Codedb, developed by Justrach, is a free MCP server tool built with Zig, designed to provide code intelligence for AI agents. It functions as a context engine, not an editor, connecting to AI agents/clients via the Model Context Protocol. Its primary role is to assist AI assistants in understanding and navigating codebases, offering structural insights to enhance production AI workflows.

What does Codedb connect to?

Codedb serves as an MCP (Model Context Protocol) server, connecting to various AI agents and clients. It operates as a code intelligence server, providing context rather than editing capabilities. The program supports the MCP protocol, which is JSON-RPC 2.0 over stdio. Compatible AI clients include Claude Code, Codex, Gemini CLI, Cursor, Windsurf, Devin, and opencode, ensuring broad integration for developers.

Core features and tools

Codedb exposes 21 or 22 MCP tools for code intelligence. Key callable tools include context, explain, callpath, list_dir, status, codedb_tree, codedb_context, codedb_symbol, codedb_search, codedb_word, codedb_callers, codedb_outline, codedb_deps, and codedb_read. The installer also provides skills like "/codedb-troubleshooting" and "/codedb-bench". The "codedb_edit" tool has been removed, reinforcing its role as a context provider.

Who is it for?

Codedb targets AI agents and developers who use AI assistants in production AI workflows. It is designed for those needing deep code understanding and efficient context delivery to large language models. This program helps agents navigate code, understand symbols, and perform structural searches. Skip this if your primary need is an in-editor code editing tool rather than a context engine.

How to get started

No credentials are required. The CLI communicates with the Codedb daemon via HTTP requests, and manual configuration involves clients launching "codedb mcp" as a stdio child process.

Use cases

• Code Navigation and Understanding: Agents use tools like codedb_tree for orientation, codedb_context for natural-language tasks, codedb_symbol for exact definitions, and codedb_callers for symbol usages.
• Providing Context to AI Agents: Codedb helps agents understand code via structural search, symbol/caller lookup, and dependency graphs, enabling edits through native client tools.
• Efficient Code Intelligence for LLMs: Providing structural code slices, like function maps, optimizes token usage for Large Language Models, allowing targeted queries.

Codedb offers a specialized solution for integrating code intelligence with AI agents, focusing on context provision rather than direct editing. Its Zig foundation and multiple distribution methods make it accessible for various environments. The program supports a range of AI clients and provides detailed code understanding tools.