{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/9e577f60de36481488a8d89e1f62ea09\" frameborder=\"0\" width=\"1920\" height=\"1440\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1440,"width":1920,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1440,"thumbnail_width":1920,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/9e577f60de36481488a8d89e1f62ea09-8b2e6a775a7cb3a9.gif","duration":156.541,"title":"Rendering Skills","description":"This Loom explains the problem with traditional AI skills and introduces “rendered skills” to make them more portable. The author describes how skills often hardcode deterministic tool-calling methodology, such as making multiple calls to a u.com API to “search for the latest market context,” which increases token usage and depends on the machine having the right tools. Rendered skills move those deterministic steps to an external compute runtime like Render and then return a fully rendered skill that can load the results into the agent’s context. The example shows executing a template rendering string with the u.com API and using the returned results to support subsequent agent work."}