1688-shopkeeper

1688选品铺货 + 商机趋势专家。用于：(1) 在1688搜索商品/选品找货源 (2) 查询已绑定的下游店铺 (3) 将商品铺货到抖音/拼多多/小红书/淘宝等平台 (4) 配置1688 AK密钥 (5) 查看即时商机热榜 (6) 查看类目/行业趋势与价格分布 (7) 生成店铺经营日报并输出主营商品选品建议。 触...

Security Scan

The skill's code, required credential (ALI_1688_AK), and runtime behavior align with its stated purpose (searching 1688, fetching details, and publishing to downstream shops); it calls 1688 APIs and stores workspace snapshots locally — nothing appears disproportionate or covert.

Purpose & Capability

The skill is an 1688 integrator: it declares a primary credential ALI_1688_AK and all network calls go to a 1688 domain (https://ainext.1688.com). The CLI commands (search, prod_detail, publish, shops, configure, etc.) and saved local data (product details, publish snapshots) are coherent with the described functionality.

Instruction Scope

SKILL.md instructs the agent to run the included CLI commands directly and to load capability reference files on first use. The runtime code reads/writes local workspace files and the OpenClaw config fallback (~/.openclaw/openclaw.json). This is expected for a CLI-style skill, but the agent will persist API responses and request previews to disk (product details, publish snapshots) so users should expect local storage of potentially sensitive shop/product data.

Install Mechanism

No external install/download specified (instruction-only install spec). The package includes Python code and a simple requirements.txt (requests). No remote installers or opaque archive downloads are used.

Credentials

The declared primaryEnv (ALI_1688_AK) matches the skill purpose. The code also reads standard environment knobs OPENCLAW_CONFIG_DIR and OPENCLAW_WORKSPACE_DIR and will read a fallback openclaw.json; these additional env vars are common for workspace/config control but are not listed in SKILL.md's metadata. The skill writes files under the workspace (~/.openclaw by default) and stores API request/response snapshots — reasonable for debugging, but users should know data is persisted locally.

Persistence & Privilege

always:false and no evidence the skill modifies other skills or system-wide agent settings. It stores its own data under user workspace paths and uses the OpenClaw config only as an AK fallback; autonomous invocation is allowed (default) which is normal for skills.

Assessment

This skill appears to be what it says: it needs your ALI_1688_AK to sign requests to 1688 and will call https://ainext.1688.com for search, detail, publish and trends. Before installing, consider: (1) providing ALI_1688_AK grants the skill ability to act on your 1688 account (search and publish) — only provide keys you intend to use; (2) the skill will create files under ~/.openclaw (or OPENCLAW_WORKSPACE_DIR if set) and will save product details and publish API request/response snapshots there, which may contain shop/product metadata; (3) the code attempts to read a fallback ~/.openclaw/openclaw.json to fetch an AK if present — it only reads the 1688-shopkeeper entry, not arbitrary other skill entries; (4) the AK parsing function first tries base64 decode then slices the string into a 32-char secret + id format — if your AK format differs the skill may fail to parse it (operational issue, not a security flaw). If you are comfortable with those behaviors and with granting the ALI_1688_AK credential, the skill is coherent with its stated purpose. If you need stricter containment, avoid providing the AK or inspect/redirect the workspace/config paths (OPENCLAW_WORKSPACE_DIR / OPENCLAW_CONFIG_DIR) before use.

统一入口：python3 {baseDir}/cli.py <command> [options]

命令速查

命令	说明	示例
search	找商品	cli.py search --query "帮我找1688上支持一件代发包邮的露营椅，100元以内" --channel douyin
prod_detail	商品详情	cli.py prod_detail --item-ids "991122553819,894138137003"
shops	查绑定店铺	cli.py shops
publish	铺货	cli.py publish --shop-code CODE --data-id ID
opportunities	商机热榜	cli.py opportunities
trend	趋势洞察	cli.py trend --query "大码女装"
shop_daily	店铺经营日报	cli.py shop_daily
configure	配置 AK	cli.py configure YOUR_AK
check	检查配置状态	cli.py check

