Daily Hot Topic Selection

What Problem It Solves

In content creation and marketing, timely capturing social media trends and leveraging them is key to boosting visibility. However, manually tracking hot lists from platforms like Douyin and Xiaohongshu is time-consuming and challenging to accurately match brand keywords and topic angles. This skill automates the process by collecting real-time hot list data, matching via algorithms, and generating a ready-to-use topic report, helping creators quickly identify content opportunities and reduce manual filtering efforts.

How It Works

Core Capabilities

• Asynchronous Task Handling: Uses POST /api/v1/content/hot-daily/tasks to create a task, immediately returning a task_id to avoid long blocking, and polls progress via GET /tasks/{id}, suitable for multi-turn interactions.
• Multi-platform Hot List Collection: Automatically scrapes today's hot lists from Douyin, Xiaohongshu, Weibo, Kuaishou, Zhihu, covering major social media with real-time updates.
• Smart Topic Matching: Based on industry, brand keywords (e.g., 兰蔻,小棕瓶), and marketing goals (e.g., 种草), algorithms assess hot topic relevance to the brand, recommend high-correlation topics with scores, and provide entry angles, copy directions (e.g., Hook, CTA), and risk warnings.
• Complete Report Generation: Outputs a Markdown report including hot list tables, topic cards, and scheduling suggestions, returned as-is from the backend to ensure consistent formatting.

Key Steps

1. Authentication Setup: Requires authentication via LY_API_KEY in config.json in the skill directory; if missing, guide users to obtain and configure it from a specified page.
2. Information Collection: Confirm necessary parameters with users: industry (required) as the industry name, optional brand_keywords (CSV of brand keywords), platforms (e.g., douyin,xiaohongshu), goal (default 种草), and count (default 8).
3. Point Deduction Confirmation: Before running the script, explicitly inform users of an estimated deduction of about 12 points (actual based on task complexity) and require confirmation to avoid unexpected costs.
4. Task Execution and Polling: Use --only-create to create a task and get task_id, then cycle with --poll-task (exit code 13 indicates in-progress) until completion (exit code 0) or failure (exit code 12, etc.).
5. Result Delivery: On success, render the Markdown report body, state actual point deduction and report file path (e.g., HOT_DAILY_REPORT_FILE); on failure, handle based on exit codes, such as point refund and ask if retry is needed.

Boundaries and Considerations

• Strong Timeliness: Hot list data is collected in real-time with platform caliber differences and delays; use topics promptly, and regeneration is needed if expired.
• Human Review Necessary: Topic recommendations and relevance/composite scores are model-based, not guaranteed accurate—serves only as inspiration; manual review for compliance and brand tone is required before publishing.
• Network and Auth Dependence: Tasks rely on stable network and valid API Key; SSL errors can be temporarily bypassed with environment variable LY_SKIP_SSL_VERIFY=1 or --insecure, but only in controlled environments.
• Point Deduction and Failure Handling: Successful tasks deduct points, failed ones (e.g., exit codes 11/12) refund points; the asynchronous interface avoids session interruption, and single poll timeout (default 90 seconds) is not an error—continue polling.
• Platform Coverage Limitations: Supports Douyin, Xiaohongshu, Weibo, Kuaishou, Zhihu, but data collection may be affected by platform rule changes; monitor abnormal exit codes (e.g., 10) and handle accordingly.