Redis MCP Server

The Redis MCP Server provides a natural language interface for agentic applications to manage and search data in Redis. It integrates with MCP (Model Content Protocol) clients, allowing AI agents to perform complex Redis operations using natural language commands.

Key capabilities include:

• Natural Language Queries: Agents can issue commands like "Store the entire conversation in a stream" or "Cache this item".
• Full Redis Support: Handles hashes, lists, sets, sorted sets, streams, and more.
• Search & Filtering: Supports efficient data retrieval and vector search.
• Transport Support: Currently supports the stdio transport. Support for stremable-http is planned for the future.

To use the Redis MCP Server with Claude Desktop, add a configuration entry to your claude_desktop_config.json. You must provide the full path to the uvx executable.

Basic Redis connection

Connects to a local Redis instance via stdio.

Azure Managed Redis with EntraID authentication

For Azure Managed Redis, provide the connection URL and set the following environment variables to handle EntraID authentication:

• REDIS_ENTRAID_AUTH_FLOW: Set to default_credential.
• REDIS_ENTRAID_SCOPES: Set to https://redis.azure.com/.default.

{
  "mcpServers": {
    "redis-mcp-server": {
        "type": "stdio",
        "command": "/FULL/PATH/TO/uvx",
        "args": [
            "--from", "redis-mcp-server@latest",
            "redis-mcp-server",
            "--url", "redis://localhost:6379/0"
        ]
    }
  }
}

The recommended way to install the Redis MCP Server is via PyPI using uvx. This command downloads the server on the fly, creates a temporary environment, and runs it.

To configure it in an MCP-compatible tool (like Claude Desktop), use the following JSON structure:

{
  "mcpServers": {
    "RedisMCPServer": {
      "command": "uvx",
      "args": [
        "--from",
        "redis-mcp-server@latest",
        "redis-mcp-server",
        "--url",
        "\"redis://localhost:6379/0\""
      ]
    }
  }
}

You can run the server directly from the GitHub repository using uvx. It is recommended to use a tagged release rather than the main branch to avoid breaking changes.

Example running version 0.2.0:

uvx --from git+https://github.com/redis/mcp-redis.git@0.2.0 redis-mcp-server --url redis://localhost:6379/0

The server uses Python's standard logging. By default, it logs at the WARNING level. You can control the verbosity using the MCP_REDIS_LOG_LEVEL environment variable.

Accepted values (case-insensitive):

• DEBUG
• INFO
• WARNING (or WARN)
• ERROR
• CRITICAL (or FATAL)
• NOTSET
• Numeric values (e.g., "10", "+20")

Note on Handlers: If the host environment (like VS Code or uv) has already installed console handlers, the server will not add its own. It will only adjust existing handler thresholds to ensure your chosen level is not filtered out.

# Set log level to DEBUG for troubleshooting
MCP_REDIS_LOG_LEVEL=DEBUG uvx --from redis-mcp-server@latest redis-mcp-server --url redis://localhost:6379/0

To use the Redis MCP Server within the OpenAI Agents SDK, install the required Python SDK and configure your OpenAI API key. You can then run the provided assistant example to start your agent workflow.

1. Install the SDK: pip install openai-agents
2. Set your API key: export OPENAI_API_KEY="<openai_token>"
3. Run the application: python3.14 redis_assistant.py

The Redis MCP Server can be configured using command line arguments or environment variables. Command line arguments take precedence over environment variables, which in turn take precedence over default values.

Command Line Arguments

Use CLI flags to specify connection details. This is useful for quick starts or when running via uvx.

Environment Variables

Use environment variables for persistent configuration. Common variables include REDIS_HOST, REDIS_PORT, REDIS_DB, REDIS_USERNAME, and REDIS_PWD.

# Basic Redis connection via CLI
uvx --from redis-mcp-server@latest redis-mcp-server \
  --host localhost \
  --port 6379 \
  --password mypassword

# Using Redis URI (simpler)
uvx --from redis-mcp-server@latest redis-mcp-server \
  --url redis://user:pass@localhost:6379/0

You can use a Dockerized deployment. You can either build your own image using the provided Dockerfile or use the official mcp/redis image from Docker Hub.

Build your own image:

docker build -t mcp-redis .

Claude Desktop Configuration (Docker):

To run the server inside a container via Claude Desktop, add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "redis": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "--name",
        "redis-mcp-server",
        "-i",
        "-e", "REDIS_HOST=<redis_hostname>",
        "-e", "REDIS_PORT=<redis_port>",
        "-e", "REDIS_USERNAME=<redis_username>",
        "-e", "REDIS_PWD=<redis_password>",
        "mcp-redis"
      ]
    }
  }
}

To use the official image, replace mcp-redis with mcp/redis in the arguments.

The Redis MCP Server supports EntraID authentication for Azure Managed Redis, providing automatic token management and renewal. You can choose from three authentication flows:

1. Service Principal: Application-based authentication using client credentials.
2. Managed Identity: For Azure-hosted applications (supports system_assigned and user_assigned).
3. Default Azure Credential: Automatic credential discovery (recommended for local development with Azure CLI).

To use EntraID, set REDIS_ENTRAID_AUTH_FLOW and the corresponding identity credentials via environment variables.

# Example: Local development with Azure CLI
az login
export REDIS_ENTRAID_AUTH_FLOW=default_credential
export REDIS_URL=redis://your-azure-redis.redis.cache.windows.net:6379

The preferred method for Augment is using the Easy MCP feature. Alternatively, you can manually import the server by adding the following JSON configuration to your Augment setup. This example uses uvx to run the latest version of the server pointing to a local Redis instance.

{
  "mcpServers": {
    "Redis MCP Server": {
      "command": "uvx",
      "args": [
        "--from",
        "redis-mcp-server@latest",
        "redis-mcp-server",
        "--url",
        "redis://localhost:6379/0"
      ]
    }
  }
}