The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands to perform code reviews via the qwen CLI and project-defined verification scripts such as npm run build or test. These operations are performed within a sandbox and require explicit user consent or are triggered in a managed GitHub Actions environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data such as issue descriptions and pull request feedback. It incorporates strong defensive instructions to ignore attempts by these inputs to override agent behavior, reveal secrets, or bypass security checks.
[SAFE]: No malicious patterns, obfuscation, or unauthorized access attempts were detected. The skill follows best practices for input validation and employs an 'assume-malicious' posture toward repository content.

autofix