libtv-skills

Why we included this project

An agent that can chat about images is common; one that can actually produce them needs an API behind it, and that is the gap this project fills. It ships as a single skill following the OpenClaw skill specification, so agent platforms that support that format can load it and call it directly. The skill manages the routine work itself: creating a session, sending the generation prompt, uploading reference files, polling for progress, and pulling the finished images or videos. It is a thin client rather than a generator, so the actual image or video is created on LibLib.tv's servers behind your access key, and the agent stays in the role of a messenger, forwarding the user's prompt and bringing back the result. If your agents are expected to hand back real media, this is the integration work you would otherwise write by hand.