{"slug":"xhs-growth-review","title":"\u5c0f\u7ea2\u4e66\u6570\u636e\u590d\u76d8","description":"\u8bca\u65ad\u5c0f\u7ea2\u4e66\u56fe\u6587\u4e0e\u89c6\u9891\u7684\u8868\u73b0\uff1a\u62c6\u66dd\u5149\u3001\u70b9\u51fb\u3001\u6536\u85cf\u4e0e\u5173\u6ce8\u8f6c\u5316\uff0c\u7ed9\u51fa\u4e0b\u4e00\u8f6e\u8981\u6539\u7684\u5177\u4f53\u52a8\u4f5c\u3002","category":"\u590d\u76d8\u4e0e\u589e\u957f","tags":["\u590d\u76d8","\u6570\u636e","\u5c0f\u7ea2\u4e66"],"author":"xintu","source_url":"https://github.com/xintu1314/xhs-creator-suite/blob/main/xhs-growth-review/SKILL.md","license":"MIT","status":"published","verification":"source_reviewed","verification_notes":"\u6765\u6e90 xintu1314/xhs-creator-suite\uff08xhs-growth-review/SKILL.md\uff09\uff0c\u672c\u5730\u6253\u5305\u4e0a\u4f20\uff1b\u81ea\u8bc4 4/5\uff5c\u6307\u6807\u62c6\u5f97\u7ec6\uff0c\u7ed3\u8bba\u843d\u5728\u300c\u4e0b\u4e00\u6279\u600e\u4e48\u6539\u300d\u4e0a\u3002 \u672a\u6267\u884c\u5305\u5185\u811a\u672c\u3002","revision":"ccb2b22ea4a6465c9f215f0bac66a8d7","created_at":"2026-10-29T05:46:17+00:00","updated_at":"2026-10-29T05:46:25+00:00","source_repo":"","source_path":"","source_commit":"","source_updated_at":"","stars":0,"proxy_score":null,"score":null,"grade":null,"dimensions":{},"evaluation":{},"installs":0,"platforms":[],"formats":[],"category_slug":"growth","category_name":"\u590d\u76d8\u4e0e\u589e\u957f","usage_guide":{},"read_url":"https://shiliuai.cn/api/v2/skills/xhs-growth-review","skill_url":"https://shiliuai.cn/api/v2/skills/xhs-growth-review/files/SKILL.md","download_url":"https://shiliuai.cn/api/v2/skills/xhs-growth-review/download","install_url":"https://shiliuai.cn/api/v2/skills/xhs-growth-review/install","files":[{"path":"SKILL.md","bytes":3047,"url":"https://shiliuai.cn/api/v2/skills/xhs-growth-review/files/SKILL.md"},{"path":"agents/openai.yaml","bytes":264,"url":"https://shiliuai.cn/api/v2/skills/xhs-growth-review/files/agents/openai.yaml"},{"path":"references/format-benchmarks.md","bytes":1878,"url":"https://shiliuai.cn/api/v2/skills/xhs-growth-review/files/references/format-benchmarks.md"},{"path":"references/metrics-schema.md","bytes":719,"url":"https://shiliuai.cn/api/v2/skills/xhs-growth-review/files/references/metrics-schema.md"},{"path":"scripts/diagnose_metrics.py","bytes":3262,"url":"https://shiliuai.cn/api/v2/skills/xhs-growth-review/files/scripts/diagnose_metrics.py"}],"content":"---\nname: xhs-growth-review\ndescription: >-\n  Diagnose Xiaohongshu carousel and video performance separately using exposure, click, retention, completion, and\n  engagement data compared to the account's own history. Produces one-variable follow-up experiments. Use for \u6d41\u91cf\u590d\u76d8,\n  \u7b14\u8bb0\u8bca\u65ad, \u89c6\u9891\u6570\u636e\u5206\u6790, \u56fe\u6587\u6570\u636e\u590d\u76d8.\n---\n\n# Xiaohongshu Growth Review\n\nTurn post data into a bounded diagnosis and a single next experiment. Read `references/format-benchmarks.md`.\n\n## Workflow\n\n1. Record `format`: `carousel`, `image`, or `video`. Do not mix formats in one baseline.