Egaki

egaki is a command line tool for AI media generation and a framework for writing video in code. One command turns a text prompt into an image, a video clip or a spoken audio file. A companion framework lets scenes be written in MDX, where each heading marks a timed section and animation components handle the motion, and the finished sequence is rendered to MP4 in the browser through WebCodecs. It is aimed at developers and motion designers who would rather express a video as source files than assemble it in a timeline. The command line covers image generation, editing and inpainting, text to video and image to video work, clip editing and extension, speech synthesis, voice cloning, transcription with word timestamps, audio stem separation, tempo detection and loudness analysis. The video framework supplies primitives for opacity, translation and scale, easing and spring curves, layout transitions, captions, server components, shader effects and a bridge to Framer Motion. Requests are routed to providers including Google, OpenAI, xAI, Fal.ai, Replicate, ElevenLabs, Cartesia, Deepgram and Groq, authenticated either with your own provider keys or with a subscription. Two subscription plans are published: Plus at $29 per month and Pro at $99 per month. Each includes a monthly budget equal to its price, and generations draw that budget down at the provider cost plus a markup. Supplying your own provider keys avoids the subscription and bills you directly from the provider. The model catalog lists approximate rates per unit of output, for example $0.02 to $0.06 for an image on some models and $15 to $100 per million characters for speech. The project describes itself as being in early release and warns that both the command interface and the MDX syntax will change, with breaking changes and migration steps recorded in a changelog.