AgentSkillsCN

video-library-pipeline-normalize-review

交互式运行的第一阶段：执行归一化/库存管理，并在提取前暂停以待人工审核。

SKILL.md
--- frontmatter
name: video-library-pipeline-normalize-review
description: Stage 1 of interactive operation. Run normalization/inventory and stop for human review before extraction.
metadata: {"openclaw":{"emoji":"🧹","requires":{"plugins":["video-library-pipeline"]}}}

Stage 1: Normalize + Human Review

Rule

  • •Use plugin tools only. Do not call Python/PowerShell scripts directly.
  • •Keep execution in the main agent turn. Do not use subagents.
  • •This stage ends with human review. Do not continue to extraction automatically.

Tool sequence

  1. •Call video_pipeline_validate with {"checkWindowsInterop": true}.
  2. •Call video_pipeline_analyze_and_move_videos with:
    • •apply=false
    • •allowNeedsReview=false
    • •optional maxFilesPerRun (default plugin value is acceptable)
  3. •Parse summary JSON in tool result stdout and extract:
    • •inventory
    • •queue
    • •plan
    • •plan_stats
  4. •Call video_pipeline_logs with {"kind":"all","tail":50}.

Human review checklist

  • •Inventory path exists and points to the expected run.
  • •Filename normalization result is acceptable.
  • •plan_stats.skipped_outside == 0 (or user explicitly accepts reason).
  • •No non-zero exit code / no runtime error.

Handoff

  • •Report inventory and queue pointers to the user.
  • •Ask user whether to proceed to Stage 2 (Extraction + YAML review).