dogaga.pages.dev

get_project_state

Read DOGAGA's current agent-safe editor state, including the Project Brief / destination, any current humanDemonstration captured through Teach by Example, any current reviewable edit plan, visual/audio tracks, canvas, loaded Asset IDs (video, image, or audio), clips, transitions, and playhead. Local File handles and object URLs are never returned. If humanDemonstration.status is ready, treat its semantic changes as a user-provided editing example: when the human asks to do the same or apply a similar treatment, infer analogous target assets or clips from the live state and use propose_edit_plan for multi-step application so the human can review it. A demonstrated add_visual_clip means the human intentionally added a loaded visual asset to a track with the captured duration/speed/fades; generalize that treatment to analogous loaded assets rather than replaying the original asset ID. When projectBrief.destination or projectBrief.goal is specific, compare that goal with the live state and proactively point out useful improvements. Do not automatically replay a human demonstration without review.

Input schema

No input properties (empty object).

{
  "type": "object",
  "properties": {},
  "additionalProperties": false
}