babysit
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from GitHub pull request comments and review summaries which could potentially contain malicious instructions.
- Ingestion points: Pull request comments and review activity are fetched using GitHub CLI commands as defined in
references/scheduling.md. - Boundary markers: The skill implements a 'Human-In-The-Loop' (HITL) boundary, pausing for human conversation and only automatically addressing bot or Copilot nits.
- Capability inventory: The skill has the ability to edit files and run shell commands via
Bash(e.g., git rebase, commit, push). - Sanitization: The instructions do not specify any sanitization or escaping of the ingested pull request content before processing.
- [DYNAMIC_EXECUTION]: The skill executes 'project gate' commands such as tests and linters to verify the integrity of the pull request.
- Execution method: Invokes project-specific tooling via the
Bashtool. - Context: This execution is part of the intended workflow to ensure PR health but involves running scripts defined within the repository context.
Audit Metadata