babysit-pr

Pass

Audited by Gen Agent Trust Hub on Nov 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from pull request feedback and CI logs, creating a potential surface for indirect prompt injection.\n
  • Ingestion points: SKILL.md workflow steps 1 and 2 involve reading pull request comments, reviews, and CI failure logs.\n
  • Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are used for the external data ingestion.\n
  • Capability inventory: The skill possesses capabilities for local file modification (Workflow Step 4), remote code publication via $file-pr (Workflow Step 7), and branch merging (Merge section).\n
  • Sanitization: Significant mitigation is provided by a mandatory Human-In-The-Loop (HITL) requirement; the agent is prohibited from committing or pushing changes without explicit user approval. Furthermore, it enforces branch protection rules and requires human-initiated status changes before merging, preventing the agent from being manipulated into unauthorized merges.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 6, 2026, 09:14 PM