MCPIndependentZilliz
Claude Context
Searches an entire codebase by meaning using an index in a vector database.
Zilliz’s server splits code along the syntax tree, indexes it in Milvus or Zilliz Cloud and combines vector search with BM25, so the agent finds snippets from a description instead of reading whole directories. The index updates incrementally; it requires a key for an embedding model (OpenAI, VoyageAI, Gemini or local Ollama) and a vector database.
Installation
Manual install
$claude mcp add claude-context -e OPENAI_API_KEY=<API_KEY> -e MILVUS_ADDRESS=<ZILLIZ_ENDPOINT> -e MILVUS_TOKEN=<ZILLIZ_API_KEY> -- npx @zilliz/claude-context-mcp@latestDetails
- Works in
- Claude Code, Claude, Codex, Gemini CLI, Cursor, Windsurf, VS Code
- Transport
- local
- License
- MIT
- Tasks
- Code
- Rating
- Source
- github.com/zilliztech/claude-context (opens in a new tab)
Tags
- semantic search
- code search
- milvus
- embeddings
- code index
A pack is your own set of skills and MCP servers. sciver loads it in your AI with one command.