The Agent Skills Directory

[SAFE]: The skill consists of purely instructional documentation without any scripts, commands, or external dependencies. No malicious patterns such as credential theft, remote code execution, or obfuscation were detected.
[PROMPT_INJECTION]: The skill defines a process for analyzing external code, which inherently presents a surface for indirect prompt injection. A malicious actor could embed instructions within code comments or data strings in an attempt to manipulate the agent's evaluation. 1. Ingestion points: Source code files and change descriptions provided to the agent for review in SKILL.md. 2. Boundary markers: None specified; instructions do not include explicit delimiters or warnings to ignore instructions found within the code being analyzed. 3. Capability inventory: The skill's primary function is generating a text-based review report; it does not perform file modifications, network operations, or subprocess executions. 4. Sanitization: No sanitization or validation of the untrusted code content is mentioned.

code-review