--- name: ketchup-video description: /ketchup as short subtitled slide video; waiting calls under it. Use for /ketchup-video. disable-model-invocation: true --- # Ketchup video Follow `/ketchup`; swap summary+visuals for one short subtitled landscape video. Window = last user message before `/ketchup-video`, not earlier `/ketchup`. Video write/render outside the repo is the only read-only exception. Reply: ketchup opening sentence, embed, leftover self-doable line, action items. Match video option names. Render fail → one line + calls only. ## Build Artifacts only, not chat prose. Draft takeaway + one line/slide; drop non-teaching; ≤10. One point/slide; evolve one scene with stable legend. No lists/tables/disconnected boxes. Labels on named things; no captions; no fact twice. Landscape mp4+, no audio; motion between slides; edge-to-edge frames—not screen-recorded. Subtitle strip below each slide: ~10 words, the slide's point in your voice, burned in, never covering it. Hold (fully on screen, excl. motion) ≥1s/3 new words incl. subtitle (skip shot text). Pre-embed: check one frame per slide at chat size; measure holds by matching frames, not a still detector; fix until readable.