MCP server

Qdrant

Semantic memory store and retrieval backed by a Qdrant vector database

Qdrant's official MCP server acts as a semantic memory layer, storing information with optional metadata and retrieving it via vector search. It runs via uvx against a remote Qdrant URL or local path, using FastEmbed models by default, and supports stdio, SSE and streamable HTTP.

Connect

  1. 1Add the server to Codex
    codex mcp add qdrant --env QDRANT_URL=YOUR_QDRANT_URL --env QDRANT_API_KEY=YOUR_QDRANT_API_KEY --env COLLECTION_NAME=YOUR_COLLECTION_NAME -- uvx mcp-server-qdrant

Configs are generated from the publisher's documented endpoint. Check their docs for plan requirements and scopes. New to this? Read the step-by-step tutorial.

Tools 2

qdrant-storeqdrant-find

Things to ask

  • “Remember that our staging DB is on port 5433”
  • “Find what I stored about the deployment process”
  • “Store this code snippet with the description 'retry with backoff'”

Related