sourcerer-mcp

Overview

Sourcerer MCP is a specialized MCP server designed to help AI agents work with codebases more efficiently. Instead of reading entire files, it allows agents to search conceptually and jump directly to specific functions, classes, and code chunks they need. The server uses Tree-sitter to parse code, extracts meaningful chunks with stable IDs, and maintains a vector database using OpenAI embeddings for semantic search. It supports multiple programming languages including Go, JavaScript, Python, and TypeScript.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• semantic_search

  Find relevant code using semantic search

  Try:Find functions related to user authenticationTry:Show me code that handles database connections
• get_chunk_code

  Retrieve specific chunks by ID

  Try:Get the code for main.go::Function::mainTry:Show me the implementation of utils.go::Class::HttpClient
• find_similar_chunks

  Find similar chunks of code

  Try:Find code similar to this authentication functionTry:Show me implementations that match this pattern
• index_workspace

  Manually trigger re-indexing of the codebase

  Try:Re-index the entire projectTry:Update the code search index
• get_index_status

  Check indexing progress of the codebase

  Try:Show me indexing statusTry:Is the code search index up to date?

Comparable tools

Installation

go install github.com/st3v3nmw/sourcerer-mcp/cmd/sourcerer@latest

brew tap st3v3nmw/tap
brew install st3v3nmw/tap/sourcerer

claude mcp add sourcerer -e OPENAI_API_KEY=your-openai-api-key -e SOURCERER_WORKSPACE_ROOT=$(pwd) -- sourcerer

{
  "mcpServers": {
    "sourcerer": {
      "command": "sourcerer",
      "env": {
        "OPENAI_API_KEY": "your-openai-api-key",
        "SOURCERER_WORKSPACE_ROOT": "/path/to/your/project"
      }
    }
  }
}

FAQ

What programming languages are supported?

Currently supported: Go, JavaScript, Markdown, Python, TypeScript. Planned: C, C++, Java, Ruby, Rust, and others.

What are the system requirements?

Requires an OpenAI API key for generating embeddings, Git repository (respects .gitignore files), and a .sourcerer/ directory added to .gitignore.

On Hacker News

Recent discussion from the developer community.

• Show HN: Sourcerer – MCP for semantic code search that reduces token waste▲ 9
  Story by st3v3nmw · 2025-08-30

Compare sourcerer-mcp with