What problem does it solve? Your agent harness may not be able to see images or generate media with a specific model. This Skill fills both gaps: it reads images through an OpenRouter vision fallback and generates images, video, or other media through the platform's metered image API backed by fal.ai, without exposing any credentials. ## Core Features & Use Cases - Image Reading Fallback: Sends base64-encoded images to an OpenRouter vision model when the harness cannot render images natively, with guidance on model selection, downscaling, and argv-size limits. - Metered Media Generation: Runs tribes-image generate with any fal.ai model id (image, video, upscaler, img2img) in a single blocking call that returns the result JSON directly. - Use Case: A user asks for an oil-painting-style lighthouse image using fal-ai/flux/schnell. The Skill submits the prompt via tribes-image generate, downloads the resulting CDN URL to a local file before it expires, and reports the saved path. ## Quick Start Ask the agent to read the contents of a screenshot it cannot see, or to generate an image with a named fal.ai model and save the result to a local file.
