Codex Plan Local Video Shots from Transcript Timestamps
Turn existing timings into a render-ready shot plan that accounts for orientation, caption space, and asset readiness.
中文Install in Codex
Provide a local `timestamps.json`, owned footage inventory, and target orientation. You can start with a planning-only exercise.
- Copy the pinned `experiments/agent-evals/news-lane/codex-project/.content-machine/skills/timestamps-to-visuals/` directory into `.agents/skills/timestamps-to-visuals/` in the project.
- Give Codex the timing input and plan output path and constrain the job to the supplied local clips.
Example task
Create a draft `visuals.json` for a fictional 12-second vertical tutorial with two timing segments and two uncaptioned clips. Note cut points and caption-safe space.What to check
Both segments map to supplied footage, the plan describes vertical framing and caption clearance, and the timing file stays intact.
Original SKILL.md on GitHub · Updated
Frequently asked question
Can I start with a manual footage list?
Yes. Include filename, duration, orientation, and whether captions are burned in so the plan can validate those constraints.