Graphify
Open-source skill that turns a codebase, docs and PDFs into a queryable knowledge graph for 20+ coding agents, with no LLM calls for code.
Overview
Graphify, from Graphify Labs, was among the fastest-growing AI repositories on GitHub in the week to October 29, 2026, gaining about 6,662 stars to reach 137.9k. It parses code locally with tree-sitter across about 40 languages with no LLM and no data leaving the machine, while docs, PDFs, images and video use the assistant's model for a semantic pass. Edges are tagged EXTRACTED or INFERRED, there are no embeddings or vector store, and graphs are queried with graphify query, path and explain, producing graph.html, GRAPH_REPORT.md and graph.json. The README reports 76% on LongMemEval-S QA, tied with dense RAG, and 82.0% key-fact coverage on ERPNext versus 70.8% for a grep baseline. It is dual-licensed Apache-2.0 and MIT; a paid Graphify Cloud also exists.
This is a sourced directory entry. Features and capabilities are reported in the linked sources; this entry does not document independent Aymorphic testing.
Key features
- Local tree-sitter parsing for about 40 languages, zero LLM credits for code
- EXTRACTED versus INFERRED edge tagging; no embeddings or vector store
- Works as a skill for 20+ assistants including Claude Code, Cursor, Codex and Copilot CLI
- README benchmarks: 76% LongMemEval-S QA, 82.0% ERPNext key-fact coverage
- Apache-2.0 and MIT dual license
Use cases
- Giving coding agents a persistent map of a large repository
- Answering cross-file and cross-document questions about legacy systems
Pricing
Open source (Apache-2.0/MIT); a paid Graphify Cloud product exists, pricing unknown.
Pricing reflects the source record and may change. Confirm current terms with the provider.
Dates and sources
- Released
- Not listed
- Record updated
- 29 Oct 2026
- Record last checked
- 29 Oct 2026