\n2. Collect metrics using `references/metrics-schema.md`. Preserve missing values as missing.\n3. Require at least three comparable posts **in the same format** for group claims; prefer five or more.\n4. Run `scripts/diagnose_metrics.py` when JSON input is available.\n5. Compare target vs account median for that format and topic when possible.\n6. Diagnose earliest supported funnel break (see format-benchmarks).\n7. Route the fix to exactly one downstream skill; do not recommend multi-variable rewrites.\n8. Propose DNA updates per the promotion rules below; never edit `account-dna.md` silently.\n\n## Promoting a learning into account DNA\n\nA single post never changes the DNA. Promote only when all three hold:\n\n- the same signal repeated across **3+ posts in the same format**\n- the direction survived at least one deliberate counter-test\n- it changes a decision (pillar, format, hook style, avoid list), not just a number\n\nWhen promoting, output a diff block for the user to approve \u2014 do not write the file directly:\n\n```\nDNA \u66f4\u65b0\u63d0\u6848\uff08\u9700\u4f60\u786e\u8ba4\uff09\n\u6bb5\u843d: \u5185\u5bb9\u652f\u67f1 2\n\u73b0\u72b6: \u5de5\u5177\u6d4b\u8bc4\n\u5efa\u8bae: \u5de5\u5177\u6d4b\u8bc4\uff08\u9650\u5b9a\u4e3a\u300c\u6211\u5b9e\u9645\u7528\u6ee1\u4e00\u5468\u7684\u5de5\u5177\u300d\uff09\n\u4f9d\u636e: 3 \u7bc7\u540c\u683c\u5f0f\u7b14\u8bb0\uff0c\u672a\u6ee1\u4e00\u5468\u7684\u6d4b\u8bc4\u6536\u85cf\u4e2d\u4f4d\u6570\u4f4e 40%\n\u7f6e\u4fe1\u5ea6: medium\n```\n\nAnything below the bar goes into `Current hypotheses` with a review date instead.\n\n## Funnel stages\n\n- Distribution \u2014 exposure weak vs baseline\n- Click \u2014 exposure exists but CTR weak\n- Early retention \u2014 carousel fold / video 3s weak\n- Completion/value \u2014 video completion or carousel save value weak\n- Interaction \u2014 comments, profile visits, follows weak\n\n## Fix routing\n\n| Bottleneck | First skill to rerun |\n|---|---|\n| exposure | `xhs-account-dna` or `xhs-topic-radar` |\n| click | title refresh in `xhs-note-writer` + `xhs-visual-lab` |\n| early retention | `xhs-video-director` opening OR carousel fold rewrite in `xhs-note-writer` |\n| completion/value | tighten proof path in production skill |\n| interaction | CTA/proof/discussion angle in `xhs-note-writer` |\n\n## Rules\n\n- Never diagnose shadow banning from one weak post.\n- Do not compare raw likes across very different exposures.\n- Video-only metrics stay missing for carousel posts.\n- Distinguish facts, calculations, and hypotheses.\n\n## Output\n\nReturn evidence table, format, earliest supported bottleneck, confidence, missing data, keep/change list, one downstream skill to rerun, and next single-variable experiment.\n","usage":"\u8bfb\u53d6 SKILL.md\uff0c\u5e76\u6309\u9700\u8bfb\u53d6\u914d\u5957\u6587\u4ef6\uff1b\u4e0b\u8f7d ZIP \u83b7\u53d6\u5b8c\u6574\u5305\u3002\u6280\u80fd\u662f\u4efb\u52a1\u65b9\u6cd5\uff0c\u4e0d\u6388\u4e88\u65b0\u589e\u8d26\u6237\u3001\u4ed8\u8d39\u6216\u7cfb\u7edf\u6743\u9650\u3002"}