bartolli/codanna

Local code intelligence MCP server and CLI for AI coding agents

What it solves

Codanna eliminates the inefficient "grep-and-read" loop that AI coding agents typically use to understand a codebase. Instead of forcing an agent to manually search for strings and open files one by one, it provides a structured, semantic understanding of the code, allowing agents to instantly retrieve symbol context, call graphs, and impact analysis in a single request.

How it works

Codanna indexes a local repository on disk and creates a semantic map of the code. It operates as a Model Context Protocol (MCP) server or a one-shot CLI tool. It uses a local embedding model to enable natural language search and parses code across 15 different languages to track relationships between symbols, callers, and callees. This allows it to perform semantic search, dependency tracking, and document RAG (Retrieval-Augmented Generation) without sending source code to a remote API.

Who it’s for

It is designed for developers using AI coding agents (such as Claude Code, Cursor, Windsurf, Codex, or Gemini) who want to give their assistants "x-ray vision" into their local projects for faster and more accurate code navigation and modification.

Highlights

  • Local-First Architecture: All indexing and embedding generation happen on the user's machine; no source code leaves the device by default.
  • Unified Context Retrieval: A single command can return a symbol's identity, signature, documentation, callers, callees, and the potential "blast radius" of changes.
  • Multi-Language Support: Supports 15 languages including Rust, Python, JavaScript, TypeScript, Java, Go, C++, and C#.
  • Flexible Integration: Works as a persistent MCP server for session-long work or a CLI for scripts and CI/CD pipelines.
  • High Performance: Features fast parser throughput and sub-10ms lookups for warm-server queries.

Related

  • Project
    colbymchenry/codegraph

    A semantic code intelligence platform that provides AI agents with a pre-built knowledge graph of a codebase to eliminate slow file-by-file discovery and reduce token costs.

  • Project
    aovestdipaperino/tokensave

    TokenSave is a local MCP server that builds a semantic knowledge graph of a codebase, letting AI coding agents query symbols, callers, and impact instead of scanning files with grep and read, cutting token usage and tool calls.

  • Project
    forloopcodes/contextplus

    An MCP server that turns large codebases into searchable, hierarchical feature graphs using RAG, AST parsing, and spectral clustering for high-accuracy AI coding context.

  • Project
    johnhuang316/code-index-mcp

    An MCP server that provides intelligent code indexing and AST-based analysis to help AI assistants navigate and understand complex codebases.

  • Project
    hitmux/hitmux-context-engine

    A semantic context search engine for MCP clients that indexes local directories into vector storage to help AI agents find relevant code and documentation.