Best for
- AI agents that need to interact with Redis
- Applications that need natural language management of Redis data
- Development workflows using MCP clients like Claude Desktop
M
MCP redis
Official
Natural language interface for managing and searching data in Redis for agentic applications.
GitHub source repository ↗
60FMRS · C
This is the official Redis MCP server, feature-rich, supports multiple data structures and vector search, easy integration, but there are risks of data mishandling and credential leakage.
The Redis MCP Server is the official MCP server from Redis, providing a natural language interface for agentic applications to manage and search data in Redis efficiently. It supports various Redis data structures (strings, hashes, lists, sets, sorted sets, streams, JSON), and offers documentation search and vector search tools. It can be installed via PyPI or Docker, configured via command-line arguments or environment variables, and supports EntraID authentication.
Tools
string
Set and get string values with expiration.
hash
Store and manipulate hash fields, can store vector embeddings.
list
Append and pop list elements, suitable for queues or message brokers.
set
Add, remove, and list set members, support set operations.
sorted set
Manage sorted sets, useful for leaderboards, priority queues, etc.
pub/sub
Publish messages, create subscriptions, and read messages.
streams
Add, read, delete stream data, manage consumer groups.
JSON
Store, retrieve, and manipulate JSON documents.
docs
Search Redis documentation and best practices.
query engine
Manage vector indexes and perform vector search.
server management
Retrieve database information.
Setup
- Install via PyPI:
pip install redis-mcp-server. - Run with uvx:
uvx --from redis-mcp-server@latest redis-mcp-server --url redis://localhost:6379/0. - Or import the JSON config above into your MCP client.
claude_desktop_config.json
{
"mcpServers": {
"RedisMCPServer": {
"command": "uvx",
"args": [
"--from", "redis-mcp-server@latest",
"redis-mcp-server",
"--url", "redis://localhost:6379/0"
]
}
}
}
Fit and risk
Not for
- Pure data pipelines without a natural language layer
- Simple KV scenarios without MCP needs
- Production systems requiring direct Redis for highest performance
Required permissions
- Requires Redis connection credentials (URL, username, password)
- Can read, write, and delete data in Redis
- Can create indexes and perform searches
Risks and side effects
- Data operation errors: AI may accidentally delete or overwrite data
- Credential exposure: Redis credentials could be exposed if misconfigured
- Resource consumption: Large vector indexes may consume memory
Troubleshooting
- Check Redis connection accessibility (default localhost:6379)
- Ensure credentials are correct, especially password and ACL permissions
- Check logs: set `MCP_REDIS_LOG_LEVEL=DEBUG` environment variable to increase verbosity
- Confirm the `redis-server` service is running
Use cases
AI assistants accessing Redis data
Chatbots managing session data
Real-time data search and analytics
Event stream processing
Supported clients
Claude DesktopFull support