The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a helper script scripts/review_diff.py that executes git commands via subprocess.run. This is a legitimate function used to retrieve file changes and diff content for the review process. The script follows security best practices by using argument lists rather than shell interpolation, which prevents command injection vulnerabilities.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data (code diffs). While this creates an attack surface for indirect prompt injection, the skill's primary purpose is to perform security audits. It explicitly instructs the agent to look for safety risks, secrets, and injection vulnerabilities in the code it reviews, effectively acting as a defensive layer.
[DATA_EXPOSURE]: The provided script contains regular expressions (RISK_PATTERNS) designed to detect AWS API keys and private keys. These patterns are used for heuristic security scanning of the analyzed code and do not represent hardcoded credentials or data exfiltration attempts.

code-reviewer