安全卖家工具包

安装指南

1. ClawHub CLI 从源直接安装技能的最快方式。
   npx clawhub@latest install safe-seller-toolkit
2. 手动安装 将技能文件夹复制到以下位置之一
   全局模式
   ~/.openclaw/skills/
   工作区
   /skills/
   优先级：工作区 > 本地 > 内置
3. 提示词安装 将此提示词复制到 OpenClaw 即可自动安装。
   请帮我使用 Clawhub 安装 safe-seller-toolkit。如果尚未安装 Clawhub，请先安装（npm i -g clawhub）。
   更倾向于下载？
   获取包含原生技能文件的ZIP压缩包。
   下载技能 (.zip)

什么是

安全卖家工具包通过实用的定价、买家消息、当面交易和寄送检查，帮助个人在点对点交易市场出售个人物品。其命令行工作流程会根据类别折旧和物品状况估算价格区间，根据九类诈骗模式评估消息，并给出判断结果和安全回复。
该工具包还会生成与物品价值相匹配的见面交易清单和寄送指南，其中包括明确的禁止条件和不得寄送的规则。它使用 Python 标准库运行，可离线使用，且不会记录任何信息，因此对于希望获得可行防护措施、又不想使用云服务的卖家来说，是 Openclaw Skills 中的实用工具。

应用场景

为二手手机、笔记本电脑、沙发、自行车或其他个人物品估算挂牌价、预期价格和底价。
回复前检查意外报价、索要验证码、超额付款、快递员取件或平台外付款等情况。
使用针对物品和交易金额定制的清单，为当面交易做好准备。
发货前检查寄送选项、包装规则、凭证照片和付款状态。
判断何时应屏蔽可疑买家，或在遵守特定安全规则的情况下继续交易。
为个人交易提供实用的起点；该工具包不适用于库存规模的企业、车辆所有权转让、房地产交易或诈骗发生后的平台纠纷。

工作原理

根据物品类别、状况、原价和使用年限为物品定价。脚本会返回挂牌价、预期价格和底价估算；请将其与同一具体型号的已售商品列表进行比较。
提交收到的买家消息进行筛查。该工具包会识别匹配的诈骗信号，计算 0–100 的可能性评分，给出判断结果并建议安全回复。
使用见面交易清单规划当面交易。金额较高的交易会触发额外防护措施，例如银行大厅交易规程以及 IMEI 或序列号检查。
如果选择寄送，请比较服务并遵循包装和凭证照片指南。只有付款完成后才能寄出，付款仍在处理中时绝不要寄出。
如遇明确的禁止条件，请立即停止。屏蔽诈骗话术，不要与发送者互动或争辩。

配置指南

技能文档提供了 Python 脚本接口，但未说明单独的软件包安装或配置流程。请在技能目录中使用 Python 3 运行随附脚本；脚本仅使用标准库，且可离线运行。
python3 scripts/seller_toolkit.py price --category iphone --condition good --original 799 --age-years 2
python3 scripts/seller_toolkit.py message --text "I'll send a 6-digit code to confirm"
python3 scripts/seller_toolkit.py meetup --item "iPhone 13" --price 350
python3 scripts/seller_toolkit.py labels --weight-lb 3.2 --dimensions 12x9x4 --value 350
针对每项销售任务使用相应命令。检查消息时，将买家的文本作为 --text 的值传入；不要向买家透露任何验证码。

数据架构

该工具包接受命令行参数，并针对当前任务返回指导。文档说明它不会记录任何信息；文档未描述持久化用户数据或生成的数据文件。
命令 输入 结果
price 类别、状况、原价、物品年限 挂牌价、预期价格和底价区间，以及市场行为提示
message 买家消息文本 各信号明细、0–100 的诈骗可能性评分、判断结果和安全回复
meetup 物品描述和售价 与金额相匹配的当面交易安全清单及适用的防护措施
labels 重量、尺寸和物品价值 寄送服务比较、包装指南、凭证包步骤及防诈骗注意事项
消息筛查涵盖九类诈骗：验证码钓鱼、超额付款、平台外付款、平台外寄送或快递员取件、好得令人难以置信的报价、托管或代理骗局、调包诱饵、施压或制造紧迫感，以及不看价格的买家。评分达到 40 或更高时结果为 BLOCK；15 至 39 分时结果为 PROCEED WITH RULES；低于 15 分则表示保持正常谨慎即可。参考资料包括 references/scam-field-guide.md 和 references/pricing-curves.md。

高级特性

使用九类诈骗模式和 40 多个触发短语评估收到的消息，提供信号级解释，并给出 BLOCK、PROCEED WITH RULES 或 LIKELY FINE 结果。
对 16 个类别和五个状况等级应用类别折旧曲线，生成价格区间；估算值仅为起始参考，不可替代对已售商品列表的核查。
根据交易金额调整见面交易防范措施，包括警察局交易区以及针对高价值电子产品的额外检查。
提供寄送方案比较、包装指南、凭证包拍照建议，以及明确的防诈骗禁寄规则。
使用 Python 标准库离线运行，且不会记录任何信息；文档所述工作流程不需要 hooks、云服务或多代理协作。

name: safe-seller-toolkit description: "Use when selling personal items on marketplaces (Facebook Marketplace, Craigslist, OfferUp, eBay) — prices used goods with category depreciation curves, triages inbound buyer messages for the 9 scam families (overpayment, code phishing, fake shipping, off-platform payment) with BLOCK/PROCEED verdicts and safe replies, generates meetup safety checklists scaled by deal value, and shipping checklists with do-not-ship fraud rules." version: 1.0.1
author: Denis Voronin
license: MIT
tags: [marketplace, selling, scam-prevention, pricing, safety]

Safe Seller Toolkit

Overview

Peer-to-peer marketplaces move enormous volumes of secondhand goods — and concentrate fraud. The Internet Crime Complaint Center logged billions in losses from non-delivery/non-payment fraud alone; marketplace scams (overpayment, Google-Voice code hijack, fake shipping labels, off-platform payment) are the everyday layer beneath that. Casual sellers list a phone, get four messages in an hour, and three are scripts written by professionals.

