w3-mcp-server-qdrant 0.1.0

MCP server for vector search with Qdrant and Ollama embeddings

Qdrant MCP Server

Model Context Protocol server for vector search using Qdrant and Ollama embeddings.

Features

• Vector Search: Search for similar documents using text queries (auto-embedded via Ollama)
• Collection Management: List and manage Qdrant collections
• Flexible Output: Markdown or JSON response formats
• Read-Only: Safe for querying without modifying data

Prerequisites

Qdrant Server

docker run -p 6333:6333 qdrant/qdrant:latest

Or install locally: Qdrant Quick Start

Ollama Server

# Install: https://ollama.ai
ollama pull nomic-embed-text
ollama serve

Other embedding models available:

• nomic-embed-text (768 dims) - recommended, lightweight
• mxbai-embed-large (1024 dims) - higher quality
• all-minilm (384 dims) - ultra-lightweight

Installation

# Clone/navigate to project directory
cd w3-mcp-server-qdrant

# Install dependencies
pip install -e .

# Or with uv:
uv sync

Configuration

Create a .env file or export environment variables:

# Qdrant
export QDRANT_URL=http://localhost:6333
export QDRANT_API_KEY=  # Optional if using API key auth

# Ollama
export OLLAMA_BASE_URL=http://localhost:11434
export OLLAMA_MODEL=nomic-embed-text

Usage

Run with stdio transport

python server.py

Test with MCP Inspector

uv run mcp dev server.py

Then open http://localhost:5173 in your browser.

Configure in Claude Desktop

Add to ~/.claude/mcp_servers.json:

{
  "mcpServers": {
    "qdrant": {
      "command": "python",
      "args": ["/path/to/server.py"],
      "env": {
        "QDRANT_URL": "http://localhost:6333",
        "OLLAMA_BASE_URL": "http://localhost:11434",
        "OLLAMA_MODEL": "nomic-embed-text"
      }
    }
  }
}

Tools

qdrant_search

Search for similar documents in a collection.

Parameters:

• collection_name (string): Name of the collection to search
• query_text (string): Text to search for (will be embedded)
• limit (integer, 1-100): Max results (default: 5)
• score_threshold (float, 0.0-1.0): Min similarity threshold (default: 0.0)
• response_format (string): "markdown" or "json" (default: markdown)

Example:

Search documents about "machine learning" in the "papers" collection
with a similarity threshold of 0.7, return top 10 results as JSON

qdrant_list_collections

List all collections in Qdrant with metadata.

Parameters:

• response_format (string): "markdown" or "json" (default: markdown)

Examples

Search documents

# Via Claude/MCP interface
qdrant_search(
    collection_name="tech_docs",
    query_text="How do vector databases work?",
    limit=5,
    score_threshold=0.6,
    response_format="markdown"
)

List collections

# Via Claude/MCP interface
qdrant_list_collections(response_format="json")

Architecture

Claude/LLM
    ↓
MCP Server (server.py)
    ├── Ollama: text → embedding
    └── Qdrant: search/store vectors

Data Flow

Search:

1. User provides text query
2. Ollama embeds the query → vector
3. Qdrant searches for similar vectors
4. Results returned with scores and metadata

Error Handling

• Collection not found: Ensure collection exists in Qdrant
• Connection error: Verify Qdrant/Ollama servers are running
• Embedding failed: Check Ollama model is loaded (ollama pull nomic-embed-text)

Development

Run tests

pytest tests/

Code formatting

black server.py
ruff check server.py

Performance Tips

• Score threshold: Use score_threshold to filter low-relevance results and reduce noise
• Result limit: Adjust limit parameter to control number of results (1-100)
• Embedding model: Choose based on quality vs. speed tradeoff:
  • nomic-embed-text: balanced (recommended)
  • all-minilm: fast, lightweight
  • mxbai-embed-large: higher quality but slower

Troubleshooting

Issue: "Cannot connect to Qdrant"

• Check: curl http://localhost:6333/health
• Start Qdrant: docker run -p 6333:6333 qdrant/qdrant:latest

Issue: "Failed to embed text"

• Check: curl http://localhost:11434/api/tags
• Pull model: ollama pull nomic-embed-text
• Start Ollama: ollama serve

Issue: "Collection not found"

• Create collection: Use Qdrant console or add data through external tools

License

MIT