Replicate

You run an MCP server that exposes replicate-style AI model inference through a focused, resource-based interface. This server enables automated image generation and management, real-time updates, and easy integration with clients like Claude Desktop, while keeping your API keys and configurations secure.

How to use

You connect to the MCP server from an MCP client and start generating images from natural language prompts. Use the available prompts to specify text-to-image tasks, model selection, and parameter hints. You can monitor progress, receive real-time updates through subscriptions, and apply quality or style presets to improve results. Use webhook notifications to integrate generation events with external alerts or workflows.

How to install

Prerequisites you need before installing the MCP server are configured tooling and accounts.

Ensure you have Python 3.11 or higher and a Replicate API key for model access.

Install the MCP server package using one of the supported methods.

Installation steps

# Installing via Smithery
npx -y @smithery/cli install @gerred/mcp-server-replicate --client claude

# Manual installation options
# Using UV (recommended)
uv pip install mcp-server-replicate

# Using UVX for isolated environments
uvx install mcp-server-replicate

# Using pip
pip install mcp-server-replicate

Claude Desktop integration

Configure Claude Desktop to communicate with the MCP server by editing its client configuration file and adding the MCP server entry.

Open the Claude Desktop configuration file for editing.

# macOS
code ~/Library/Application\ Support/Claude/claude_desktop_config.json

# Windows
code %APPDATA%\Claude\claude_desktop_config.json

MCP server configuration for Claude Desktop

{
  "globalShortcut": "Shift+Alt+A",
  "mcpServers": {
    "replicate": {
      "command": "uv",
      "args": ["tool", "run", "mcp-server-replicate"],
      "env": {
        "REPLICATE_API_TOKEN": "APITOKEN"
      },
      "cwd": "$PATH_TO_REPO"
    }
  }
}

Set your Replicate API key

Provide your Replicate API key in one of two ways.

# Option 1: Environment variable
export REPLICATE_API_TOKEN=your_api_key_here

# Option 2: .env file in your home directory
echo "REPLICATE_API_TOKEN=your_api_key_here" > ~/.env

Usage after setup

With the server connected, you can generate images, browse recent generations, search, and check status using natural language prompts.

Troubleshooting

If the server does not show up in Claude Desktop, review logs and configuration to identify issues.

tail -n 20 -f ~/Library/Logs/Claude/mcp*.log

Documentation and development

You can explore implementation details, API reference, and contribution guides to extend or customize the MCP server. Development steps include cloning the project, installing dependencies, and running tests.

Available tools

image_generation

Create images from natural language prompts with style and quality presets, track progress, and receive updates.

model_selection

Help choose the right AI model for a given task based on capabilities and presets.

parameter_help

Explain and configure model parameters for better control over outputs.

webhook_notifications

Receive external notifications when generation events occur.

subscription_updates

Subscribe to real-time status updates during generation workflows.