engineering / skill

graphify

Map a codebase, with its docs and other files, into a knowledge graph your coding assistant can query instead of searching files.

Repository owner Graphify-Labs

Not rated yet
0 comments

Like 0 Save

Published 10 Oct 2026Listing updated 12 Oct 2026

At a glance

Purpose: Map a codebase, with its docs and other files, into a knowledge graph your coding assistant can query instead of searching files.
Runs in: Local agent with shell, MCP server, Web service
Operating systems: macOS, Linux, Windows
Needs: Runtime: Python 3.10+; Needs internet: Not stated; Needs accounts or API keys beyond your agent: Not stated
Files and data: Reads files: Yes; Writes files: Yes; Sends data to other services: Not stated
Tool type: Skill
Delivery: From GitHub
Version: Latest on the default branch
Access: Code on GitHub
Compatibility: Claude Code, Cursor, Codex and Gemini CLI, among other assistants, according to the repository. The MCP server works with MCP clients.
External costs: The open-source tool has no charge (Apache-2.0). The semantic pass over docs, PDFs, images and video uses your assistant's model or a configured API key, which can cost money.
Permissions: Reads the files in the folders you point it at and writes its output into your project. Registering the skill changes your assistant's configuration. Optional commands install git hooks and assistant hooks, and graphify can fetch URLs and clone repositories you name.
Usage terms: Apache License 2.0, as stated in the repository. See its LICENSE file for the exact terms.

Code on GitHub

Graphify-Labs/graphify

Apache-2.0

Python125,921 stars12,153 forksLast pushed 12 Oct 2026 UTC

Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing, every edge explained, no vector store.

Topics: ai-agents antigravity ast claude-code code-analysis code-search codex cursor

Public GitHub data fetched 13 Oct 2026 UTC · 0 days old

View on GitHub ↗

What it does

graphify adds a /graphify skill to AI coding assistants. According to its README, it maps a project, including code, docs, PDFs, images and video, into a knowledge graph. You can then ask questions, trace the path between two things or explain one concept, instead of searching files one by one.

The README says code is parsed locally with tree-sitter, without a language model, and that each relationship carries a confidence tag: extracted, inferred or ambiguous. It describes graphify as a graph rather than a vector index, with no embeddings and no vector store.

The graph can also be served as an MCP server for repeated tool calls, locally or over HTTP for a team.

Who it is for

Developers and teams who use AI coding assistants such as Claude Code, Cursor, Codex or Gemini CLI.

Example tasks

Map a repository, then ask how two parts of it are connected.
Read a report of the most-connected concepts and surprising links between modules.
Query the graph from the terminal.
Serve the graph to your assistant as an MCP server.

Getting started

Install the command-line tool, register the skill with your assistant, then run /graphify in your project. According to the README, the package is named graphifyy (double y); the README has the exact commands.

Limitations

Docs, PDFs, images and video need a language model for the semantic pass: your assistant's model, or an API key you configure. The headless graphify extract command has a --code-only option that skips them. Installation needs Python 3.10 or later; the README recommends uv, with pipx or pip as alternatives. Some languages need optional extras; the README lists the supported file types.

Data practices

According to the README, code is processed on your machine and video and audio are transcribed on your machine; docs, PDFs, images and video then go through a semantic pass with your assistant's model or with a provider whose API key is set in your environment. The README says there is no telemetry, and its settings table says a local query log is off unless you switch it on.