Understand your code — don't grep it. We turn any folder of code, docs, PDFs, images, or video into a knowledge graph you query, trace, and prove things about instead of searching by hand.

Website · graphify (open source) · PyPI · Discord · LinkedIn

We ship two things, one idea: a graph of how your project actually fits together. The open-source tool builds that map on your machine for free. The enterprise platform builds a smarter, verified map — one that proves your changes are safe, remembers, learns, and runs entirely inside your own infrastructure.

Type /graphify in your AI coding assistant and it maps an entire project into a persistent knowledge graph. Code is parsed locally with tree-sitter (deterministic, no LLM, nothing leaves your machine); docs, PDFs, images, and video get a semantic pass through your assistant's model or a configured key. You get three files: an interactive graph.html, a plain-language GRAPH_REPORT.md, and a GraphRAG-ready graph.json.

• A real graph, not a vector index. No embeddings, no vector store. You traverse it — ask a question, trace the shortest path between two things, or explain one concept and its neighbours.
• Code maps for free. tree-sitter AST across ~40 languages resolves calls / imports / inherits edges with no API cost and no code leaving your machine.
• An honest audit trail. Every edge is tagged EXTRACTED (explicit in the source) or INFERRED (resolved by graphify) — you always know what was read vs derived.
• Structure you wouldn't think to ask for. God nodes surface what everything flows through; Leiden community detection groups the graph into labelled subsystems.

uv tool install graphifyy && graphify install     # register the skill with your assistant
/graphify .                                        # Claude Code, Cursor, Codex, Copilot, Gemini CLI…

Docs: how it works — the extraction pipeline, community detection, and confidence scoring.

The open-source tool tells you how your code fits together. Enterprise tells you whether a change is safe — without your code ever leaving your building. It's the same graph, made trustworthy for teams shipping AI-written code.

• Verified, not just reviewed. Where a reviewer guesses a change looks fine, Graphify runs formal verification: it proves an edit behaves identically, or hands you the exact input that breaks it — as a runnable test. When it genuinely can't be sure, it says so. No false confidence.
• Runs in your VPC — code never leaves. On-prem by default: CPU-local graph, retrieval, memory, and verification. No source or chat logs in someone else's cloud. The privacy claims are true by construction, not by policy.
• A grounded memory + fact-checker for your AI agents. Over MCP, Cursor / Claude Code / Codex get provenance-carrying, honest-abstain answers — and can verify their own changes before you ever see the PR.
• PR review + merge gate for teams. Every pull request gets a structural review and a codebase-aware verification, posted as a check — with an enforceable gate.
• It learns from you. When a trusted maintainer says a finding is wrong, it remembers and suppresses it going forward — and always cites why. Preferences are advisory; correctness always wins.
• Enterprise-ready. SSO, project scoping, a tamper-evident audit trail, and secret redaction before any model call.

The one-liner: every other AI code reviewer guesses. Ours proves — and it runs in your house.

Start free on the open-source tool. Upgrade when you need to prove your code is right, share across a team, and keep everything inside your own infrastructure.

Talk to us: founders@graphifylabs.ai · graphify.com · Discord

We think you're gonna like it here.