ChunkHound
A local-first codebase intelligence tool that enables AI assistants to research codebases using semantic search, multi-hop relationship discovery, and structural parsing. It allows users to extract architectural patterns and institutional knowledge a
What is ChunkHound?
ChunkHound is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to local-first codebase intelligence tool that enables ai assistants to research codebases using semantic search, multi-hop relationship discovery, and structural parsing. it allows users to extract arch...
A local-first codebase intelligence tool that enables AI assistants to research codebases using semantic search, multi-hop relationship discovery, and structural parsing. It allows users to extract architectural patterns and institutional knowledge a
This server falls under the Knowledge & Memory category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- A local-first codebase intelligence tool that enables AI ass
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx chunkhoundConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use ChunkHound
ChunkHound is a local-first codebase intelligence tool that indexes your source code with semantic chunking (the "cAST Algorithm") backed by DuckDB and Tree-sitter, then exposes semantic search and multi-hop relationship discovery to any MCP-aware AI assistant. It supports 32 programming languages, watches files in real time, and can query code changes across Git commit ranges — making it ideal for understanding large codebases, tracking architectural decisions, and surfacing institutional knowledge without sending code to the cloud.
Prerequisites
- Python 3.10 or higher installed on your machine
- uv package manager installed (curl -LsSf https://astral.sh/uv/install.sh | sh)
- An MCP client such as Claude Desktop or Claude Code
- Optional: a VoyageAI or OpenAI API key for semantic (vector) embeddings; regex search works without one
- A local Git repository to index
Install ChunkHound via uv
Install ChunkHound as a system-wide tool using uv. This makes the chunkhound CLI available globally.
uv tool install chunkhoundCreate a project configuration file
In the root of the codebase you want to index, create a .chunkhound.json file. At minimum you can leave it empty or specify an embedding provider. VoyageAI is recommended for best results, but you can omit the embedding block to use regex-only search.
{
"embedding": {
"provider": "voyageai",
"api_key": "pa-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
},
"llm": {
"provider": "claude-code-cli"
}
}Index your codebase
Run the indexer from your project root. ChunkHound parses all supported source files with Tree-sitter, chunks them semantically, and stores the index in a local DuckDB database. File watching keeps the index up to date as you edit.
chunkhound indexTest search from the CLI
Verify the index works by running a semantic search query from the terminal. The --last-n flag limits the search scope to the N most recent commits.
chunkhound search "authentication logic" --last-n 20Add ChunkHound to your MCP client configuration
Register ChunkHound as an MCP server in your Claude Desktop or Claude Code config. Point it at the project directory you want to make searchable.
{
"mcpServers": {
"chunkhound": {
"command": "uvx",
"args": ["chunkhound", "mcp", "--project-dir", "/path/to/your/project"]
}
}
}Restart your MCP client and verify connectivity
Restart Claude Desktop (or reload Claude Code). ChunkHound should appear as an available MCP server. You can now ask the AI to search your codebase directly.
ChunkHound Examples
Client configuration
Claude Desktop configuration pointing ChunkHound at a local project directory with VoyageAI embeddings.
{
"mcpServers": {
"chunkhound": {
"command": "uvx",
"args": ["chunkhound", "mcp", "--project-dir", "/Users/you/projects/myapp"],
"env": {
"VOYAGE_API_KEY": "pa-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
}
}
}
}Prompts to try
Example prompts you can send to Claude once ChunkHound is connected.
- "Search the codebase for how database connections are managed"
- "Find all functions related to user authentication and explain how they interact"
- "What changed in the payment module in the last 30 commits?"
- "Show me the entry points for the HTTP request pipeline"
- "Find any code that handles JWT token validation"Troubleshooting ChunkHound
chunkhound: command not found after installation
Run 'uv tool update-shell' to add the uv tools bin directory to your PATH, then open a new terminal session.
Semantic search returns no results or errors about embeddings
Confirm your embedding API key is correct in .chunkhound.json. For regex-only search without an API key, remove the 'embedding' block from the config and re-run 'chunkhound index'.
MCP server is not listed in Claude Desktop after adding config
Ensure the project-dir path is absolute and the directory contains a valid .chunkhound.json. Check Claude Desktop logs (Help > Show Logs) for startup errors from the chunkhound process.
Frequently Asked Questions about ChunkHound
What is ChunkHound?
ChunkHound is a Model Context Protocol (MCP) server that local-first codebase intelligence tool that enables ai assistants to research codebases using semantic search, multi-hop relationship discovery, and structural parsing. it allows users to extract architectural patterns and institutional knowledge a It connects AI assistants to external tools and data sources through a standardized interface.
How do I install ChunkHound?
Follow the installation instructions on the ChunkHound GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with ChunkHound?
ChunkHound works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is ChunkHound free to use?
Yes, ChunkHound is open source and available under the MIT license. You can use it freely in both personal and commercial projects.
ChunkHound Alternatives — Similar Knowledge & Memory Servers
Looking for alternatives to ChunkHound? Here are other popular knowledge & memory servers you can use with Claude, Cursor, and VS Code.
MemPalace
★ 52.6kA local AI memory system that stores all conversations verbatim and organizes them into navigable structures. It provides 19 MCP tools for AI assistants to search and retrieve past decisions, debugging sessions, and architecture debates automatically
Kratos
★ 25.7k🏛️ Memory System for AI Coding Tools - Never explain your codebase again. MCP server with perfect project isolation, 95.8% context accuracy, and the Four Pillars Framework.
Context Mode
★ 15.4kAn MCP server that preserves LLM context by intercepting large data outputs and returning only concise summaries or relevant sections. It enables efficient sandboxed code execution, file processing, and documentation indexing across multiple programm
Memu
★ 13.7kMemory for 24/7 proactive agents like OpenClaw.
MemOS
★ 9.3kMemOS (Memory Operating System) is a memory management operating system designed for AI applications. Its goal is: to enable your AI system to have long-term memory like a human, not only remembering what users have said but also actively invoking, u
Everos
★ 5.4kBuild, evaluate, and integrate long-term memory for self-evolving agents.
Browse More Knowledge & Memory MCP Servers
Explore all knowledge & memory servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up ChunkHound in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use ChunkHound?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.