Claude Context
A semantic code search MCP server: indexes the project into a vector database and gives the agent only relevant snippets
Install
claude mcp add claude-context \
-e OPENAI_API_KEY=sk-your-openai-api-key \
-e MILVUS_ADDRESS=your-zilliz-cloud-public-endpoint \
-e MILVUS_TOKEN=your-zilliz-cloud-api-key \
-- npx @zilliz/claude-context-mcp@latestRequires Node.js 20+.
This is third-party code. Review the repository files before installing.
What it does
Claude Context splits project code into chunks, builds embeddings and stores them in Milvus or Zilliz Cloud. When the agent needs context, it runs a semantic query and gets relevant code without multi-step file searching. This saves tokens on large repositories. The server works with Claude Code, Codex, Gemini CLI, Cursor, Windsurf, Claude Desktop and other MCP clients.
Who it is for. For developers on large codebases where the agent struggles to find code with plain search.
Good fit when
- The repository is too large for the agent to read in full
- The agent spends many steps finding the right files
- You need to search code by meaning rather than exact names
Not a fit when
- The project is small and grep is enough
- Code must not be sent to external embedding services
Example request
Index this repository and find where user authentication is handledLimitations
It needs a vector database, Zilliz Cloud or your own Milvus, and an embedding provider key, OpenAI by default. The OpenAI API is not reachable from Russia without a VPN. Code chunks are sent to the embedding service and stored in the vector database. Node.js 20+ is required.
How to disable. Run claude mcp remove claude-context or delete the claude-context block from the client's MCP config.
MCP
- Transport
- stdio
- Authentication
- API key
| Environment variables | |
|---|---|
| OPENAI_API_KEY required, secret | OpenAI key for the embedding model |
| MILVUS_ADDRESS | Milvus address or Zilliz Cloud public endpoint |
| MILVUS_TOKEN required, secret | Zilliz Cloud API key or Milvus token |
Security check
- Reads the whole project codebase
- Sends code chunks to an embedding service and vector database
README in short
The README explains why to index code in a vector database and lists requirements: Zilliz Cloud or Milvus and an OpenAI key. It gives the claude mcp add command and ready configs for about ten clients, including Codex, Gemini CLI, Qwen Code and Cursor. It also covers the VS Code extension and the core library. MIT licensed.
FAQ
Can I avoid Zilliz Cloud?
Yes, point MILVUS_ADDRESS at your own Milvus instance.
Is there more than MCP?
The repository also has a VS Code extension and the @zilliz/claude-context-core package.
Related
A skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review
Skills for real engineers by Matt Pocock
Skills For Real Engineers
Small composable skills for engineering with agents: plan grilling, TDD, bug diagnosis, code review and architecture
GitHub toolkit for spec-driven development: the specify CLI adds agent commands and skills to a project, from principles to implementation
Reference MCP servers
Model Context Protocol servers
Official reference MCP servers: Filesystem, Fetch, Git, Memory, Sequential Thinking, Time and Everything