Show HN: Choreo – an animation DSL that verifies its own rendered output (github.com)

Choreo is a new domain-specific language (DSL) for web animation designed to empower AI agents to create animations that are verifiable against specified conditions. Unlike existing tools like GSAP and Lottie, which focus on human-authored animations, Choreo allows for a semantic description of animations that automatically compiles into self-contained SVG and CSS. A key innovation of Choreo is its built-in verifier that ensures the rendered output meets its defined spatio-temporal checks, such as ensuring animations occur in the correct sequence and timing. This development is significant for the AI/ML community because it enhances the reliability of animations generated by large language models (LLMs), improving their accuracy from about 59% to approximately 94% when a formal verify-and-correct loop is implemented. Choreo serves as both a substrate for generating animations and a mechanism for validating them, filling a crucial gap that traditional visual regression testing tools do not address. The project, while still in research stages and not intended as a supported product, showcases the potential for automated animation creation that's both reliable and adaptable for future AI applications.