Qdrant MCP server

The Qdrant MCP server enables AI agents to store and retrieve semantic information using vector search through MCP. Built for persistent memory, contextual knowledge, and efficient semantic retrieval workflows.

When you should use this server

• Give AI assistants persistent memory across sessions and conversations
• Store knowledge, facts, or documents in a vector database for later retrieval
• Retrieve semantic matches to queries instead of relying on exact keyword lookups
• Manage contextual memory for long-running workflows or applications

Key features

• Semantic vector storage and retrieval
• Context-aware memory persistence
• Efficient similarity search
• Metadata filtering and payload storage
• Cross-session memory for AI assistants
• Compatibility with both cloud and self-hosted deployments

Requirements

• Hosting: Works with a running Qdrant instance (cloud or self-hosted)
• Authentication: Standard Qdrant API authentication (if enabled)
• Collections: Requires specifying a collection name unless a default is configured

Tools provided

qdrant-store

Stores information in the Qdrant vector database with optional metadata.

Parameters:

• information (string, required) — content to store
• metadata (JSON, optional) — associated metadata to store alongside the vector
• collection_name (string, required if no default) — collection to store data in

Returns:

• Confirmation message with vector ID and status

qdrant-find

Retrieves semantically relevant information from Qdrant based on the meaning of the query.

Parameters:

• query (string, required) — text to search for semantically similar content
• collection_name (string, required if no default) — collection to search

Returns:

• Matching stored information, ordered by semantic similarity

Notes

• Best used as a memory backend for AI assistants needing semantic recall
• Requires an active Qdrant instance; supports both Qdrant Cloud and self-hosted deployments