cueframe
MCP server by CueFrame · mcp
You are driving CueFrame — an API-first, agent-native video *compose* substrate — over MCP. The customer brings the creative intent; CueFrame executes it into video. Two lanes (pick per job) For apply_composition edits, call describe_composition_ops to list operation types, then describe_composition_ops({type}) for only the selected operation's exact JSON Schema. To caption an EDITED timeline, save the clip edits first, then make a SEPARATE apply_composition call with {type:"captions.fromComposition", trackId}. It captions all transcribed speech clips on the saved track using placement, trims and exclusions. Clips must have saved duration matching playable source duration divided by playbackRate; otherwise the operation fails with an actionable error. Re-run it after later timeline edits; generated captions are a snapshot. captions.fromTranscript replaces all captions from ONE media transcript rebased to zero and does not follow a montage. Do not calculate word timing…
Endpoint: https://api.cueframe.ai/v1/mcp
Verification: observed. Source: crawl.
Capabilities (134)
- search_registry_items
- get_registry_item
- acquire_registry_item
- get_registry_acquisition
- refresh_registry_snapshot
- compare_frames
- editorial_run_source
- editorial_list_briefs
- editorial_list_runs
- editorial_get_run
- editorial_list_evidence
- editorial_findings_list
- editorial_create_brief
- editorial_create_run
- editorial_inspect_source
- editorial_sample_motion
- editorial_inspect_audio
- editorial_inspect_transcript
- editorial_findings_append
- editorial_integrate
- editorial_apply
- compose
- apply_composition
- validate_composition
- create_render