Redis MCP Server

Redis's own MCP server, packaged as a Gemini CLI extension. It gives your agent a natural-language interface to a Redis database, so it can store and read data, work with Redis data structures and run vector search without you writing the commands. What you can do with it - Set and get strings with an expiry, for caching and session data - Work with hashes, lists, sets, sorted sets and streams, including stream consumer groups - Store, read and update JSON documents by path - Create vector indexes and run vector and hybrid searches - Publish messages to channels and subscribe to them - Scan keys, set expiries, rename and delete keys, and read server info and stats - Search the Redis documentation for answers How it connects Install it with gemini extensions install https://github.com/redis/mcp-redis. It runs on your machine through uvx (the redis-mcp-server package on PyPI) and connects to the Redis URL you set in REDIS_URL, such as redis://localhost:6379/0, or a rediss:// address for Redis Cloud and other TLS connections. With no URL set, it connects to a local Redis at 127.0.0.1:6379. Good to know - It starts with uvx, so you need uv installed. - For read-only access, connect as a Redis ACL user without write permissions; Redis's README shows how to create one. - The server also supports Redis Cluster mode and Microsoft Entra ID sign-in for Azure Managed Redis. - Outside Gemini CLI, the same server runs as a local MCP server with uvx or Docker.