Purpose: fal-mcp-server is a server based on the Model Context Protocol (MCP) that enables Claude Desktop and other MCP clients to call models on Fal.ai to generate images, videos, music, and audio.
Main capabilities: Provides 18 media generation tools, including generate_image (text-to-image), generate_image_structured (control composition, lighting, subject), generate_image_from_image (style transfer), remove_background (remove background to generate transparent PNG), upscale_image (2x/4x upscaling), edit_image (natural language editing), inpaint_image (masked local editing), resize_image (social media sizes), compose_images (watermark/Logo overlay), generate_video, generate_video_from_image, generate_video_from_video, generate_music; also list_models, recommend_model, get_pricing, get_usage, upload_file. Through dynamic discovery, 600+ models on the Fal.ai platform can be used, supporting full model IDs or friendly aliases, with TTL caching.
Usage: Requires Python 3.10 or above, a Fal.ai API key (with free quota), and Claude Desktop or a compatible MCP client. It can be run directly with uvx, or installed through the Claude Code plugin marketplace or using the official Docker image. Configuration is written to Claude Desktop's claude_desktop_config.json (macOS: ~/Library/Application Support/Claude/claude_desktop_config.json; Windows: %APPDATA%\Claude\claude_desktop_config.json), then restart Claude Desktop. HTTP/SSE modeThe SSE endpoint below is http://localhost:8080/sse（Docker）或 http://localhost:8000/sse（pip），消息端点为 POST http://localhost:8080/messages/。Docker When using upload_file, a volume must be mounted, and a path inside the container must be used.
Applicable scenarios: generating images in a Claude conversation, converting images to video, generating music, text-to-speech or audio transcription, and retrieving available models by category or keyword.
Limitations: Video and music generation may take several minutes, which is within the normal range; you can switch to a faster model variant or reduce the duration and resolution; rate limits may be encountered, requiring waiting or upgrading the Fal.ai plan; FAL_KEY must be set before use; local file upload under Docker requires additional volume mount configuration.
