SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsdreamina-canvas-manage-timeline

dreamina-canvas-manage-timeline

Agents & models

View repository

Updated 10-15 21:47

About this skill

Dreamina Canvas Timeline

This Skill owns the Canvas timeline node. A timeline has two tracks: a visual track (--clip) and an audio track (--audio-clip). The most common surprise is that passing either flag rebuilds the corresponding track, throwing away any unsaved work and reissuing clip identities.

Reference detail is in references/timeline-contract.md.

When to use

  • The caller wants to assemble a video with both visual and audio tracks on the canvas.
  • The caller wants to swap an individual clip with explicit trim / speed / volume options.
  • The caller wants a metadata-only edit (title / tags) without touching the tracks.

When not to use

  • For video generation parameters — that is dreamina-canvas-generate-video.
  • For audio generation parameters — that is dreamina-canvas-generate-audio.
  • For approval or paid execution — that is dreamina-canvas-quote-and-run.

Create

dreamina-canvas --format json node create timeline --title "<title>" \
  --clip <visualNodeId1> \
  --clip <visualNodeId2>,trim=1000-4000,speed=2 \
  --clip <imageNodeId>,duration=3000 \
  --audio-clip <musicResourceId>,start=0,volume=0.5
  • Either --clip or --audio-clip (or both) must be supplied on creation.
  • An audio-only timeline is valid; the server adds an empty visual track for the canvas to remain consistent.

Clip options

--clip and --audio-clip accept a comma-separated option list:

Option Applies to Unit Default
duration=<ms> image clips milliseconds canvas default image duration
trim=<start>-<len> video / audio milliseconds full source duration
speed=<factor> video / audio multiplier 1
volume=<0-1> / muted video / audio boolean / fraction full
start=<ms> audio clips milliseconds 0

Omitted options are left empty and the server fills them in from the source asset. Never guess duration or trim locally.

Track replacement is destructive

This is the rule that trips every first-time caller:

Supplying either --clip or --audio-clip to node edit timeline rebuilds the corresponding track. The server deletes the existing clips, re-creates them per the supplied list, and issues new clip identities. Any unsaved per-clip edits the user had on the original timeline are lost.

Before calling node edit timeline with either flag, the Skill must:

  1. Confirm the user wants a full track replacement (not metadata-only).
  2. Confirm the user accepts that clip identities will change.
  3. Build the complete new list for the affected track; do not attempt to "add a clip" by mixing old and new.

If only the title / tags / description needs to change, do not pass either --clip or --audio-clip.

Metadata-only edit

dreamina-canvas --format json node edit timeline \
  --node-id <nodeId> \
  --title "<new title>"

Sparse update; the tracks are untouched.

Clear tracks

dreamina-canvas --format json node edit timeline \
  --node-id <nodeId> \
  --clear-tracks

Mutually exclusive with --clip and --audio-clip.

Reference syntax on the timeline

  • --clip and --audio-clip take bare node IDs (node_xxx) or lowercase canonical UUIDs for resources. Do not use the node: or res: prefix — those return cli.invalid_resource_reference (exit 2).

Paid execution

Hand off to dreamina-canvas-quote-and-run. This Skill never calls --run.

Failure → recovery

Failure requiredAction What to do next
node_xxx rejected on --clip / --audio-clip none Verify the bare ID; do not add node: prefix.
Two frames disagree on ratio none Confirm trim / duration units are in milliseconds.
--clip / --audio-clip accidentally included in a metadata-only edit human_intervention Surface to the user that the track was rebuilt; this is irreversible from the client.

What this Skill will not do

  • Approve spend.
  • Persist tokens, signed URLs, cookies, or session material.
  • Pass node: or res: prefix on --clip / --audio-clip.
  • Silently merge old and new clip lists; track replacement is destructive and the user must consent.
  • Guess missing options locally; let the server fill them from the source.

Policy

Invocation requires:

  • Confirmed installation of dreamina-canvas
  • Explicit user consent for any track replacement
  • For paid execution: explicit handoff to dreamina-canvas-quote-and-run

Forbids:

  • Calling --run from this Skill
  • Passing node: or res: prefix on --clip or --audio-clip
  • Silently merging old and new clip lists
  • Persisting tokens, signed URLs, cookies, or session material

Default prompt:

Passing --clip or --audio-clip to node edit timeline rebuilds the corresponding track and issues new clip identities. Confirm with the user before destructive edits; for title-only changes omit both flags. Hand off paid execution.