| name | lightflow-auto-edit |
| description | This skill should be used when the user asks to run or change LightFlow automatic video editing with lightflow.video_auto_edit or lightflow.video_auto_edit_plan. |
| version | 0.4.0 |
LightFlow Automatic Video Editing
Run lightflow.video_highlights with VideoScore before planning or rendering.
Pass its clips output unchanged to lightflow.video_auto_edit_plan or
lightflow.video_auto_edit.
Set the same runtime-only LIGHTFLOW_VIDEOSCORE_EVIDENCE_KEY for highlight
generation and automatic editing. Require every clip object to contain path
and highlight. Keep any explicit
start / end range. Require highlight.source_path to resolve to the same
source, highlight.start_seconds / end_seconds to match the selected range
within 0.001 seconds, score to be numeric from 1 through 4, and non-empty
model and reason values. Require workflow to be
lightflow.video_highlights, model to be TIGER-Lab/VideoScore-v1.1, and
evidence to be the lowercase-hex HMAC-SHA256 tag generated by
lightflow.video_highlights. Do not pass string-only sources or invent
highlights from scene, silence, filename, or transcript heuristics.
Use the package-owned Rust runner.v1 path through lfw run; do not invoke
repository scripts directly. Update this skill whenever the clip provenance
contract or common workflow invocation changes.
API Usage
curl -sS -X POST http://127.0.0.1:5174/workflows/lightflow.video_auto_edit/run \
-H 'content-type: application/json' \
-d '{"inputs":{"sources":[{"id":"clip-1","path":"media/long-interview.mp4","start":12.0,"end":18.5,"highlight":{"workflow":"lightflow.video_highlights","source_path":"media/long-interview.mp4","start_seconds":12.0,"end_seconds":18.5,"aggregate_score":3.2,"model":"TIGER-Lab/VideoScore-v1.1","reason":"clear walk-around","evidence":"<hmac-sha256-hex>"}}],"brief":"Energetic highlight cut.","output_path":"output/auto-edit.mp4"}}'
