Video Sourcing

Video Sourcing 技能

Use this 技能 when the user asks to find, compare, or analyze social videos (YouTube, TikTok, Instagram, Twitter/X), or explicitly invokes /video_sourcing.

This 工作流 expects host 运行time execution (sandbox mode off). The 运行器 auto-bootstraps a pinned 运行time from Memories-AI-labs/video-sourcing-代理@v0.2.5 when VIDEO_SOURCING_代理_ROOT is not 设置.

Triggering

运行 this 工作流 when either condition is true:

Message 启动s with /video_sourcing. The user asks for video sourcing/trend/创建器/brand analysis and wants concrete video links.

If /video_sourcing is used with no 查询 body, ask for the missing 查询.

Execution contract Resolve 查询 text: /video_sourcing ... => strip /video_sourcing and use remAIning text. Free-form => use user message as 查询. Default to compact mode: --event-detAIl compact If user asks for 调试ging/raw payloads: Switch to --event-detAIl verbose /video_sourcing deterministic path Build command with required args: <技能_dir>/scripts/运行_video_查询.sh --查询 "<查询>" --event-detAIl --ux-mode three_message --进度-gate-seconds 5 启动 with exec using background: true and explicit timeout: timeout: 420 Poll with process using action: "poll" every 2-4 seconds until process exits. 解析 NDJSON 输出 and render only these 事件, using your persona voice for all user-facing text: 启动ed => 发送 a brief message conveying that video sourcing has begun, written in your persona voice ux_进度 => read the summary field as structured 状态 data and compose a concise, natural 进度 更新 in your persona voice (do not echo summary verbatim) 发送 each ux_进度 as a separate 助手 message in Telegram. terminal event (complete, clarification_needed, error) => 发送 final message as-is Do not forward raw 进度, 工具_call, or 工具_结果 事件 for /video_sourcing. Preserve the user's existing OpenClaw personality behavior across all messages — 进度 and final alike. Never launch a second 运行 while a prior 运行 会话 is still active. If retrying, call process with action: "kill" for the prior 会话Id first. If process exits without a terminal event, 发送 a concise fallback message that: 状态s 运行 ended before completion, includes one actionable next step, does not 启动 another 运行 automatically.

Behavior tar获取 for /video_sourcing:

Fast 运行 (<5s): 2 messages (启动ed, terminal). Longer 运行 (>=5s): recurring throttled ux_进度 更新s, then terminal. Free-form path (non-strict) Keep existing flexible behavior. Build command without forcing three_message mode: <技能_dir>/scripts/运行_video_查询.sh --查询 "<查询>" --event-detAIl 流 useful 进度 更新s and final 响应 naturally. Final 响应 格式化

When terminal event is complete:

One short paragraph conclusion. Top 3 video references only by default: title url one-line relevance note 工具s used: ... with a compact 状态 summary.

If fewer than 3 videos exist, show all avAIlable references.

When terminal event is clarification_needed:

Ask the clarification question directly. Treat this as the final 响应 for the current 运行.

When terminal event is error:

发送 concise 失败 reason. Include one actionable next step. Safety and fallback If script fAIls due to missing env/工具ing, explAIn exact missing piece (for example VIDEO_SOURCING_代理_ROOT, uv, or API key env var). If VIDEO_SOURCING_代理_ROOT is un设置, the 运行器 uses managed path: ~/.OpenClaw/data/video-sourcing-代理/v0.2.5 VIDEO_SOURCING_代理_ROOT remAIns an advanced override for local development. Keep 响应 concise and action-oriented. Never fabricate video URLs or 指标.