Security audit
Security checks for vulnerabilities and agentic risk
This skill appears to be a local Azure OpenAI proxy guide, but it asks for broad agent authority and gives inconsistent, under-scoped guidance around when it runs and how API keys are handled.
Install only if you specifically need a local Azure OpenAI compatibility proxy. Keep it bound to 127.0.0.1, avoid storing Azure API keys in persistent client config when possible, review any script you add or run, and be aware that the skill's requested write/exec authority is broader than its Markdown guide clearly justifies.
The phrase "Use when 需要代码生成、编程辅助、调试测试、开发部署时使用" describes very broad scenarios that overlap with common everyday developer requests rather than narrow invocation triggers. This makes it unclear when the skill should activate versus when a more specific Azure-proxy context is required, increasing the chance of unintended invocation.
The manifest declares tools twice and inconsistently: once as read/exec and again as read/write/exec. In a skill that is documented as a local request-routing proxy, granting write capability expands the agent's authority beyond the stated purpose and can enable unintended file modification or persistence if the skill is invoked.
The documentation claims the proxy rewrites Authorization: Bearer into api-key, but the setup later instructs users to disable auth header forwarding and manually provide api-key. This inconsistency can cause operators to misconfigure authentication, accidentally place secrets in static client configs, or incorrectly assume the proxy is handling credential translation safely.
The skill instructs users to supply Azure API keys and even shows them in client-side configuration examples, but does not prominently warn about the risks of storing secrets in config files or exposing a local proxy that accepts credentialed requests. In this context, the skill directly mediates access to a paid cloud AI endpoint, so weak credential-handling guidance can lead to key leakage, misuse, or unintended external exposure.
No suspicious patterns detected.
