Reveyes Skill

使用 Reveyes API 批量抓取亚马逊商品评论，支持 20 个站点。 输出评论列表、星级分布统计和差评摘要。 Use when: 用户提到抓评论、查差评、分析竞品口碑、给出 ASIN 编号需要评论数据。 NOT for: 分析已经抓好的本地评论文件，或查询亚马逊商品价格/销量。

Security Scan

The skill’s requests, instructions, and included code align with its stated purpose of fetching Amazon reviews via the Reveyes API and nothing in the package indicates unrelated or hidden behavior.

Purpose & Capability

Skill name/description, README, clawhub.json, SKILL.md, and scripts/fetch.py consistently describe using the Reveyes service to fetch Amazon reviews. The only required credential is REVEYES_API_KEY and the Python reveyes SDK is declared — both are appropriate and proportional to the task.

Instruction Scope

SKILL.md instructs extracting ASIN and parameters then invoking scripts/fetch.py. The instructions reference only the Reveyes API, the REVEYES_API_KEY env var, and local script execution; they do not ask the agent to read unrelated files, other environment variables, or send data to third parties outside reveyes.cn.

Install Mechanism

There is no automatic install spec in the skill bundle (instruction-only), and the README/clawhub.json recommend installing the Python package 'reveyes' via pip. This is an expected dependency for a Python-based API client and does not involve arbitrary binary downloads or suspicious URLs.

Credentials

Only REVEYES_API_KEY is required. No unrelated secrets, system config paths, or multiple credentials are requested. The script reads that single env var and fails cleanly if it's missing.

Persistence & Privilege

The skill is not always-enabled, does not request privileged/persistent system changes, and does not modify other skills or system-wide settings.

Assessment

This skill appears coherent and limited to calling the Reveyes API. Before installing: 1) Verify the Reveyes service and the 'reveyes' PyPI package are from the expected publisher (check the GitHub repo and PyPI page) to avoid installing a malicious package impersonating Reveyes. 2) Only provide a dedicated REVEYES_API_KEY (avoid reusing keys with broader privileges). 3) Be aware that scraped review content will be returned in plaintext/JSON and may appear in agent logs—avoid sending highly sensitive ASINs or private text. 4) Confirm you understand Reveyes pricing/credits (README notes 1 credit/page). If you want extra caution, inspect the reveyes SDK source before installing or run the script in a constrained environment.

When to Run

• 用户发送 ASIN 编号并要求抓取评论（如「帮我抓 B08N5WRWNW 的评论」）
• 用户说「帮我看看这个竞品的差评」「分析一下这款产品的用户反馈」
• 用户要对比同一商品在多个亚马逊站点的口碑
• 用户要导出评论数据做分析

Parameters

从用户消息中提取以下信息：

参数	说明	默认值
asin	亚马逊 ASIN，10 位字母数字（必须）	—
marketplace	站点代码，见下方列表	US
pages	抓取页数（1-10），1 页约 10 条评论	2
filter_star	星级筛选：all_stars positive critical five_star…one_star	all_stars

支持的站点代码： US CA MX UK DE FR IT ES NL SE PL BE IE JP IN SG AE SA AU BR

Workflow

1. 提取参数：从用户消息中识别 ASIN（10 位）、站点代码、页数、星级筛选

   • 若用户只给 ASIN，其余使用默认值
   • 若用户说「差评」自动设置 filter_star=critical
   • 若用户说「好评」自动设置 filter_star=positive

2. 执行抓取：调用 scripts/fetch.py，传入参数

   python scripts/fetch.py <ASIN> <marketplace> <pages> <filter_star>
   

3. 等待完成：脚本内部自动轮询（最多 5 分钟），完成后输出 JSON 结果

4. 格式化输出：

   • 汇总：总评论数、平均评分、各星级占比
   • Top 5 差评（如请求的是 critical 或 all_stars）：标题 + 正文前 150 字
   • 如评论数 > 20，额外输出高频关键词（从差评标题提取）

5. 错误处理：

   • AuthenticationError → 提示用户检查 REVEYES_API_KEY
   • InsufficientCreditsError → 提示前往 https://www.reveyes.cn 充值
   • BadParamsError → 提示检查 ASIN 或站点代码是否正确
   • TimeoutError → 告知任务仍在运行，提供 task_id 供后续查询

Output Format

📦 ASIN: B08N5WRWNW  |  站点: US  |  共 87 条评论

⭐ 评分分布：
  ★★★★★  42 条 (48%)  ████████████
  ★★★★☆  18 条 (21%)  █████
  ★★★☆☆   8 条  (9%)  ██
  ★★☆☆☆   5 条  (6%)  █
  ★☆☆☆☆  14 条 (16%)  ████

📝 典型差评（最近 5 条）：
1. ★★ "Stopped working after 2 weeks"
   Bought this expecting quality but it broke down...

2. ★★ "Not as described"
   The color is completely different from the photos...

[查看完整结果或导出 CSV，请告诉我]