Graphify

Graphify-Labs/graphifyReviewed by AI

What it does

Graphify reads a project folder and builds a graph of how its parts connect: which function calls which, what a file imports, which document explains what. You get an HTML page you can click through, a Markdown report and a graph.json you can ask questions about from the terminal or from your AI coding assistant with /graphify. Code is parsed on your machine with tree-sitter and needs no AI key; documents, PDFs, images and video only use a model if you set one up. For programmers, with or without an assistant such as Claude Code.

How easy it is to install

C · install 1 or 2 tools · Windows

Windows C
macOS C
Linux C

Same work on all three platforms, and macOS is the shortest path (`brew install [email protected] uv`, then `uv tool install graphifyy` and `graphify install`); Windows needs `winget install astral-sh.uv` first. There is no installer: you need Python 3.10 or newer plus uv or pipx, and the only required program is Python. An AI assistant is the recommended way in (`/graphify .`) but not a requirement: the README says "Beyond your AI assistant, query `graph.json` straight from the CLI with `graphify query`, `graphify path`, and `graphify explain`" and "A code-only corpus requires no API key: `graphify extract` runs fully offline". Docker is optional, for the one use the repository's Dockerfile covers: "graphify MCP server as a shared HTTP service".

Numbers on GitHub

Stars: 140,040on Oct 31, 2026
Downloads: 11,9321,033 this week
Growth: +3,878stars in 5 days

Stars over time

Facts

Runs on: Windows, macOS, Linux
You need to install: Python
To build from source: Python, Docker
Disk space: 100 MB–300 MB · estimate
Accounts or keys: OpenAI, Anthropic, Google Gemini, DeepSeek, Moonshot AI (Kimi), Graphify Cloud (the paid service)
Cost: Freemium · The command-line tool is free under Apache-2.0. The same README sells Graphify Cloud, a paid always-on service with a "free 14-day trial". Running the semantic pass over documents without an assistant needs a paid AI key.
License: Apache-2.0 (Permissive)
Can replace: Graphify Cloud, Mem0, Supermemory
Last update: Nov 1, 2026

Risk signals (8)

Attentionunusual star rate
140041 stars in 211 days (~664 per day)
automatic check
AttentionInstall command runs a script from the internet
Install command pipes a remote script into a shell: curl -LsSf https://astral.sh/uv/install.sh \| sh
automatic check
Attentionsimilar package names on pypi
README, Install: "Official package: The PyPI package is `graphifyy` (double-y). Other `graphify*` packages on PyPI are not affiliated. The CLI command is still `graphify`." Typing the obvious name installs someone else's package.
AI review
Infoqueries written to a log file
README, Privacy: "every `graphify query`, `graphify path`, `graphify explain`, and MCP `query_graph` call is logged to `~/.cache/graphify-queries.log` in JSON Lines format (timestamp, question, corpus, nodes returned, duration)". It stays on your machine and `GRAPHIFY_QUERY_LOG_DISABLE=1` turns it off. The same section says there is no telemetry.
AI review
Infochanges your assistant and your repository
`graphify install` writes instruction files and hooks for your assistant (for example `.claude/skills/graphify/SKILL.md`, `AGENTS.md`, `.cursor/rules/`, a `PreToolUse` hook, a native Pi extension), and `graphify hook install` adds git hooks that rebuild the graph on every commit and branch checkout. `graphify uninstall --purge` removes them.
AI review
Infofree tool promotes a paid cloud service
The README repeats the Graphify Cloud trial pitch in four places, including the claim "Cut agent token spend by 80%" and a Y Combinator S26 badge. The local tool works without it, but the page is also a sales page for the company's paid product.
AI review
Infobenchmarks are self reported
The README's comparison table ("0.497 recall@10" against mem0 and supermemory) was produced by the project's own harness and judge, as described in BENCHMARKS.md. No independent party ran it.
AI review
Infooptional extra downloads videos from websites
The `video` extra installs yt-dlp, and `/graphify add <youtube-url>` fetches and transcribes a video. Downloading videos this way is against the terms of service of YouTube and similar sites, and that extra is not installed by default.
AI review

Evidence is shown in its original language.

What Radar could not check

Whether the graph is accurate for your project: the README itself warns that a file parsed as the wrong language still produces nodes, so the graph can look full while the real symbols are missing.
How long a build takes and how much disk the graph uses on a large repository.
What the optional semantic pass over documents and PDFs costs with a paid API key.
Whether the hooks it installs really make your assistant read fewer files, or just add instructions it may ignore.
How 140,041 stars accumulated in 211 days, which is about 664 a day.