所有命令输出 JSON：{"success": bool, "markdown": str, "data": {...}} 展示时直接输出 markdown 字段，Agent 分析追加在后面，不得混入其中。

使用流程

Agent 根据用户意图直接执行对应命令，无需每次先执行 check。 各命令在 AK 缺失、店铺异常等情况下会自行返回明确错误，Agent 按下方「异常处理」应对即可。

选品铺货典型路径：search → 用户筛选 → shops（确认目标店铺） → publish

商品详情使用指引：

• prod_detail：用户想看商品标题、价格、类目、SKU、CPV 属性、商家信息时使用。

商机/趋势使用指引：

• opportunities：用户想看“此刻/近1小时/热榜/趋势商机”。
• trend：用户询问某宽泛类目/行业的周期趋势、价格分布、热品概况；关键词过长或过细时先按用户原词尝试，若空结果再提示用户换更宽泛/拆分词后重试。
• shop_daily：用户想看“店铺经营日报/今日动销/渠道经营诊断/主营商品选品建议”。

安全声明

风险级别	命令	Agent 行为
只读	search, shops, check, opportunities, trend, shop_daily	直接执行
配置	configure	提示影响范围后执行
写入	publish	必须先 dry-run 预检查；仅当写入目标不唯一时追问，目标唯一则直接执行

全局写入规则（适用于所有写操作）：

1. 必须先执行 dry-run。
2. 仅当写入目标不唯一（商品或店铺存在多个候选）时，才向用户追问一次做歧义消解。
3. 目标唯一时，dry-run 成功后直接执行正式写入，不再二次确认。

异常处理

任何命令输出 success: false 时：

1. 先输出 markdown 字段（已包含用户可读的错误描述）
2. 再根据关键词追加引导：

markdown 关键词	Agent 额外动作
"AK 未配置" 或 "签名无效"/"401"	输出下方 AK 引导话术
"授权过期"	提示用户在 1688 AI版 APP 中重新授权
"店铺未绑定" 或 shops 返回 0 个店铺	输出下方 开店引导话术
"限流"/"429"	建议用户等待 1-2 分钟后重试
其他	仅输出 markdown 即可

详细错误码说明见 references/common/error-handling.md。

执行前置（首次命中能力时必须）

• 首次执行 search 前：先完整阅读 references/capabilities/search.md
• 首次执行 prod_detail 前：先完整阅读 references/capabilities/prod_detail.md
• 首次执行 shops 前：先完整阅读 references/capabilities/shops.md
• 首次执行 publish 前：先完整阅读 references/capabilities/publish.md
• 首次执行 configure 前：先完整阅读 references/capabilities/configure.md
• 首次执行 opportunities 前：先完整阅读 references/capabilities/opportunities.md
• 首次执行 trend 前：先完整阅读 references/capabilities/trend.md
• 首次执行 shop_daily 前：先完整阅读 references/capabilities/shop_daily.md
• 同一会话内后续重复调用同一能力可复用已加载知识；仅在规则冲突或文档更新时重读。

AK 引导话术

"需要先配置 AK。打开 1688 AI版 APP（没装的话点链接下载），首页点击「一键部署开店Claw，全自动化赚钱🦞」，进入页面获取 AK，然后告诉我：'我的AK是 xxx'"

开店引导话术

"还没有绑定店铺。打开 1688 AI版APP → 首页「一键开店」，开好后告诉我。"

FAQ 知识库（按需加载）

用户问经营或技术问题时，先加载对应文件再回答，不凭经验泛泛而谈。

用户话题	加载文件
AK 问题、铺货失败、支持平台、收费	references/faq/base.md
选哪个平台、抖店/拼多多/淘宝	references/faq/platform-selection.md
选品风险、品类、节日选品	references/faq/product-selection.md
运费模板、定价、加价倍率	references/faq/listing-template.md
发货超时、中转费、偏远地区	references/faq/fulfillment.md
退货、仅退款、运费险、售后	references/faq/after-sales.md
新店破零、服务分、推广	references/faq/new-store.md
素材审核、白底图、标题优化	references/faq/content-compliance.md