What problem does it solve? Running multimodal AI models for image generation, video creation, speech synthesis, transcription, OCR, and image editing normally requires managing GPUs or integrating many separate APIs. This Skill wraps Replicate's pay-per-use model catalog in ready-to-run Node.js scripts with sensible defaults, model selection guidance, and prompting rules. ## Core Features & Use Cases - Media Generation: Create images (FLUX, nano-banana, gpt-image, ideogram), SVG vectors (recraft), video (Veo, Kling, Runway), music (MusicGen), and text-to-speech (Kokoro, ElevenLabs) through simple CLI commands. - Editing & Processing: Upscale images, remove backgrounds, restore faces, inpaint regions, run OCR, and transcribe audio with speaker diarization. - Model Selection Guidance: Decision tables map each task to the right model with cost estimates, plus detailed prompting guides per model family. - Use Case: A marketer needs a product photo with readable label text. Run gen-image.mjs with the nano-banana model, which renders text reliably, then refine it with conversational editing instead of re-rolling. ## Quick Start Generate an image by asking the agent to run gen-image.mjs with your prompt and save the result to a file, making sure REPLICATE_API_TOKEN is set in .env first.
