The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process and analyze external, untrusted code changes via git diff and file reads. This creates a surface where malicious instructions hidden in code comments or logic could attempt to mislead the agent or influence its analysis.
Ingestion points: Code content and diffs retrieved via shell commands in SKILL.md and references/initial-assessment.md.
Boundary markers: The instructions do not define specific delimiters to separate untrusted code from agent instructions.
Capability inventory: Access to shell commands (git) and file reading for analysis.
Sanitization: No explicit sanitization or filtering of reviewed content is described.
[COMMAND_EXECUTION]: The skill utilizes local shell commands (git diff, git log, bash) to perform initial assessments and schema validation. While these are standard development tools, they involve executing processes on the host environment.
[SAFE]: The references/security-review.md file contains examples of SQL injection and XSS vulnerabilities. These are documented for educational purposes with clear distinctions between 'vulnerable' and 'safe' patterns, posing no risk to the execution environment.

code-review-analysis