Security audit
Security checks for vulnerabilities and agentic risk
This is a disclosed QA gatekeeping skill that tells an agent to classify code changes, run appropriate tests, and make limited fixes before delivery.
Install this if you want an agent to enforce testing before handing over code changes. Be aware that it uses broad Chinese trigger phrases and may push the agent to run validation commands and make minimal repair edits when tests fail.
The description is written entirely in Chinese and presents the skill behavior as mandatory, with no indication that the user can choose another language or locale. Per the policy, forcing a specific language without opt-in is a natural-language policy violation unless the locale restriction is explicitly justified.
The trigger list uses broad, generic phrases such as '代码验证', '质量门禁', and '发布前验证', which can cause the skill to activate in many ordinary QA or delivery contexts without clear user intent. In an agent system, overbroad activation is dangerous because it can unexpectedly take over workflows, steer execution toward running tests/commands, and apply its gatekeeping logic where it was not explicitly requested.
No suspicious patterns detected.
