
Redis MCP Server
OfficialOfficial Redis server: natural-language access to keys, search, and data structures in Redis.
Add to your client
Copy the config for your MCP client and paste it into its config file.
uvx --from redis-mcp-server@latest redis-mcp-server --url redis://localhost:6379/0Paste into ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"redis-mcp-server": {
"command": "uvx",
"args": [
"--from",
"redis-mcp-server@latest",
"redis-mcp-server",
"--url",
"redis://<user>:<password>@<host>:6379/0"
]
}
}
}Requires `uv` (the Python package runner). Install it from https://docs.astral.sh/uv/ if `uvx` is not found.
Before you start
- Python 3.14+ and the uv package manager (uvx is used to run the published package)
- A reachable Redis instance (local, Redis Cloud, or Azure Managed Redis) and its connection URL or host/port
- Redis credentials if your instance requires auth — a password and/or ACL username; for Azure Managed Redis you can use EntraID/Managed Identity instead
- For vector search / JSON tools, a Redis with the Query Engine and JSON modules (e.g. Redis Stack or Redis 8+)
About Redis MCP Server
Redis MCP Server is the official Model Context Protocol server from Redis. It gives AI agents a natural-language interface to read, write, search, and manage data in Redis, mapping MCP tools onto Redis commands across every major data type.
It covers strings, hashes (including vector embeddings), lists, sets, sorted sets, streams with consumer groups, JSON documents (RedisJSON), and pub/sub messaging. It also exposes a query-engine tool group for creating vector indexes and running vector similarity search, plus server-management and documentation-search helpers.
The server is written in Python and distributed as redis-mcp-server, typically launched with uvx. It connects to any Redis instance via a redis:///rediss:// URL or individual host/port/credentials, and supports ACL users, TLS, and EntraID/Azure AD authentication for Azure Managed Redis.
Tools & capabilities (11)
stringSet/get string values with optional expiration
hashManage hash field-value pairs, including storing vector embeddings
listAppend to and pop from Redis lists
setAdd, remove, and list set members
sorted setScore-based add/range operations for leaderboards and priority queues
streamsAdd, read, and delete stream entries and manage consumer groups
JSONStore, retrieve, and manipulate JSON documents (RedisJSON)
pub/subPublish messages and manage subscriptions, including pattern matching
query engineCreate and manage vector indexes and run vector similarity search
server managementInspect database/server info
docsSearch Redis documentation
When to use it
- Use it when you want an agent to cache and retrieve values in Redis using plain language instead of raw commands
- Use it when building agent memory or session state on top of Redis hashes, JSON, or strings
- Use it when you need vector storage and similarity search for RAG, using Redis as the vector database
- Use it when working with streams and consumer groups for event/task queues from an AI workflow
- Use it when you want quick introspection of a Redis instance (keys, data structures, server info) during debugging
Quick setup
- 1Ensure Python 3.14+ and uv are installed
- 2Run the server with uvx: uvx --from redis-mcp-server@latest redis-mcp-server --url redis://localhost:6379/0 (or install via pip)
- 3Provide connection details via --url / REDIS_URL or REDIS_HOST/REDIS_PORT/REDIS_USERNAME/REDIS_PWD, plus TLS or EntraID options if needed
- 4Add the command to your MCP client config (type: stdio)
- 5Restart your MCP client and verify by asking it to set and get a test key
Security notes
The Redis connection URL can include the password, so avoid embedding production credentials in plaintext config and prefer a dedicated ACL user. An agent with write access can flush or overwrite keys, so scope permissions carefully.
Redis MCP Server FAQ
How does it authenticate to Redis?
Pass a redis:// or rediss:// URL with embedded credentials, or set REDIS_USERNAME/REDIS_PWD. It supports Redis ACL users, TLS with certificate verification, and EntraID/Azure AD (Service Principal, Managed Identity, or Default Azure Credential) for Azure Managed Redis.
Do I need Redis Stack for vector search and JSON?
Yes for those features. The JSON and query-engine (vector index/search) tools require the RedisJSON and Query Engine modules, available in Redis Stack or Redis 8+. Core data-type tools work on any standard Redis.
What transport does it use?
stdio. The server runs over standard input/output for local MCP clients; HTTP streaming transport is noted as planned for the future.
How do I run it without cloning the repo?
Use uvx with the published package: uvx --from redis-mcp-server@latest redis-mcp-server --url <your-redis-url>. There's no npm package — it's a Python project.
Alternatives to Redis MCP Server
Google's official MCP server with prebuilt BigQuery tools, querying datasets via Application Default Credentials.
Read/write Postgres access plus index tuning, EXPLAIN plans, and database health analysis for AI agents.
Official Supabase server: manage tables, run SQL, branches, configs and edge functions from your AI client.
Compare Redis MCP Server with: