编程

Cheap

试用

Find the cheapest meaningful visible price for a product across major Chinese shopping platforms

它能做什么

Find the cheapest meaningful visible price for a product across major Chinese shopping platforms. Use when the user provides a product name and wants the lowest current visible price, the cheapest platform, or a quick cross-platform price comparison. Ask questions like "哪个最便宜", "哪里买最便宜", "帮我比价", "全…

技能文档

Cheap

Find the cheapest meaningful visible price for a product across major Chinese shopping platforms. Use when the user provides a product name and wants the lowest current visible price, the cheapest platform, or a quick cross-platform price comparison. Ask questions like "哪个最便宜", "哪里买最便宜", "帮我比价", "全网找最低价", "哪个平台便宜", "哪里下单最划算", or "帮我找最低价".

Usage Scenarios

Scenario 1: Cross-Platform Price Comparison

User input: "帮我比价小米手环 8 Pro,哪个平台最便宜?" Expected output: The skill searches across Taobao, Tmall, JD.com, Pinduoduo, and other relevant platforms. Compares like-for-like listings (same model, specs, bundle). Returns: Cheapest Found (platform and price), Comparable Options with prices on other platforms, Match Confidence (whether listings look like the same product), Caveats (variant match, coupon pricing, shipping), and Final Advice.

Scenario 2: Ambiguous Product Name

User input: "哪里买 iPhone 最便宜?" Expected output: The skill identifies the product as too broad (iPhones have many models), asks one short clarifying question: "您具体想看哪个型号?比如 iPhone 15 Pro, iPhone 15, 还是 iPhone 14?不同型号价格差异很大。" Then proceeds with the comparison once clarified.

Scenario 3: Price-Limited Mode (Weak Results)

User input: "帮我找这款小众耳机的全网最低价,只有一个淘宝店在卖" Expected output: The skill attempts cross-platform search. If only one platform has this product (Price-limited mode), switches to Mode 2. Outputs: Current Best Signal (the single available price), Why No Strong Cheapest Claim (product too niche for cross-platform comparison), What Remains Unverified (whether other platforms may have it at different prices), and Final Advice with cautious conclusion.

Scenario 4: 想买便宜机票

User input: "想去成都玩几天,从深圳出发,怎么买机票最便宜?什么时候买最划算?" Expected output: 推荐比价策略:同时查去哪儿、携程、飞猪、航司官网(如南航、深航)的价格,注意含税总价。建议提前3-4周购票,选中周三/周四出发。使用各平台的'低价日历'和'预约提醒'功能。给出深圳到成都的近期待票价区间参考,并建议关注航司会员日(如南航每月28号)。

Workflow

  1. Identify the product.

    • Accept a product name.
    • If the product is too broad or ambiguous, ask one short clarifying question for the most decision-relevant detail, such as brand, model, size, or key variant.
  2. Check price accessibility.

    • Search across relevant Chinese platforms.
    • Decide which mode applies:
      • Mode 1 if enough visible price evidence is accessible and comparable.
      • Mode 2 if platform access, listing clarity, or price conditions are too weak for a strong cheapest-price claim.
  3. If Mode 1, compare meaningful prices.

    • Prefer Taobao, Tmall, JD.com, Pinduoduo, and other clearly relevant platforms.
    • Use only currently visible price evidence.
    • Match comparable listings by brand, model, specs, quantity, version, and bundle contents.
    • Distinguish straightforward visible price from coupon-only, member-only, deposit/pre-sale, bundle-distorted, or variant-mismatched prices.
  4. If Mode 2, do not fake precision.

    • State that the currently accessible evidence is too weak for a strong cheapest-price claim.
    • Explain whether the issue is platform access, weak product matching, or distorted price conditions.
    • Give only a limited conclusion when still useful.
  5. Give the result. Cover:

    • cheapest platform or listing found when supportable
    • visible price evidence actually available
    • comparable alternatives when supportable
    • matching confidence
    • any caveats about variants, bundles, coupons, seller trust, or access limits

Output

Use this structure unless the user asks for something else.

Mode 1 output

Use when enough price evidence is accessible.

Cheapest Found

State the lowest meaningful visible price found and on which platform.

Comparable Options

List the main comparable prices found on other platforms.

Match Confidence

State whether the compared listings look like the same product:

  • High
  • Medium
  • Low

Caveats

State any important risks, such as:

  • unclear variant match
  • bundle differences
  • coupon or membership pricing
  • deposit or pre-sale pricing
  • seller quality concerns
  • incomplete platform access

Final Advice

Give a direct buying suggestion in plain language.

Mode 2 output

Use when accessible evidence is too weak for a strong lowest-price claim.

Current Best Signal

State the weakest still-usable price clue, if any.

Why No Strong Cheapest Claim

Explain whether the issue is:

  • platform access limits
  • weak listing comparability
  • unclear price conditions
  • incomplete platform coverage

What Remains Unverified

State what cannot yet be confirmed.

Final Advice

Give a cautious conclusion without pretending a definitive lowest price was found.

Quality bar

Do:

  • compare like-for-like listings
  • say when the cheapest visible option may not be the best option
  • distinguish real low price from misleading low price
  • be explicit when matching confidence is weak

Do not:

  • compare different variants as if they were the same item
  • treat coupon-only or pre-sale prices as normal prices without warning
  • claim full-market coverage when platform access is incomplete
  • invent prices or listing details

Limitation handling

If the product is too ambiguous:

  • ask one short clarifying question

If platform access is incomplete:

  • switch to Mode 2 unless a narrower Mode 1 comparison is still supportable
  • say which platforms were actually checked
  • avoid claiming a definitive all-platform lowest price
  • do not present a partial platform scan as a full-market cheapest-price result

If listing comparability is weak:

  • lower confidence
  • explain what may not match cleanly

If the visible lowest number depends on unclear conditions:

  • do not present it as the straightforward cheapest option without warning

相关技能

把自然语言描述转为结构化 JSON,并由 mcp-diagram-generator MCP 服务生成 Draw.io、Mermaid 或 Excalidraw 图表文件。

作者 nssa.io1.0k 次安装47 星标

从 AdMapix API 拉取广告创意、应用、榜单和收入预估等数据,原样返回结构化 JSON。

作者 fly0pants4.3k 次安装296 星标

以 AI 机器人身份加入视频会议,提供语音、虚拟形象与屏幕共享四种模式。

作者 johnpatternai22 次安装8 星标

诊断生产力系统反复失效的根因,给出最小干预——容量测算、瓶颈定位、可靠的本地记录。

作者 Iván855 次安装69 星标

在本地磁盘以分类纯 Markdown 文件保存需要长期留存的事实,与智能体内置记忆并存。

作者 Iván559 次安装18 星标

harrylabsj 的更多技能

浏览全部技能

淘宝与京东、拼多多、唯品会同款比价,输出可执行的购买建议。

作者 harrylabsj52 次安装2 星标

根据美团可见的配送费、包装费、满减门槛、ETA、评价等公开信息,给出一个明确的下单、换店、凑单、再等等或不点的判断。

作者 harrylabsj44 次安装1 星标

在下单前拆解拼多多低价:百亿补贴、拼团门槛、店铺风险、退货条款一次过。

作者 harrylabsj34 次安装1 星标

定位技能瓶颈,把练习日志转化为可执行的下一次训练方案。

作者 harrylabsj35 次安装

用通俗中文解读常规体检和化验报告,标注异常、分清轻重、给出下一步动作。

作者 harrylabsj27 次安装

为盒马鲜生购物规划清单、评估鲜度、推荐配送时段,并整理手动下单步骤。

作者 harrylabsj23 次安装