Turn your codebase into an AI-ready knowledge graph
CodeGenome maps your repository's internal architecture into a live dependency network. It feeds your AI tools (Cursor, Copilot, or Claude Code) instant, flawless context on how your files interact—slashing AI response latency and cutting your LLM token costs by up to 70%
pip install codegenome
codegenome analyze . && codegenome tui
-
Launch TUI
-
Set Workspace
-
Analyze / Check Workspace
-
Analyze (Ends)
-
Export (Ends)
-
AI Rules Generation
-
Live Graph
-
Auto Rebuild
-
LAN / Localhost
-
Live Panel, AI Chat
-
-
-
-
MCP Integration
-
HTTP Transport
-
LAN or Local
-
AI Agents Comm.
-
-
-
-
-
-
Build an AST
Uses localized tree-sitter loops to convert source code files into Abstract Syntax Trees — fast, language-aware, and zero runtime overhead.
Graph Dependencies
Links imports, function calls, and class relationships into an on-disk SQLite knowledge graph that lives in your .genome/ folder.
Serve to AI Agents
Acts as an MCP Server to stream ultra-low-token context windows straight into Cursor, Claude, or any compatible coding agent.
Nodes are Neurons
Your functions and classes are processing units. Import statements and function calls act as digital synapses linking them together.
Graph Decomposition
Splits computation into cellular Micro-Graphs and abstract Macro-Graphs, mapping the modular hierarchy without Out-of-Memory crashes.
Terminal UI (TUI)
A powerful, interactive terminal interface for seamless monitoring and navigation of your codebase architecture right from your command line.
LAN Live Graph Support
Respects your .gitignore automatically. Broadcast live graph updates to your LAN instantly for collaborative debugging.
Integrated AI Chat
Live Graph AI Chat lets you query your architecture directly via Ollama Local, Ollama Cloud, GROQ, OpenAI, Gemini, and more.
Agentic MRI Scan
Audit the autonomous AI agent storm and deeply inspect your codebase to uncover hidden complexities automatically.
Data-Driven Refactoring
Don't guess. Highlight your "God Nodes" (hyper-connected components) using empirical Degree-Based Analysis and Betweenness Centrality.
Model Context Protocol (MCP)
Exposes CodeGenome's live architecture graph, dependencies, and symbols as native tools to any MCP-compatible AI agent for deeper context.
Warp Speed Onboarding
Ditch the unindexed documentation. Interactively zoom into modules, trace critical execution pathways, and understand the macro-system in minutes.
Connect CodeGenome with your favorite AI tools and coding agents.
LLM Providers for Chat
Supported Coding Agents (MCP)
Join the community and help build the future of codebase mapping.
-
GitHub Stars
-
Forks
-
PyPI Downloads