How to Set Up the Redis MCP Server (2026)

Updated July 13, 2026• Web MCP Guide

The Redis MCP server connects Claude, Cursor, or VS Code straight to a Redis instance using a standard connection string. Once configured, your AI assistant can inspect keys, read data structures, run commands, and query vector search indexes — no separate Redis client code needed.

Prerequisites

• A running Redis instance — local, Docker, self-hosted, or Redis Cloud
• Connection details: host, port, and credentials (or a full REDIS_URL)
• Python 3.10+ with uv/uvx installed, or Node.js 18+ depending on which server build you run
• An MCP-compatible client: Claude Desktop, Cursor, or VS Code

Step 1 — Gather Your Redis Connection Details

You need either a single connection string or the individual host, port, and auth pieces. For Redis Cloud, both are visible on your database's details page. For a local instance, the default is localhost:6379 with no password unless you set one.

redis://default:YOUR_PASSWORD@your-host.redis.cloud:12345

If your password contains special characters like @ or :, URL-encode them or use the split environment variables instead of a single connection string.

Step 2 — Install the Redis MCP Server

Redis maintains an open-source MCP server (redis/mcp-redis) that can be run without a separate install step using uvx:

uvx --from redis-mcp-server redis-mcp-server

Check the official GitHub repository for the exact package name and installation flags — MCP server packaging changes periodically, and some clients prefer a pip-installed version over uvx.

Step 3 — Configure Your MCP Client

Add the server to your client config, passing connection details as environment variables. Replace the placeholders with your actual Redis credentials.

Claude Desktop

Config: ~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "redis": {
      "command": "uvx",
      "args": ["--from", "redis-mcp-server", "redis-mcp-server"],
      "env": {
        "REDIS_HOST": "your-host.redis.cloud",
        "REDIS_PORT": "12345",
        "REDIS_USERNAME": "default",
        "REDIS_PASSWORD": "YOUR_PASSWORD",
        "REDIS_SSL": "true"
      }
    }
  }
}

Cursor IDE

Config: ~/.cursor/mcp.json

{
  "mcpServers": {
    "redis": {
      "command": "uvx",
      "args": ["--from", "redis-mcp-server", "redis-mcp-server"],
      "env": {
        "REDIS_URL": "redis://default:YOUR_PASSWORD@your-host.redis.cloud:12345"
      }
    }
  }
}

Restart your MCP client after saving. Redis should appear in your available tools list.

Step 4 — Test the Connection

Try a query in your client's chat:

"List the first 20 keys in Redis matching the pattern session:*"

A real list of matching keys confirms the connection is working. If you get a connection error, verify the host, port, and credentials with redis-cli -h HOST -p PORT -a PASSWORD ping before troubleshooting the MCP config further.

What You Can Do With the Redis MCP Server

• List and search keys by pattern, and check their type and TTL
• Read the contents of hashes, sets, lists, and sorted sets
• Run individual Redis commands and interpret the output in plain language
• Query vector search indexes when RediSearch or Redis Stack is enabled
• Inspect memory usage and key expiration to debug caching issues
• Cross-reference application code with live cache state, e.g. “why is this key stale?”

If your stack also uses Redis alongside cloud file storage, the filesystem MCP server guide covers giving your assistant safe, scoped access to local files as well.

Troubleshooting

Frequently Asked Questions