The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were detected in the skill's instructions or code references.
[COMMAND_EXECUTION]: The skill contains logic for the agent to guide users through integration tasks, but it explicitly forbids the agent from using tools like write_to_file or replace_in_file to modify the user's project directly.
[DATA_EXFILTRATION]: All code examples use standard WeChat Pay domains (e.g., api.mch.weixin.qq.com) and utilize placeholders (e.g., 'xxx', '19xxxxxxxx') for sensitive credentials like API keys and merchant IDs.
[PROMPT_INJECTION]: The skill uses role-playing instructions ('金融支付系统技术专家') to set a technical context for quality assessment, which is a benign use of persona to improve task accuracy and does not attempt to bypass core safety filters.
[EXTERNAL_DOWNLOADS]: The skill references standard libraries such as OkHttp, Gson, and Bouncy Castle, which are well-known and trusted dependencies in the developer community.

wechatpay-payscore