SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsjuya

juya

General & other

View repository

Updated 10-15 21:52

About this skill

Juya AI Daily Report

Get the daily AI report updated by Juya, and output morning-tea-style HTML for human reading.

Execution Principles

  • Get the current time to determine the daily report date:date "+%Y-%m-%d %H:%M:%S"
  • The daily report is usually published around 09:00 every day; the latest date can be determined based on the time.
  • Set the current skill path to WFP_PATH, pointing to the directory where webfetch-plus is located.
  • Any webfetch-plus command execution must first cd "$WFP_PATH"

Retrieval Methods (only three, by priority)

Priority Method Script / Tool webfetch count Description
1 Preferred RSS rss-to-html.py 0 HTTP pull RSS, best formatting
2 Alternative WeChat Official Account wx-to-html.py 1～2 Used when RSS fails; after converting the body to an RSS item, use the unified rendering
3 Fallback Bilibili collection webfetch-plus → wx-to-html.py -u 2～3 Extract the WeChat link from the video description, then export HTML

Execution Flow

用户请求早报 →
  ① RSS 脚本（rss-to-html.py）
      ↓ 失败 / 无当日条目
  ② 微信脚本（wx-to-html.py）
      ↓ 失败
  ③ B站合集取最新视频简介中的微信链接 → wx-to-html.py -u <链接>
      ↓
  输出 juya-YYYY-MM-DD.html → 打开浏览器

Method 1: RSS (preferred)

python3 <skill-path>/scripts/rss-to-html.py [输出目录]
python3 <skill-path>/scripts/rss-to-html.py [输出目录] -d 2026-06-09
  • RSS source:https://daily.juya.uk/rss.xml
  • Output:juya-YYYY-MM-DD.html
  • No webfetch-plus needed, no stealth needed

Method 2: WeChat Official Account (alternative)

Use when RSS is unavailable or the current day's entry is missing. Flow: locate the article -> extract the body -> convert to an RSS item -> call the same rendering pipeline as Method 1.

python3 <skill-path>/scripts/wx-to-html.py [输出目录]
python3 <skill-path>/scripts/wx-to-html.py [输出目录] -d 2026-06-09

Reduce webfetch calls (try in order):

Scenario Command webfetch count
Already have the WeChat article link wx-to-html.py [目录] -u '<微信链接>' 1
Repeated execution on the same day / cache not expired (6h) Run the script directly 1
The RSS in the script can resolve the WeChat link Automatically skip the album 1
First time or need to refresh the album index wx-to-html.py [目录] --refresh-album 2

Common parameters:

--url / -u          微信文章直链，跳过专辑抓取
--date / -d         指定日期 YYYY-MM-DD（默认最新一期）
--refresh-album     忽略 RSS/缓存，强制重抓专辑
--no-open           生成后不自动打开浏览器

At the end of the script it prints webfetch 调用次数: N。

Official account album (used internally by the script):

  • __biz: MzIyMDk0MDY1OA==
  • album_id: 4066755324999598083

Method 3: Bilibili collection (fallback)

When the WeChat script also fails, get the WeChat link from the latest video description in the Bilibili AI daily report collection, then use -u to fetch only the body.

Step 1 — Open the collection and find the latest video BV number:

cd "$WFP_PATH"
echo "https://space.bilibili.com/285286947/channel/collectiondetail?sid=572036 --wait 3000" | bash bin/wfp.sh

Step 2 — Fetch the video page and extract the WeChat link from the description:

echo "https://www.bilibili.com/video/<BV号> --wait 3000" | bash bin/wfp.sh

Example description format:相关链接和文字版请看：https://mp.weixin.qq.com/s/xxxxx

Step 3 — Export HTML using the direct link (only 1 webfetch):

python3 <skill-path>/scripts/wx-to-html.py [输出目录] -u 'https://mp.weixin.qq.com/s/xxxxx'

Failure Handling

  1. Read the attempt_{n}_{N}.metadata.json output by webfetch-plus, in suggestion field
  2. Adjust parameters and retry as suggested (WeChat pages usually need --stealth, already built into the script)
  3. Retry at most 3 times at the same level, then downgrade to the next priority method

Output Format

The daily report HTML contains the following sections:

  • Overview Highlights: key news of the day
  • Model Releases: new model release information
  • Developer Ecosystem: development tools, API updates
  • Product Applications: new product features
  • Technology and Insights: technical analysis, security reports
  • Industry Trends: Policy, financing, industry news
  • Foresight and rumors: Future plans and rumors

Each news item has a number (#1-#N), including a title, detailed description, and a link to the original article.