Skip to content

Summary

Local code intelligence MCP server that returns symbol context, call graphs and impact analysis in a single call — 15 languages, indexed on your own machine.

Features

  • One MCP call returns symbol context, call graph and impact analysis, pre-correlated
  • Semantic and symbol search across an on-disk index of your repository
  • Call-graph and dependency tracking with exact call sites for callers and callees
  • Recursive impact analysis to estimate the blast radius of a change
  • Runs as a persistent MCP server (stdio, HTTP, HTTPS) or as a one-shot CLI with no daemon
  • --watch keeps the index current as files change
  • Parses 15 languages including Rust, Python, TypeScript, Go, Java, Swift and C++
  • Indexes locally — no source code leaves your machine

Installation

Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.

Any MCP-compatible agent

Most agents (Claude, Cursor, Windsurf, VS Code, and more) read a standard mcpServers configuration.

  1. Open your agent's MCP configuration file.
  2. Merge the snippet below into it.
  3. Restart the agent — the "Codanna" tools will be available.
{
  "mcpServers": {
    "codanna": {
      "command": "codanna",
      "args": [
        "--config",
        ".codanna/settings.toml",
        "serve",
        "--watch"
      ]
    }
  }
}

Description

Codanna is a local-first code intelligence server for AI coding agents. It indexes a repository on disk and exposes symbol search, semantic search, call graphs, dependency tracking, document RAG and impact analysis over the Model Context Protocol, so an agent can ask one question instead of running the usual grep-open-read loop across dozens of files.

The headline tool is semantic_search_with_context, which fuses five separate lookups into one response: for every semantic match it returns the symbol identity, its signature and docstring, the functions it calls with exact call sites, the functions that call it, and the recursive blast radius of changing it. For an agent working in a large codebase, that collapses a long chain of tool calls into a single pre-correlated result.

Codanna runs two ways from the same toolset. As a persistent MCP server (codanna serve, over stdio, HTTP or HTTPS) it suits session-long agent work, and --watch keeps the index hot as files change. As a one-shot CLI (codanna mcp find_symbol name:"…") it needs no daemon, which makes it easy to wrap in slash commands, bash hooks, CI steps or an Agent Skill in any harness that can run a shell command.

Written in Rust and distributed via an install script, Homebrew, PowerShell or Cargo. It parses 15 languages — Rust, Python, JavaScript, TypeScript, Java, Kotlin, Go, PHP, C, C++, C#, Clojure, Lua, Swift and GDScript — and works with Claude Code, Cursor, Windsurf, Codex, Gemini and any MCP-compatible client. Indexing happens entirely on your machine, so no source code is sent to a third-party service. Apache-2.0 licensed and free.

Covered in the Weekly

Related MCP Servers

MCP: Hugging Face

by Hugging Face

New

Hugging Face's official MCP server — search the Hub's models, datasets and Spaces, read repository files, and call thousands of Gradio applications as tools from one remote endpoint.

New

Qase test management over MCP: 36 task-oriented tools for cases, runs, results, defects and QQL search, hosted at mcp.qase.io or self-run with your own API token.

New

Arm's official MCP server for porting and tuning software on Arm — semantic docs search, codebase migration scanning, container architecture checks and LLVM-MCA assembly analysis.

New

Official AntV MCP server that turns a description of your data into a rendered chart — 26 visualization types from bar and line to sankey, mind-map, fishbone and geographic maps.

Browse all MCP servers →