The structure is as follows:
mcp-server-redis/
├── src/
│ ├── __init__.py
│ ├── main.py # Main entry point
│ ├── config.py # Configuration
│ ├── connection.py # Redis connection management
│ ├── resources/ # Resource implementations
│ │ ├── __init__.py
│ │ ├── status.py # Connection status resources
│ │ └── keys.py # Key-related resources
│ └── tools/ # Tool implementations
│ ├── __init__.py
│ ├── basic.py # Basic Redis operations
│ ├── lists.py # List operations
│ ├── hashes.py # Hash operations
│ ├── sets.py # Set operations
│ └── pubsub.py # Pub/Sub operations
├── tests/ # Test directory
│ └── __init__.py
├── README.md
└── pyproject.toml
To use this server, you'll need to:
Install the required dependencies:
[If not already installed] https://docs.astral.sh/uv/
uv venv
source .venv/bin/activate
uv sync
Refer to the .env.example file to configure your Redis connection by adding the appropriate variables.
Run the server:
python src/server.py
Resources:
- redis://status - Current connection status
- redis://info - Redis server information
- redis://keys/{pattern} - List keys matching a pattern
Tools:
- Basic operations:
- get_value - Get a value
- set_value - Set a value with optional expiry
- delete_key - Delete a key
- increment - Increment a numeric value
- List operations:
- list_push - Push values to a list
- list_range - Get range of list values
- Hash operations:
- hash_set - Set hash fields
- hash_get - Get hash fields
- Set operations:
- set_add - Add to a set
- set_members - Get set members
- Pub/Sub:
- publish_message - Publish a message
Each tool includes error handling and returns meaningful responses. The server will automatically reconnect to Redis if the connection is lost.
You can test it with the MCP Inspector:
mcp dev src/server.py
Or install it in Claude Desktop:
mcp install src/server.py
User: What's the current Redis connection status?
LLM: Let me check the Redis connection status for you.
[Reading resource redis://status]
The Redis server is currently connected and running on localhost:6379, database 0.
User: Can you store my name "Alice" in Redis?
LLM: I'll help you store your name in Redis using the set_value tool.
[Calling set_value with key="name" value="Alice"]
Successfully set key 'name'
I've stored your name "Alice" in Redis under the key "name". Would you like me to verify the value was stored correctly by retrieving it?
User: Yes please verify it
LLM: I'll retrieve the value using the get_value tool.
[Calling get_value with key="name"]
Alice
Yes, I can confirm that your name "Alice" was successfully stored in Redis and can be retrieved correctly. The value matches exactly what we stored.
This implementation provides a solid foundation for Redis integration through MCP. You can extend it further by adding more Redis commands as needed for your specific use case.