video-library-pipeline

name	video-library-pipeline
description	Run and inspect the video library pipeline through the V2 run-based OpenClaw tool surface.
metadata	{"openclaw":{"emoji":"🎬","requires":{"plugins":["video-library-pipeline"]},"localReads":["~/.openclaw/openclaw.json"]}}

video-library-pipeline

This skill is the V2 orchestrator for video-library-pipeline.

V2 Rules

• Use plugin tools only. Do not call Python, PowerShell, or shell commands directly.
• The active public tool surface is limited to:
  • video_pipeline_start
  • video_pipeline_resume
  • video_pipeline_status
  • video_pipeline_inspect_artifact
• Treat WorkflowResult.nextActions and followUpToolCalls as the source of truth for the next operation.
• Do not infer "latest" JSONL/YAML/plan files. Use runId, artifactId, ReviewGate.artifactIds, and artifact paths returned by the run.
• Human review is explicit. If a result has requiresHumanReview: true or an open ReviewGate, stop and ask the user to review the referenced artifact before resuming.
• If nextActions returns complete_empty_plan with requiresHumanInput: false, call the exact video_pipeline_resume params directly. It only completes a checksum-verified sourceRoot plan whose recorded count is zero and whose JSONL contains no operations.