The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, specifically pull request diffs and descriptions. It lacks explicit boundary markers or 'ignore' instructions when interpolating this data into AI prompts (e.g., review_prompt and security_analysis_prompt). This creates a surface where an attacker could embed malicious instructions in a pull request to influence the behavior of the reviewing agent.
[COMMAND_EXECUTION]: Multiple scripts and workflows (GitHub Actions, Python orchestrator) use subprocess calls to execute shell commands like sonar-scanner, semgrep, and trufflehog. While these are standard tools for the described use case, they represent a significant capability tier that could be abused if the input arguments are manipulated.
[EXTERNAL_DOWNLOADS]: The skill references an external script (<support-root>/scripts/ai_review.py) as a prerequisite and suggests using GitHub Actions like actions/checkout@v4 and actions/github-script@v7. These are well-known services and repositories, making the references standard for DevOps workflows.

code-review-ai-ai-review