executing-plans
executing-plans
Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available
sealeap-bifang-amazon-change-validation-by-rankValidate each listing or advertising change on Amazon by recording the change and its previous value, sampling organic and sponsored keyword positions for the main terms before and after under the same conditions, waiting for the effect window, then keeping, revising or reverting the change based on the observed movement rather than on intuition. Use for 改了…
Independent GitHub snapshot · Automated classification · Original documentation governs
Validate each listing or advertising change on Amazon by recording the change and its previous value, sampling organic and sponsored keyword positions for the main terms before and after under the same conditions, waiting for the effect window, then keeping, revising or reverting the change based on the observed movement rather than on intuition.
用户未指定时采用“诊断”。
缺失项必须标为 NEEDS_EVIDENCE;不得猜数字、补属性或把不同站点、ASIN、变体、币种和时间窗混在一起。
先读取 references/playbook.md,确认该方法适用于当前对象。按以下顺序执行:
最后做数据充分性检查,并把结论分成 FACT / ESTIMATE / HYPOTHESIS / UNKNOWN。若关键证据不足,状态写 HOLD。
仅在自有数据不足且当前任务确实需要外部证据时,读取 references/mcp-data-plan.md,再使用 scripts/mcp_research.py。本 Skill 的外部取数目的:批量抓取主词的自然位与广告位作为前后对比的代理观测。
doctor,再 search-tools 和 describe;工具名及参数以实时 tools/list 与 inputSchema 为准。tools/call 或 Actor 可能计费;先展示 Provider、工具、无密钥业务参数、预计成本与输出位置,核对已有授权覆盖后才加 --allow-cost;该标志不是费用上限。方案状态使用 READY FOR REVIEW / DRAFT / HOLD / STOP;如已执行,另行记录实际结果及回读证据。未得到明确批准时,不得声称已修改线上对象。