edits assets the user already has into one finished video: clips in sequence, transitions between them, text and logo overlays, and layered audio. prices a render without rendering, and polls it.

These three are named directly rather than discovered because they are first-party Layer tools rather than models, with no competing alternative in the catalogue.

This is an assembly step, not a generation step. It composites assets that already exist. Generating the clips is , and the shots should be approved before any of them reach a timeline.

If a sibling skill named here is missing from your available skills, ask the user to install it (); unattended, proceed from tool schemas and flag the gap.

Before assembling, ask whether one video model could render the whole piece. A model that produces a multi-shot sequence in one generation holds its own continuity across the cuts, which assembly cannot do. Assembly is right when the shots must be individually approved, when they come from different sources, or when the piece is longer than any model will render.

Every clip, image, and audio file must already be in the workspace and referenced by . Generated assets carry one, so they feed the timeline directly. Local files upload first.

Two things to settle before the first render, because both are expensive to fix afterwards:

Shot length carries pace. A montage that holds every shot for the same three seconds reads as a slideshow, so vary it: short shots for energy, longer holds where the viewer must read something.

Transitions are a decision, not decoration. A straight cut is right most of the time. Use a dissolve for a time or location change, and keep any one piece to a single transition vocabulary, because mixing wipes, dissolves, and fades reads as unfinished rather than varied.

For a piece with a CTA, hold the end card long enough to read twice at the viewer's pace, not the editor's.

Text overlays must survive the platform. Keep titles and CTAs inside the safe area, well clear of the edges, since social players crop and overlay their own UI at top and bottom. Set type against a plate, a scrim, or a deliberately quiet part of the frame rather than over busy motion.

Generated video is the worst possible background for small text. Where the deliverable has more than a few words, favour a held frame or a plain plate behind them.

"Cut these five approved shots into a 20-second vertical ad with our logo and a CTA."