Install
openclaw skills install @harrylabsj/review-buying-advisorAnalyze a product from name plus platform, read public reviews, and turn evidence into a practical buying recommendation
openclaw skills install @harrylabsj/review-buying-advisorAnalyze a product from name plus platform, read public reviews, and turn evidence into a practical buying recommendation. Use when the user asks whether a product is worth buying on Tmall, Taobao, JD.com, Pinduoduo, Amazon, Best Buy, or similar platforms; asks for help reading reviews; wants a real user-feedback summary; wants recurring complaints, hidden risks, pros and cons, buying advice, who the product is suitable for, or how to avoid buying mistakes before ordering. Especially relevant for requests like "值不值得买", "帮我看评论", "看看口碑", "有什么坑", "适不适合买", or "帮我避坑".
User input: "帮我看看这款蓝牙耳机的口碑,在京东上,值得买吗?" Expected output: The skill identifies the product and platform, determines review accessibility (Mode 1), samples a representative set of public reviews (positive, negative, mixed, recent), groups signals into themes (sound quality, battery life, comfort, connectivity issues), identifies recurring complaints and their severity, and outputs: Overall Verdict (Recommend / Not Recommended), Best For, Main Positives, Main Risks, Watch Before Buying, Final Advice, and Confidence level.
User input: "这个洗发水在小红书上口碑怎么样?" Expected output: The skill attempts to access reviews. If the platform blocks or weakens review content, switches to Mode 2. Outputs: Not enough evidence for a strong recommendation, explains the limitation (platform blocking review access), gives limited conclusion with low confidence, and advises what the buyer should verify independently before purchasing.
User input: "我想买这个扫地机器人,但怕有坑,帮我看看有什么常见问题" Expected output: The skill searches for the product on the platform, samples reviews looking specifically for recurring complaints and failure modes, separates repeated issues from isolated complaints, weighs severity as well as frequency, and outputs: common issues (laser navigation fails after 6 months, water tank leaks), severity assessment (medium - some users report partial refund success), and buyer fit recommendations (best for homes without thick carpets, avoid if you have mostly dark floors).
User input: "我想买个扫地机器人,预算2000以内,在京东看了追觅和石头的小米款,评论区有说好的有说坏的不敢下手。" Expected output: 从海量评价中提取有效信息:追踪追觅和石头在该价位段各型号的差评关键词(如'容易卡住'、'噪音大'、'APP总掉线');比较30天内好评率和追评内容;参考抖音/b站up主实测视频的结论。建议在京东自营购买利用7天无理由,并注意哪些店铺支持上门售后。
Identify the product.
Check review accessibility.
If Mode 1, analyze review evidence.
If Mode 2, do not fake completion.
Give the final answer. Cover:
Use this structure unless the user asks for something else.
Use when public review evidence is available.
Choose one:
2-4 bullets with the strongest evidence.
Who is most likely to be satisfied.
Most credible repeated strengths.
Most important risks, including repeated issues or severe but less frequent ones.
What the buyer should verify before ordering.
A direct recommendation in plain language.
High / Medium / Low, with a brief reason.
Use when public review evidence is not sufficiently accessible.
Usually:
Explain what was and was not accessible.
State whether the issue is platform blocking, weak public text, ambiguous product matching, or incomplete review visibility.
State what remains unverified.
Give a cautious conclusion without pretending the reviews were validated.
Usually Low.
Do:
Do not:
If the product cannot be identified confidently:
If the platform does not expose enough public review content:
If public review evidence is sparse or mixed but still partially usable:
If reviews may mix multiple variants: