The Agent Skills Directory

[SAFE]: The skill defines a structured methodology for reviewing code quality, focusing on maintainability, error handling, and regression coverage. It does not contain any instructions for network access, file system modification, or command execution.
[INDIRECT_PROMPT_INJECTION]: As a code review tool, the skill naturally processes external source code which constitutes untrusted data. However, the instructions are confined to generating analysis reports and recommendations for the user. The skill lacks capabilities for code execution or network exfiltration, and the metadata specifies a read-only sandbox mode, effectively mitigating the risk of instructions embedded in analyzed code causing harmful actions.

code-reviewer