CodeWeave Mcp
@semihkayan
Save tokens while coding — your AI agent gets structured code context, not file dumps.
Overview
What is CodeWeave Mcp?
CodeWeave Mcp is an MCP server that gives AI agents structured code intelligence by querying local indexes — AST, call graph, type graph, and hybrid semantic search — instead of dumping entire files into context. It is designed for developers using AI coding assistants like Claude Code and VS Code who want cheaper, more precise code understanding.
How to use CodeWeave Mcp?
Run npx @codeweave/mcp inside your project directory. The setup wizard automatically installs dependencies, configures your MCP client, and indexes your project. After setup, open your project in Claude Code or VS Code and ask questions using the 8 available tools (e.g., semantic_search, get_function_source, get_impact_analysis).
Key features of CodeWeave Mcp
- 8 MCP tools for code understanding workflow
- Semantic search with 6-stage hybrid retrieval pipeline
- tree-sitter AST parsing for 7 programming languages
- Local vector embeddings via Qwen3-Embedding-0.6B and LanceDB
- Call graph and type graph for impact analysis
- Incremental file watcher for automatic reindexing
Use cases of CodeWeave Mcp
- Search for functions by meaning when you don't know exact names
- Get the source code of a specific function without reading entire files
- Analyze the blast radius of a code change before making edits
- Discover stale or outdated docstrings across your codebase
- Understand dependencies between functions in large monorepos
FAQ from CodeWeave Mcp
What runtime does CodeWeave Mcp require?
Node.js 20 or later. Ollama must also be installed locally to run the embedding model.
Where does CodeWeave Mcp store its data?
Indexes live on disk in a .code-context directory. LanceDB stores vector embeddings; call and type graphs are stored as JSON.
What happens if the embedding service is unavailable?
The system degrades gracefully: search falls back to full-text only, and AST-based tools (dependencies, impact analysis, module summary) work without any embedding infrastructure.
Which transport protocol does CodeWeave Mcp use?
It uses the MCP protocol over stdio transport. It works with any MCP-compatible client, including Claude Code, VS Code, and Cursor.
What are the known limits or current status?
CodeWeave Mcp is under active development. The first run requires a one-time download of Ollama and the embedding model (takes a few minutes).
