explaingit

killop/codedb-mcp

24RustAudience · developerComplexity · 3/5Setup · moderate

TLDR

A fast Rust server that indexes your codebase and lets AI coding assistants search it by exact text or by meaning in milliseconds, connecting via the MCP protocol to tools like Claude or Cursor.

Mindmap

mindmap
  root((codedb-mcp))
    What it does
      Index codebase
      Exact text search
      Semantic search
      Dependency graph
    Performance
      6ms exact search
      100ms semantic search
      45s initial index build
    MCP integration
      AI assistant tool calls
      Chain operations
      Reduce round trips
    Query types
      Symbol lookup
      File dependencies
      Import graph export
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Things people build with this

USE CASE 1

Give your AI coding assistant fast semantic and exact-text search over a large codebase by connecting this MCP server as a tool.

USE CASE 2

Query function definitions, file dependencies, and import graphs across a codebase with sub-100ms response times.

USE CASE 3

Export the full dependency graph of a codebase for external analysis or visualization.

Tech stack

RustMCP

Getting it running

Difficulty · moderate Time to first run · 30min

Requires the Rust toolchain to build the server and an MCP-compatible AI coding assistant to connect to.

No license information is mentioned in the explanation.

In plain English

This repository is a Rust-based server that indexes a software codebase and then answers questions about it very quickly. It is designed to plug into AI coding assistants through a protocol called MCP, which lets AI tools call external services as tools during a conversation. The server scans source code files, parses them to understand the structure of functions and classes, builds a graph of how files depend on each other, and creates two types of search indexes. One index handles exact text and regular expression searches. The other handles semantic or meaning-based searches, where you describe what you are looking for in plain terms rather than exact strings. Both indexes work together to return relevant code chunks when queried. The benchmark data in the README shows what this looks like in practice. On a large Unity game client codebase with nearly 19,000 files and over 275,000 symbols, the server answers exact text searches in about 6 milliseconds and semantic searches in under 100 milliseconds. Building the index from scratch takes around 45 seconds, but once built it stays loaded in memory and subsequent queries are fast. Beyond search, the server can list the symbols defined in a file, show which files a given file depends on, find which files import a given file, and export the full dependency graph for further analysis. You can also chain these operations in a single request to reduce round trips when working with an AI assistant. The README compares this tool to a standard text search utility called rg (ripgrep). For raw grep on arbitrary files, rg is still the better choice. This server is intended for repeated code-aware work where an AI needs to understand structure and meaning, not just find text matches. The README is partially truncated, so the full installation and configuration instructions are not visible here.

Copy-paste prompts

Prompt 1
I want to connect killop/codedb-mcp to Claude. Walk me through building the Rust server, indexing my codebase, and registering it as an MCP tool in my Claude config.
Prompt 2
Using codedb-mcp, how do I run a semantic search for code that handles user authentication and then chain it with a file dependency lookup in a single request?
Prompt 3
I have a large Unity project with nearly 20,000 files. Walk me through indexing it with codedb-mcp and querying which files depend on a specific module.
Open on GitHub → Explain another repo

← killop on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.