Qdrant MCP
The Model Context Protocol (MCP) is an open standard that lets AI assistants call external tools. Qdrant provides three MCP servers. Each one does a different job, and you can connect any combination of them to the same assistant.
| Server | What it does | Runs | Sign-in |
|---|---|---|---|
| Qdrant Cloud MCP | Manages Qdrant Cloud clusters, serverless spaces, backups, access, and billing. | Remote, hosted by Qdrant | OAuth |
| Qdrant Docs MCP | Searches the Qdrant documentation, articles, blog, and courses, and finds code snippets. | Remote, hosted by Qdrant | None |
| Qdrant MCP (open source) | Stores and retrieves information in a Qdrant collection, as semantic memory or code search. | On your machine or your infrastructure | Qdrant URL and API key |
Which One Do I Need?
- To create, back up, or monitor clusters from your assistant, use the Qdrant Cloud MCP server. It works on infrastructure, not on your data.
- To have your assistant answer Qdrant questions with current documentation, use the Qdrant Docs MCP server. It needs no account and no setup beyond the URL.
- To give your assistant a memory, or to search your own content or code, use the open source Qdrant MCP server. It reads and writes a collection in any Qdrant deployment, including a local one.
The Qdrant Cloud MCP server also exposes documentation search, so if you already use it, you do not need the Docs server as well.
Related Tools
MCP servers give an assistant the ability to act. Agent skills give it the knowledge to act well. They work together: connect a server, load a skill, and your assistant can both operate Qdrant and tune it.
Ready to try Qdrant?