WeChat Video Short Video Decomposer (Full Version)

Challenge in Analyzing WeChat Video Account Content

In the WeChat Video Account ecosystem, content creators often face a core challenge: how to systematically identify viral patterns from short videos and extract replicable strategies? Traditional manual analysis is time-consuming and subjective, lacking structured data support. For instance, a video's success may stem from specific script types, pacing segments, or emotional triggers, but these are hard to systematize without data. This leads creators to rely on intuition rather than data-driven insights when optimizing content.

How the Skill Works

This skill leverages a remote service to automate the decomposition of Video Account short videos, generating structured analysis reports. Its core capability transforms video content into quantifiable metrics and recommendations, following key steps:

• Input Recognition and Authentication: The skill accepts two input types: a WeChat Video Account share link (e.g., https://weixin.qq.com/sph/...) or a local video file path (supporting formats like mp4, mov, up to 100MB). First-time use requires API Key configuration via environment variable LY_API_KEY or a config.json file, ensuring secure service access.
• Decomposition Execution:
  1. Deduction Confirmation: Before initiating a new task, it prompts an estimated point deduction of about 128 points (actual varies by complexity), requiring user confirmation to avoid unexpected consumption.
  2. Local File Upload: For local videos, a pre-signed upload flow (get upload URL, direct transfer, confirm upload) exchanges for a video_id, which is then used to create the decomposition task.
  3. API Calls and Polling: The script scripts/analyze_wx_video.py calls the remote endpoint POST /api/v1/social-analytics/collector/wx-video-analyses to create a task, obtains an analysis_task_id, and polls status—typically taking 1–5 minutes. Progress is reported in real-time, with status changes like current_stage relayed to the user.
• Report Generation and Delivery: Upon completion, the skill returns a Markdown-formatted report containing structure segmentation, script type identification, viral factor analysis, six-dimensional scoring, and actionable strategies. The report is rendered directly in the conversation and can also be saved locally, with the path specified by WX_VIDEO_REPORT_FILE.

Applicable Scenarios and Considerations

This skill is designed for WeVideo Account content creators, operators, or data analysts, suitable for the following scenarios:

• Content Optimization: Decompose competitor or own viral videos to extract replicable elements (e.g., script structure, pacing points).
• Data Validation: Use quantitative metrics like six-dimensional scoring to assist decisions, reducing bias from subjective judgment.

However, note its boundaries:
- Platform Limitation: Only supports WeChat Video Account videos; links from Douyin, Bilibili, or YouTube are not directly processed. If non-Video Account links are provided, users are guided to use local file methods.
- Technical Constraints: Local file size is capped at 100MB; larger files need compression. Authentication failures (exit code 3) require re-obtaining the API Key; insufficient balance (exit code 4) necessitates topping up on the 01Claw platform.
- Point Deduction Mechanism: Each new task consumes points; resuming timed-out tasks (using --task-id) does not double-deduct. Failed tasks (e.g., exit codes 5, 6) refund points, but users must manually confirm retries.

In summary, this skill automates decomposition to elevate video analysis from perception to data-driven insights, but it depends on stable network connectivity and valid authentication—ideal for engineers and creators seeking deep content understanding.