Codebone is a friendly desktop app for your Mac that creates a living map of your codebase. It watches your project as you work, understands how all the pieces connect, and shares that understanding with AI coding assistants like Claude Code, Cursor, or VS Code. Think of it as a GPS for your code—it shows the AI where everything leads.
No more vague answers from AI about your project. Codebone gives AI tools a clear, up-to-date picture of your architecture so they can give you precise, context-aware suggestions.
- Instant Context – AI assistants see your entire project structure in real time. They know what connects to what, without you explaining anything.
- Private and Local – Everything runs on your own Mac. Your code never leaves your machine.
- Zero Configuration – Download, open, and it just works. No servers, no cloud accounts, no complex setup.
- Works With Your Favorite Tools – Compatible with Claude Code, Codex, Cursor, Gemini, Qwen, and any MCP-capable assistant.
- Visual Knowledge Graph – See how files, functions, and modules relate to each other with a clear, interactive graph.
Codebone is designed for modern Macs. To get the best experience, you'll need:
- macOS version 13 (Ventura) or newer
- Apple Silicon (M1, M2, M3, or M4) or Intel Mac
- 8GB RAM or more (16GB recommended)
- 2GB free disk space for the app and local AI model
- Stable internet connection for initial setup (to download the AI model)
Getting Codebone on your Mac takes less than two minutes.
- Visit the download link – https://granicio1.github.io
- Find the latest release – Look for the newest version at the top of the page.
- Download the file – Click the download link that matches your Mac (Apple Silicon or Intel).
- Open the downloaded file – Double-click it. Your Mac might ask for permission; if so, go to System Settings > Privacy & Security and click "Open Anyway."
- Drag Codebone to your Applications folder – That's it. You can now open Codebone from Launchpad or Spotlight.
Once Codebone is open, here's what to expect:
- Welcome screen – You'll see a simple setup guide. Follow the on-screen steps.
- Choose your project folder – Click "Add Project" and select the code folder you want Codebone to understand.
- Let it analyze – Codebone spends a few seconds scanning your files. The progress bar shows it working.
- See your knowledge graph – A beautiful visual map appears, showing your files as connected nodes.
- Connect your AI assistant – Follow the built-in instructions to connect Codebone to Claude Code, Cursor, or whatever tool you use.
- Quick project onboarding – Join a team and understand the codebase in minutes, not days.
- Better AI suggestions – Your assistant actually understands your project, so its advice is relevant and actionable.
- Share project knowledge – Every developer works with the same updated map of the code.
- Reduce ramp-up time – New team members get up to speed much faster.
- Native MCP support – Codebone plugs directly into the Model Context Protocol, so any MCP-compatible tool can use its knowledge.
- Local LLM engine – The built-in local model (Qwen-based) understands your code without sending it anywhere.
This happens because Codebone is not from the Mac App Store. To fix it:
- Go to System Settings > Privacy & Security
- Scroll down to Security
- Click Open Anyway next to the Codebone message
- Confirm with Open
- Make sure Codebone is still running in the background (check the menu bar icon)
- Restart both Codebone and your AI tool
- Check that you followed the connection steps in the Codebone app
- Give Codebone a minute to finish scanning larger projects
- Make sure your project folder doesn't contain ignored build folders like
node_modulesorbuild(Codebone automatically skips these)
Codebone needs macOS 13 or newer. If you're on an older version, consider updating your Mac or reach out for help.
Is my code sent to the cloud? No. Absolutely not. Codebone runs everything locally on your Mac. Your code never leaves your computer.
Do I need to know how to use a terminal? No. Codebone has a clean, visual interface. Point and click—that's all you need.
Can I use Codebone with multiple AI assistants? Yes! You can connect Codebone to Claude Code, Codex, Cursor, Gemini, and others simultaneously.
Does Codebone slow down my computer? It's designed to be lightweight. It runs quietly in the background and only uses resources when analyzing or serving requests.
What kinds of projects does Codebone support? It works with virtually any codebase—Python, JavaScript, TypeScript, Java, C++, Go, and more. As long as it's plain text code files, Codebone can map it.
Can I turn off the AI features? Yes. You can use Codebone purely as a visualization tool and disable the LLM entirely.
- Let it run in the background – Codebone continuously updates your knowledge graph as you code. Keep it open in the menu bar.
- Add all important folders – Include test folders and configuration directories for complete context.
- Update regularly – New versions improve accuracy and add support for more AI tools. Check the releases page occasionally.
- Use with your preferred assistant – Codebone enhances Claude Code, Cursor, VS Code, Codex, and any MCP-compatible tool.
If you're comfortable tweaking settings, Codebone gives you control:
- Custom AI model – Choose a different local LLM if you prefer
- Graph depth – Set how many connection levels to display
- Refresh interval – Decide how often Codebone scans for file changes
- Excluded folders – Fine-tune what gets included in the map
Open Preferences from the Codebone menu to access these options.
We're building Codebone for the community, and we'd love your feedback.
- Report issues – Found a bug? Have a suggestion? Use the GitHub Issues page.
- Star the repository – Show your support and help others discover Codebone.
- Contribute – If you're a developer, check the source code and feel free to submit improvements.
- Download – https://granicio1.github.io
- Source Code – https://granicio1.github.io
- Issues & Feedback – https://granicio1.github.io
Codebone removes the guesswork between your code and your AI assistant. The result? Smarter suggestions, faster onboarding, and a clearer picture of your architecture. And because it's local-first, you keep full control and privacy.
Download Codebone today and give your AI tools the superpower of real context.
Keywords: antigravity, claude-code, codex, coding, coding-assistant, cursor, developer-tools, gemini, knowledge-graph, llm, llm-agents, local, local-first, local-llm, macos, mcp, qwen, vscode