evidence-based-review

Pass

Audited by Gen Agent Trust Hub on Nov 3, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill consists entirely of natural language instructions and metadata. No executable scripts, command-line operations, or network exfiltration patterns were detected. The instructions are aimed at improving code review quality by requiring specific evidence for every claim.
  • [INDIRECT_PROMPT_INJECTION]: As the skill is designed to process external code from pull requests, it inherently possesses a surface for indirect prompt injection. However, the skill's requirements for agents to quote exact lines and cite concrete failure modes provide a manual verification mechanism for human reviewers, mitigating the risk of the agent being misled by instructions hidden in the code being reviewed.
  • Ingestion points: Pull request code, file names, and line content are processed by the agent using this skill.
  • Boundary markers: None explicitly defined in the skill instructions.
  • Capability inventory: The skill mentions the use of gh pr for inline-comment anchors, implying interaction with the GitHub CLI.
  • Sanitization: The skill encourages direct quoting of source material, which facilitates manual verification by the PR author.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 3, 2026, 08:26 PM