audit-pr

Pass

Audited by Gen Agent Trust Hub on Sep 8, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill reviews untrusted code and pull request metadata, creating a potential surface for indirect prompt injection. The skill includes specific design-level protections to neutralize this threat.
  • Ingestion points: Pull request diffs, titles, descriptions, commit messages, and linked issues are explicitly identified as the content under review in SKILL.md.
  • Boundary markers: The 'Data handling' section of SKILL.md provides a strict instruction that any command found within the data is to be treated as a finding to report on, rather than a command to follow.
  • Capability inventory: The skill is scoped to use file-search, file-read, and project-standard audit tools (formatters, linters, and tests).
  • Sanitization: The skill mandates the use of a [REDACTED] placeholder for all credentials found in the code, ensuring sensitive data is not leaked in the final report.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 8, 2026, 12:22 AM
Security Audit — agent-trust-hub — audit-pr