wechat-comic-factory

Generate WeChat comic articles and publish them to the WeChat Official Account draft box by executing local Python pipeline scripts. Use when the user wants...

This skill appears internally coherent, but take these practical precautions before running it: 1) Do not place real API keys in the repo or share config.json; keep config.json in the local skill folder and fill only the required fields (text_llm, image_llm, wechat). 2) Review the included scripts (especially publish_draft.py and any helper functions in common.py) to confirm they call only the expected endpoints (WeChat API, configured LLM/image APIs) and don't exfiltrate data to unknown hosts. 3) Run in an isolated environment (virtualenv/container) and inspect output/logs/ output/latest_task.json before claiming success. 4) Be aware tokens may be cached under logs/wechat_access_token.json and task results under output/; treat those files as sensitive. 5) If you need stricter guarantees, run the pipeline manually from a terminal rather than allowing autonomous agent invocation, and verify responses from the scripts include success=true and draft_status=success before announcing a publish.

Type: OpenClaw Skill Name: wechat-comic-factory Version: 1.0.0 The skill bundle is a legitimate automation pipeline for generating WeChat comic articles using LLMs and image generation APIs. It orchestrates a multi-stage process (planning, image generation, compression, and HTML formatting) and includes a final step to push content to the WeChat Official Account draft box. The code follows secure practices by using a local 'config.json' for sensitive credentials (API keys for DeepSeek, DashScope, and WeChat) and includes defensive instructions in SKILL.md to prevent the AI agent from hallucinating success or bypassing configuration checks. No evidence of data exfiltration, malicious execution, or harmful prompt injection was found across the scripts (run_pipeline.py, common.py, etc.).