This skill gives the seller four instruments:

price — 16 category depreciation curves × 5 condition grades → list/expect/floor price band plus market-behavior notes.
message — inbound buyer text scored against 9 scam families (40+ trigger phrases) → 0-100 scam likelihood, BLOCK/PROCEED verdict, and a safe one-line reply.
meetup — value-scaled in-person safety checklist (police exchange zones, IMEI verification, bank-lobby protocol for high-value deals) plus hard no-go conditions.
labels — shipping service comparison, packaging rules, proof-pack photography, and the do-not-ship fraud list.

Pure stdlib, offline, logs nothing.

When to Use

"Someone offered to pay full price without questions — is this a scam?"
"How much should I list my 2-year-old iPhone / laptop / couch for?"
"A buyer wants me to ship to their agent / send a code / take Zelle"
Preparing a first in-person sale (meetup checklist)
"Got an offer on Marketplace, feels off, check this message"
Selling high-value items ($300+ phones, bikes, cameras, laptops)

Don't use for: business/inventory-scale selling (different economics), real-estate or vehicle-title transfers (legal process beyond scope), or as a substitute for platform-specific dispute processes once fraud already happened (report + chargeback paths differ by platform).

How It Works — Numbered Steps
Price the item: python3 scripts/seller_toolkit.py price --category iphone --condition good --original 799 --age-years 2 → list $447 / expect $388 / floor $330. Cross-check against sold listings for the exact model; the band is a starting anchor, not gospel.
List with the safe-reply ready. When messages arrive, triage each: python3 scripts/seller_toolkit.py message --text "I'll send a 6-digit code..." → per-signal breakdown, score, verdict (BLOCK / PROCEED WITH RULES / LIKELY FINE), and a safe reply line.
Arrange the meetup with the checklist: python3 scripts/seller_toolkit.py meetup --item "iPhone 13" --price 350 High-value deals escalate automatically (bank lobby, IMEI/serial checks).

If shipping instead: python3 scripts/seller_toolkit.py labels --weight-lb 3.2 --dimensions 12x9x4 --value 350 Ship only after payment reads COMPLETED, never 'pending'.

Walk away on any no-go condition. The sale you lose to a scammer costs more than the sale you decline from a weird-but-honest buyer.
The 9 Scam Families (message triage)

Family Points Core move

verification-code phishing 40 hijacks your number (Google Voice port)
overpayment / 'accidental' extra 35 fake payment + refund of real money
off-platform payment 30 Zelle/Venmo/gift cards/crypto = no protection
off-platform shipping / courier 30 fake labels, fake payment emails
too-good offer 30 pays above asking — always a setup
escrow / agent theater 20 third-party payment misdirection
switch / swap bait 18 hands back a different item
pressure / urgency 15 rushes you past your own judgment
price-blind buyer 12 zero questions = reseller or switch scam

Score ≥40 → BLOCK. 15-39 → PROCEED WITH RULES (one terms reply, second offense = block). Under 15 → normal caution.

Common Pitfalls

Sharing ANY verification code. The "buyer" is porting your number to Google Voice to run scams as you. Codes are never shared, ever.
Accepting 'pending' payments. Zelle/Venmo/payments can be reversed or fake-screenshotted. Cleared and completed, or cash counted in hand.
Refunding an overpayment. Their payment bounces; your refund is real.
Shipping on the buyer's label. Fake labels get your package stolen or delivered nowhere; you have no proof of delivery for disputes.
Meeting at home because it's heavy. Couch sellers get robbed. Meet in the driveway with a friend, or pay a mover — never alone inside.
Pricing from active listings. Active = unsold = overpriced. Search SOLD/completed listings and undercut the cheapest comparable.
Answering scam scripts with engagement. Every reply confirms your account is active and human. Block, don't debate.
Skipping the proof pack. Photograph the item powered on + serial before boxing. It's your only evidence in a 'item not as described' scam.

Example

Seller listing a 2-year-old iPhone 13, gets an instant message.

$ price --category iphone --condition good --original 799 --age-years 2
→ List $447 / expect $388 / floor $330 (retention 49%)

Buyer: "I'm out of town, my shipper will pick it up. I'll send a 6-digit
code to confirm, zelle you extra $50, text me asap."

$ message --text "
"
→ [+40] code phishing [+30] off-platform payment
[+30] courier shipping [+15] urgency
→ SCAM LIKELIHOOD 100/100 — VERDICT: BLOCK
Safe reply: "Cash in person only. No codes, no shipping, no agents."

Second buyer: "Is the battery health above 85%? Can we meet at the
police station exchange zone tomorrow?"
→ 0 signals, LIKELY FINE → meetup checklist, station flag on.

References

references/scam-field-guide.md — each scam family anatomy: the script, the psychology, the exact counter, and what to do if already hit.
references/pricing-curves.md — the depreciation model, category table, condition multipliers, and local-market calibration method.
eSIM 旅行设置
深度清洁作战规划器