juicebro-content-hub Skill

Overview

• Full-platform Search: What updates are available across all platforms today/in the last N days/in the last week?
• Single-platform Query: Queries for specified platforms such as Weibo, Xiaohongshu, Douyin, Bilibili, Xueqiu, Toutiao, WeChat Official Account, Xiaoyuzhou, Ximalaya, etc.
• Topic Query: Filter content on specific topics like Bitcoin, Gold, US Stocks, AI, Trump, etc.
• Content Type Query: Filter content types such as video, article, audio, short post, image-text post, etc.
• Summary Report: Generate structured summaries in the form of daily reports, weekly reports, or summaries by platform/topic/content type.
• Navigation Recommendation: Recommend platforms suitable for specific content types.

Account Matrix (Authoritative Data Source)

data/accounts.json

Usage Flow

Step 1: Intent Recognition

data/query_intents.json

intent_type: all_platform_query | platform_single_query | topic_query |
             content_type_query | aggregation_report | navigation_recommend

• platforms

  : Platforms specified by the user (select all platforms with is_queryable=true if none specified)

• time_range

  : today / last_3_days / last_7_days / last_30_days

• topic

  : Topic tags (refer to

  data/topic_keywords.json

  )

• content_type

  : video / article / audio / short_post / image_post

• output_format

  : list / digest / report

Step 2: Platform Adaptation

1. Retrieve account information from

   data/accounts.json

2. Check query strategies in

   data/platform_rules.json

3. Refer to

   adapters/{platform}.md

   for field extraction methods and failure handling
4. If the platform has

   is_queryable: false

   , directly output restriction instructions and skip this platform
5. If the platform is queryable but access results are insufficient, clearly state this without speculating or supplementing content

Step 3: Content Normalization

NormalizedPost

schemas/normalized_post.schema.json

{
  "platform": "weibo",
  "account": "果汁哥6688",
  "publish_time": "2024-01-15T10:30:00+08:00",
  "title": null,
  "summary": "今天聊聊比特币最新走势...",
  "content_type": "short_post",
  "topic_tags": ["bitcoin", "crypto"],
  "source_locator": "https://weibo.com/...",
  "dedupe_hash": "a3f2c1..."
}

Step 4: Deduplication & Aggregation

• Calculate

  dedupe_hash

  to merge duplicate content across platforms
• Select aggregation prompts based on intent:
  • Time-based aggregation →

    prompts/summarize_recent_posts.md

  • Platform-based aggregation →

    prompts/summarize_by_platform.md

  • Topic-based aggregation →

    prompts/summarize_by_topic.md

  • Field extraction →

    prompts/extract_post_fields.md

  • No results or restricted →

    prompts/fallback_response.md

Step 5: Output

schemas/normalized_result.schema.json

Important Constraints

1. No Content Fabrication: If a platform is inaccessible or public results are insufficient, explain the reason without speculating or inventing content
2. No Login Bypassing: Directly state restrictions for paid/login-required content without attempting to bypass login
3. Source Labeling: Each piece of content must be labeled with platform + account + source_locator
4. Deduplication Priority: Merge identical content across platforms for display, do not list duplicates
5. Access Status Disclosure: Label the access status (success/restricted/no content) of each platform in the output for every query

Extension Points

• Add New Platforms: Add

  adapters/{platform}.md

  + update

  data/accounts.json

  and

  data/platform_rules.json

• Add New Topics: Update

  data/topic_keywords.json

• Improve Output: Optimize prompt templates in

  prompts/*.md

• Connect Real Data Sources (V2): Implement platform methods in

  scripts/optional_fetcher.py

Reference Documents

• Account Matrix
• Platform Rules
• Query Intents
• Topic Keywords
• Normalized Schema
• Architecture Description
• Supported Queries List
• Safety Statement