AI Education Daily

Problem

AI education updates are scattered across WeChat accounts, web pages, and policy releases, with heavy noise, short half-life, and commercial promotion. Engineers or product owners need a quick view of ministry signals, real classroom practice, and parent feedback without manually scanning and filtering.

How It Works

AI Education Daily runs on a schedule or manual trigger and produces a structured digest. It uses two retrieval paths: wechat-article-search and coze-web-search. It builds queries around policy signals, classroom practice, and parent/user experience, requiring at least one practical or parent-side angle.

Key steps:
- Filtering: removes product pitches, coupons, clickbait, and items older than 7 days.
- Scoring: rates timeliness, usefulness, source authority, and audience fit out of 10; only items at 4.5 or higher are kept.
- Output: organizes keywords, action advice, parent cases, practice, policy, briefs, and quotes; by default it sends the result to Feishu and writes a local Markdown file only after the user confirms archiving.

Boundaries

This skill is useful for a lightweight education AI news feed, not for deep research, competitive analysis, or compliance review. It depends on the availability of two sub-skills, and result quality varies with keyword coverage. Since it does not archive by default, users must explicitly request archiving to generate the dated Markdown file.