Sourcerer MCP is an MCP server for semantic code search and navigation that helps AI agents work efficiently without burning through costly tokens. Instead of reading entire files, agents can search conceptually and jump directly to the specific functions, classes, and code chunks they need. The server builds a semantic index of your codebase by parsing sources with Tree-sitter into ASTs, extracting meaningful chunks (functions, classes, methods, types) with stable IDs, and attaching source code, location data, and contextual summaries. Chunks are identified with IDs like file.ext::Type::method. It watches for file changes with fsnotify, respects .gitignore files, and automatically re-indexes changed files, storing modification times as metadata. A chromem-go backed vector database stores persistent embeddings generated via OpenAI, enabling semantic similarity searches rather than exact text matching. MCP tools include semantic_search, get_chunk_code, find_similar_chunks, index_workspace, and get_index_status to help AI agents retrieve relevant code with reduced token usage and cognitive load.