# Graphify
Graphify turns any folder of code, docs, papers, images, or videos into a **queryable knowledge graph** that AI coding assistants consume as a skill. Instead of grepping files, you ask natural-language questions about the whole codebase. Installed as the `graphifyy` Python package (`uv tool install graphifyy` / `pipx install graphifyy`), then registered with your assistant.
## Key points
- **Outputs** — `graph.html` (interactive visualization), `GRAPH_REPORT.md` (god nodes, surprising connections, suggested questions), `graph.json` (programmatic querying).
- **Extraction** — code parsed locally via tree-sitter (28 grammars, no API calls); docs/PDFs/images/video sent to your chosen LLM for semantic extraction. Relationships tagged `EXTRACTED` / `INFERRED` / `AMBIGUOUS` for transparency.
- **Querying** — `graphify query/path/explain`, community-detection clustering, and an [[Model Context Protocol (MCP)]] server (`query_graph`, `get_node`, `shortest_path`).
- **Integrations** — [[Claude Code]], [[Cursor.com|Cursor]], [[GitHub Copilot]] CLI, [[Gemini CLI]], Codex, Aider, 15+ assistants; LLM backends Claude / Gemini / OpenAI / DeepSeek / Ollama / Bedrock / Azure.
- **Privacy & team** — code stays local, no telemetry; `graphify-out/` commits to git, auto-merges, post-commit incremental rebuilds; PR intelligence (`graphify prs --triage`).
- **License** — MIT; Python 3.10+.
## Why it matters
Graphify is a structured **context layer** between raw code search and RAG ([[Retrieval-Augmented Generation (RAG)]]): the graph lets an agent grasp architecture without re-reading the whole repo each turn. Notably for PKM, it can **generate [[Obsidian]] vaults and Markdown wikis** from the graph — making it a bridge between codebase understanding and a personal knowledge base, and a skill in the same ecosystem as [[Google AI Skills]] / [[NVIDIA SkillSpector]].
## References
- Website: https://graphify.net
- Repository: https://github.com/safishamsi/graphify
- PyPI: https://pypi.org/project/graphifyy/
## Related
- [[Atomic]]
- [[Model Context Protocol (MCP)]]
- [[Obsidian]]
- [[Retrieval-Augmented Generation (RAG)]]
- [[Claude Code]]
- [[AI Agent Skills]]