capture_breakdown_frame

Grab a still from a play session's stored video at a time (seconds) and attach it to a moment as an immutable image asset. Supply momentId (uses the moment time unless time is given) or sourceId + time to create a new moment there. Returns the receipt plus momentId and frame. The session video must be ready. Details: /docs/mcp/capture_breakdown_frame.

Grab a still from the stored session video. Give momentId to attach at that moment's time (or an explicit time), or sourceId (a play session) + time to create a moment there; height (180–1080, default 720) sets the JPEG size. The frame is stored once as an immutable asset and referenced from the moment; the same requestId replays the same frame ID. Use it when the transcript mentions something visual (a screen, menu, HUD, enemy) so a finding can cite both words and picture. Read the picture with get_asset_downloads or render nothing: frames are ordinary image assets.