Claude Context

Claude Context

Bring your codebase as context to Claude Code

4,554
Stars
415
Forks
0
Releases

Overview

Claude Context MCP server is an MCP (Model Context Protocol) plugin that injects deep, semantic code search results from your entire codebase into Claude Code and other AI coding agents. It indexes your project with semantic search, retrieving relevant code chunks from millions of lines and bringing them directly into the AI’s context, enabling more accurate answers without loading the whole repository per request. The server stores your codebase in a vector database and leverages a suite of embedding providers (OpenAI, VoyageAI, Ollama, Gemini) to convert code into dense vectors. It supports hybrid search combining BM25 and dense vectors to locate pertinent code, and uses intelligent AST-based code chunking for efficient search. It is designed to be scalable, configurable, and compatible with a wide range of MCP clients and IDEs via the standard MCP protocol and stdio transport. It emphasizes cost efficiency by only including relevant code in the active context, rather than loading entire directories, and supports incremental re-indexing for changes.

Details

Owner
zilliztech
Language
TypeScript
License
MIT License
Updated
2025-12-07

Features

Semantic code search context

Provides context from the entire codebase to Claude Code via MCP by retrieving relevant code chunks instead of loading whole directories.

Cost-efficient context provisioning

Stores code in a vector database and uses only related code in context to reduce token usage and costs.

Hybrid code search

Supports BM25 plus dense vector search to locate pertinent code efficiently.

Incremental indexing

Efficiently re-indexes only changed files using Merkle-tree-based change tracking.

Intelligent code chunking

Utilizes AST-based code chunking for more meaningful search units and results.

Scalable vector database integration

Works with Milvus or Milvus Cloud for scalable, managed vector search.

Flexible embedding providers

Supports OpenAI, VoyageAI, Ollama, Gemini embedding providers.

Broad MCP client compatibility

Follows the standard MCP protocol via stdio transport and works with numerous MCP clients and IDEs.

Audience

AI developersIntegrate semantic code search context into Claude Code and other AI coding agents to improve search relevance and contextual accuracy.
Software teams / engineersIndex and search large codebases to provide contextual code snippets to AI copilots and assistants.
MCP client integrators / IDE vendorsConfigure and deploy Claude Context MCP server across a broad set of MCP clients and development tools.

Tags

semantic code searchMCP serverClaude Codecodebase contextMilvusMilvus CloudOpenAI embeddingembedding providersBM25 + dense vectorAST-based chunkingincremental indexingMCP compatibility