It is a thin wrapper, and that is the appeal: one call, four options, no image pipeline to maintain. The background argument is the detail worth knowing — asking for transparent up front saves the cut-out work that usually follows a generated asset. Note that both keys are yours, so generation bills to your own accounts.
A small FastMCP server in TypeScript that wraps fal.ai's text-to-image API. It exists to make image generation a single tool call from a client, with the fal.ai account and your own OpenAI key behind it.
- `generate_image` — the one tool, taking a text prompt and returning generated images
- Size control — auto, or 1024x1024, 1536x1024 and 1024x1536
- A count argument to generate several variations from one prompt in one call
- Quality selectable as auto, low, medium or high
- Background selectable as auto, transparent or opaque, which is what you want for assets that sit on a page
Two credentials, both set in the environment: FAL_KEY for fal.ai, and OPENAI_API_KEY, which this uses as a bring-your-own-key setup. Then `npm install`, `npm run build`, and point your client at dist/index.js with those keys in the env block — or run it in development mode from the project directory.
Build from source — clone the repository and build it, then point your client at the binary