diffusionstudio/lottie

Text-to-Lottie is an open-source tool that lets you generate polished, production-ready Lottie animations by describing what you want in plain language to an AI coding assistant like Claude Code or Codex. Lottie animations are lightweight, scalable animated graphics used widely in apps and websites for things like loading spinners, onboarding flows, and icons. Instead of manually animating in a tool like After Effects, you describe the motion you want, and the AI builds the animation file for you. To use it, you install a "skill" package into your coding agent with a single command. Then you prompt the agent with a description of the animation you want, for example, asking it to reveal an SVG path with a gradient, use ease-in-out timing, and run at a specific frame rate. The agent sets up a workspace with a built-in player that loads each animation as a "scene" within a project. As the agent works, the player live-updates so you can scrub through the timeline, inspect the result, and refine it in real time. This tool is built for designers, developers, product managers, or anyone who needs animated graphics but doesn't want to hand-animate every frame. A startup founder could use it to quickly generate a loading animation for their app. A product designer could prototype a micro-interaction by describing the motion they have in mind. The results come out as standard Lottie JSON files, which work across web, iOS, Android, React Native, and Flutter, and can also be imported into After Effects for further tweaking. The README offers practical guidance on getting good results: ground your prompts with concrete assets like SVGs or screenshots, use motion design terminology like "ease-in" and "ease-out," and think like a camera operator by describing pans, zooms, and pushes. You can also request specific controls, frame rates, and durations. By default, outputs expose a background color control, but you can ask the agent to add more customization options.