode, Kimi, and several more alongside the original Claude, Gemini, OpenAI, LangChain, LlamaIndex, Haystack, Pinecone, ChromaDB, FAISS, and Qdrant outputs. The same input data pipelines now cover 17 source types: - Documentation websites - GitHub repositories - PDFs, Word docs, EPUB e-books, PowerPoint presentations - Jupyter notebooks and OpenAPI specs - YouTube and local videos with transcript and frame extraction - Confluence wikis and Notion pages - RSS/Atom feeds and man pages Each source type plugs into the same create command with auto-detection. One asset can be exported to all 12 platforms without re-scraping the source. Single-Run Multi-Source Builds A single config file can combine sources: bash skill-seekers create facebook/react https://react.dev/ ./my-components/ notes.pdf skill-seekers package output/react --target claude --target cursor --target langchain Conflict detection flags overlapping content between sources before the build runs, preventing duplicate chunks in the output. The Claude Code Plugin A companion repository, skill-seekers-plugin, exposes 35 MCP tools directly inside Claude Code sessions. This means Claude ca