arbor

<p align="center"> <img src="https://raw.githubusercontent.com/Anandb71/arbor/main/docs/assets/arbor-logo.svg" alt="Arbor" width="60" height="60" /> </p> <h1 align="center">arbor-mcp</h1> <p align="center"> <strong>Model Context Protocol server for Arbor</strong><br> <em>Let Claude walk your code graph</em> </p> <p align="center"> <a href="https://crates.io/crates/arbor-mcp"><img src="https://img.shields.io/crates/v/arbor-mcp?style=flat-square&color=blue" alt="Crates.io" /></a> <a href="https://registry.modelcontextprotocol.io"><img src="https://img.shields.io/badge/MCP-registered-purple?style=flat-square" alt="MCP" /></a> <img src="https://img.shields.io/badge/license-MIT-green?style=flat-square" alt="License" /> </p> --- ## Overview `arbor-mcp` is the **AI Bridge** for [Arbor](https://github.com/Anandb71/arbor). It implements the [Model Context Protocol](https://modelcontextprotocol.io/) to let LLMs like Claude Desktop navigate your codebase as a graph. ## MCP Tools | Tool | Description | |------|-------------| | `get_context` | Retrieve semantic neighborhood of a node | | `find_path` | A* shortest path between two nodes | | `analyze_impact` | Predict blast radius of changes | | `list_symbols` | Fuzzy search across the graph | ## Why MCP? Instead of RAG-style "find similar text," Arbor lets the AI: - **Walk the call graph** to understand control flow - **Trace imports** to find the real source of a symbol - **Predict impact** before making changes ## Usage ```bash cargo install arbor-graph-cli arbor bridge # Starts MCP server over stdio ``` ### Claude Desktop Config ```json { "mcpServers": { "arbor": { "command": "arbor", "args": ["bridge"] } } } ``` ## Links - **Main Repository**: [github.com/Anandb71/arbor](https://github.com/Anandb71/arbor) - **MCP Registry**: `io.github.Anandb71/arbor`