ai-pr-review

Pass

Audited by Gen Agent Trust Hub on Mar 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill requires 'Bash' tool access to execute the GitHub CLI ('gh') and git commands. This is used to fetch pull request metadata, analyze diffs, and perform the automated 'fix-and-verify' workflow, which involves committing and pushing code changes.
  • [PROMPT_INJECTION]: The skill has a significant indirect prompt injection surface (Category 8) evidenced by: 1. Ingestion points: Pull request unified diffs, PR titles/bodies, and acceptance criteria from linked GitHub issues. 2. Boundary markers: There are no explicit instructions or delimiters in 'SKILL.md' to treat these external inputs as untrusted data or to ignore embedded instructions. 3. Capability inventory: The agent has the ability to execute shell commands, modify the filesystem, and push code to remote branches. 4. Sanitization: No content validation, sanitization, or filtering of the fetched external data is defined before the agent processes it to generate reviews or fixes.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 2, 2026, 10:20 PM