om-auto-merge-triage

Pass

Audited by Gen Agent Trust Hub on Nov 21, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted content from the repository and tracker (such as Pull Request descriptions, code diffs, and issue comments) to perform "AI judgment" on merge eligibility.
  • Ingestion points: External data is retrieved using tracker operations like list-prs, get-pr, get-pr-diff, and list-issue-comments as documented in SKILL.md and references/ai-judgment.md.
  • Boundary markers: The skill provides defensive instructions in references/agentic-setup.md under the "Untrusted content boundary" section, directing the agent to treat repository content as data and ignore any embedded directives or instructions encountered during analysis.
  • Capability inventory: The skill possesses the capability to perform automated actions such as applying labels (apply_label) and posting or updating comments on pull requests (comment-pr, update-comment).
  • Sanitization: Input validation is mandated in references/agentic-setup.md, which requires validation of externally-sourced identifiers and branch names against strict regular expressions (^[A-Za-z0-9._/-]+$) and ensures they are quoted before interpolation into shell commands or tracker calls.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 21, 2026, 04:53 PM