skills/openai/codex/babysit-pr/Gen Agent Trust Hub

babysit-pr

Pass

Audited by Gen Agent Trust Hub on Apr 18, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • Command Execution: The skill uses subprocess.run to interact with the gh (GitHub CLI) tool. This is a core component of its functionality, allowing it to poll PR status, view check results, and trigger reruns of failed CI jobs. The implementation includes structured error handling and uses the CLI's JSON output capabilities for safer data parsing.
  • Git Safety Guardrails: The instructions include explicit rules for safe operations, such as working only on the PR head branch, avoiding destructive commands, and checking for uncommitted local changes before making edits.
  • External Interaction: The skill communicates exclusively with GitHub APIs via the official gh tool. It retrieves review comments and CI status from trusted repository authors (owners, members, and collaborators) to prevent processing instructions from unauthorized external actors.
  • State Management: The script maintains a state file in /tmp to track retry attempts and seen comments, ensuring it adheres to the configured flaky retry budget (default of 3) and avoids redundant processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 18, 2026, 12:14 AM