The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from PR diffs, Git diffs, and project files (plan.md, report.md, etc.) which could theoretically contain malicious instructions. However, this is an inherent risk of its primary function to analyze code changes.
Ingestion points: SKILL.md (Step 2) reads data from gh pr diff, git diff, and various local markdown/HTML files.
Boundary markers: None explicitly defined in the instructions for the generated HTML content.
Capability inventory: The skill uses Bash, Read, Glob, Grep, Write, and AskUserQuestion tools.
Sanitization: Not explicitly mentioned; however, the output is a static HTML quiz intended for human review.
[COMMAND_EXECUTION]: The skill uses Bash to execute standard developer commands like gh pr view, gh pr diff, and git diff. These operations are well-scoped to the stated purpose of analyzing code changes within a repository.

quiz