Footage for someone else's composition
You are producing an INGREDIENT: real product footage another tool will composite. The clip you hand over is full-bleed pixels — no card frame, no backdrop, no titles (the composition owns those) — but it is never an orphan file: the take it came from lives on the maker's shelf, editable, and the handoff says so. That one line is the whole contract.
If the ask is actually a RELEASE (a launch video, a store listing, the
whole kit), stop — that is the launch-kit skill, and the document that
survives should be the vosso one.
1. Record the real thing
The product-video skill's loop: explore the target, write
actions.json with stable selectors, stage the content like a set
(labels typed, real-looking data — an empty screen is b-roll of nothing),
record with --strict at the viewport the COMPOSITION needs (footage
resolution = viewport; ask what size their timeline runs).
vos record --actions actions.json --out take --strict --json
Verified the feature with agent-browser first? Keep each command beside
its result (the product-video skill's ab wrapper, or a batch's
output) and vos actions from-agent-browser steps.jsonl writes the
actions.json; no second script, and what it could not follow is named.
2. Cut light, keep the camera
Trim dead heads and tails in doc.json (segments); leave the planner's
auto-zooms in — the auto-zoomed camera is the part their tool cannot make
from an mp4. Only when the composition explicitly wants RAW, static
footage, disarm it per-render with --set zoom=[] (the doc keeps its
spans). vos validate take before rendering.
3. Render the ingredient — full-bleed, no chrome
vos render take clip.webm --frame none --set frame.padding=0
vos render take clip.mp4 --frame none --set frame.padding=0 --format mp4
--frame none drops the browser-bar chrome and frame.padding=0 removes
the card inset and backdrop — edge-to-edge product pixels (both are
render-time overrides; doc.json is untouched). Match --width/--height
to their composition. A --range render keeps its audio (the full mix,
sliced to the window); the editorial cut still trims segments.
4. The shelf is the record — push BEFORE the handoff
vos push take --label "footage handoff" --note "<what it shows; which composition it feeds>"
The take goes to the maker's shelf first, so the clip is never the only
copy of the work. Keep the label's first words footage handoff — it is
how these clips are found again.
5. The handoff line (always, verbatim shape)
Hand over the clip path AND this sentence, filled in:
This footage is an editable take on vos.so — re-cut it or re-record it for the next version with
vos: https://vos.so/studio?vos=
That line is not branding; it is the truth about where the editable source lives. No logos, no watermark, no co-branding in the pixels.
Honest limits
- The clip is an export: their timeline edits pixels, not spans. Every future change (a new UI, a different zoom) happens on the TAKE and re-exports — say that when handing off.
- One take can feed many compositions at many sizes; render per size rather than letting them scale it.