Fal Reve

You run a Model Context Protocol (MCP) server that exposes Reve, a high-quality text-to-image model. It supports real-time and queued generation, image editing, automatic local image downloads, and flexible output formats. This guide walks you through how to use the server with an MCP client, how to install it, and how to configure and troubleshoot common scenarios so you can start generating and editing images with ease.

How to use

Connect to the Fal Reve MCP Server from your MCP client to generate and edit images using Reve. You can run real-time requests for immediate results or queue requests for batch processing. Use prompts to describe scenes, add edits to existing images, and specify output formats and aspect ratios. Each request returns both remote URLs and local file paths for downloaded results, and the server handles missing API keys gracefully with informative error reporting.

How to install

Prerequisites you need before installing: Node.js 18 or higher and a FAL AI API key.

# 1) Get your FAL AI API Key
# Visit FAL AI and follow the sign-up flow to obtain an API key.
# 2) Install and run via universal npx configuration
# 3) Install dependencies and build locally if you prefer a local setup

Additional setup steps

Universal npx configuration provides a portable setup that works on any machine with Node.js. This method runs the MCP server directly via npx and pulls the latest code from the repository when you start.

If you prefer to run a local installation, build the project first to generate executable JS files, then point your MCP client to the built entry point.

{
  "mcpServers": {
    "fal-reve": {
      "command": "npx",
      "args": [
        "-y",
        "https://github.com/PierrunoYT/fal-reve-edit-mcp-server.git"
      ],
      "env": {
        "FAL_KEY": "your-fal-api-key-here"
      }
    }
  }
}

Other configuration options

If you want to run the server locally from a built index, you can use a local path to the generated build index. The example below shows the approach used by Claude Desktop and Kilo Code MCP settings.

{
  "mcpServers": {
    "fal-reve": {
      "command": "node",
      "args": ["path/to/fal-reve-edit-mcp-server/build/index.js"],
      "env": {
        "FAL_KEY": "your-fal-api-key-here"
      }
    }
  }
}

Security and best practices

Keep your FAL API key secure. Do not expose it in client-side code or public repositories. Use environment variables to pass the key to the MCP server and ensure proper access controls in your MCP client configuration.

Troubleshooting

If the MCP server does not appear in your client, verify you are using a valid MCP URL or stdio command, confirm Node.js 18+ is installed, and check that the server has started without critical errors. If you encounter image generation or editing failures, verify your FAL API key and credits, ensure prompts comply with policy, and try simpler prompts or smaller batch sizes. If running locally, ensure the build artifacts exist at the expected path and that the path in the config is absolute.

Examples

Basic image generation and editing prompts can be combined with specific aspect ratios, output formats, and batch sizes to fit your workflow. Use real-time tools for quick results and async tools for longer or multi-image tasks.

Available tools

reve_generate

Generate images with Reve using real-time processing. Parameters include prompt, sync_mode, num_images, output_format, and aspect_ratio. Returns image URLs, metadata, and local download paths.

reve_generate_async

Generate images with async queue processing for longer requests. Supports the same parameters as reve_generate with queue-based progress updates and timeouts.

reve_edit

Edit an existing image by uploading or providing an image URL and transforming it via a text prompt. Returns edited image URLs, metadata, and local download paths.

reve_edit_async

Edit images using async queue processing for batch edits with progress polling and timeout handling.