daffy0208__ai-dev-standards/mcp-servers/vector-database-mcp

Vector Database MCP Server

MCP server for vector database operations supporting Pinecone, Weaviate, and Chroma.

Installation

npm install
npm run build

Usage

Pinecone

// Connect
{
  "provider": "pinecone",
  "apiKey": "your-api-key",
  "collection": "your-index-name"
}

// Insert vectors
{
  "vectors": [
    { "id": "vec1", "values": [0.1, 0.2, 0.3], "metadata": { "text": "sample" } }
  ]
}

// Search
{
  "query": [0.1, 0.2, 0.3],
  "topK": 5,
  "filter": { "category": "test" }  // optional
}

Weaviate

// Connect
{
  "provider": "weaviate",
  "environment": "http://localhost:8080",
  "apiKey": "your-api-key",  // optional
  "collection": "YourClass"
}

// Same insert/search format as above

Chroma

// Connect
{
  "provider": "chroma",
  "environment": "http://localhost:8000",
  "collection": "your-collection"
}

// Same insert/search format as above

Tools

Tool	Description
connect	Connect to vector database
insert_vectors	Insert vectors with metadata
search_vectors	Semantic search by vector similarity
delete_vectors	Delete vectors by IDs
list_collections	List available collections/indexes

Running

npm start

Testing

npm test