The Agent Skills Directory

[SAFE]: The skill provides advisory content and instructions for developers to diagnose performance bottlenecks in Django applications. It does not perform any automated actions that could compromise a system.
[COMMAND_EXECUTION]: The instructions suggest that the developer manually run pytest in their own environment to profile database queries. The provided code snippet uses standard Django testing utilities (CaptureQueriesContext) and does not involve the execution of untrusted or dynamically generated strings.
[INDIRECT_PROMPT_INJECTION]: The skill asks developers to provide endpoint URLs and APM traces for analysis. However, the agent's role is limited to generating reports and Jira tickets based on this data; it does not possess capabilities (such as file writing or network requests) that could be exploited via injected instructions in the provided traces.

perf-audit