WeChat RSS

通过 wcrss.com API 获取并展示微信公众号的最新文章。当用户请求查看微信公众号文章、获取公众号最新发布内容、阅读微信RSS订阅或浏览公众号内容时使用此技能。该技能会从环境变量 WCRSS_API_KEY 中读取 API Key，并调用 wcrss.com 的接口来获取文章数据。

WeChat RSS Skill

Fetch and display the latest articles from WeChat public accounts using the wcrss.com API.

Prerequisites

Before using this skill, the user must:

1. Register an account at https://wcrss.com/
2. Add favorite public accounts at https://wcrss.com/publishers
3. Generate an API key at https://wcrss.com/settings
4. Save the API key as an environment variable WCRSS_API_KEY

Python Scripts

This skill uses Python scripts for data fetching:

• Location: scripts/wechat_rss.py (relative to skill directory)
• Commands:
  • python scripts/wechat_rss.py fetch <recentDays> <num> - Fetch articles (auto-cached for 1 hour)
  • python scripts/wechat_rss.py get <index> - Get single article by index
  • python scripts/wechat_rss.py count - Get total articles count

Workflow for Agent

Step 1: Fetch Articles List

Execute the Python script to fetch articles:

python scripts/wechat_rss.py fetch 3 10

Step 2: Get Total Articles Count

python scripts/wechat_rss.py count

Step 3: Process Each Article

For each article index from 0 to (count-1), call:

python scripts/wechat_rss.py get <index>

This returns:

{
  "title": "文章标题",
  "author": "公众号名称",
  "content_html": "<p>HTML内容...</p>",
  "url": "https://example.com/s/article",
  "publish_time": "%Y-%m-%d %H:%M",
  "description": "文章简介"
}

Step 4: Summarize Each Article

For each article, use the LLM to summarize the content_html into key points (bullet list).

LLM Prompt:

请从以下文章内容中提取关键要点，以条目的形式列出：

标题：{title}
作者：{author}
内容：{content_html}
时间：{publish_time}

请提取3-8个核心要点，每个要点用一句话概括。

Step 5: Display Results

For each article, display in this format:

【{序号}】{标题}
作者：{公众号名称}  {时间}
文章要点：
- 要点1
- 要点2
- 要点3
原文链接：{url}
---

Error Handling

• If "WCRSS_API_KEY environment variable not set" error occurs, guide the user to configure the API key
• If API returns an error, display the error message to the user
• If no articles are found, inform the user and suggest checking their followed publishers