pipeline-report
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements comprehensive data protection guidelines, explicitly forbidding the inclusion of secrets, API tokens, connection strings, and row-level data samples in outbound messages.
- [SAFE]: It enforces strict credential handling best practices, instructing the agent to use interactive configuration tools or environment variables (BRUIN_CLOUD_API_KEY) rather than passing secrets as command-line arguments.
- [SAFE]: The skill requires mandatory human approval for critical messages and the first post of a session, providing oversight for all outbound communication.
- [INDIRECT_PROMPT_INJECTION]: The skill processes finding files generated by other specialist skills (e.g., from .context/) and interpolates them into reports. While this is a potential ingestion point for untrusted data, the risk is mitigated by the summary-only nature of the report and the requirement for human approval before delivery.
- Ingestion points: Finding files in
.context/(e.g., drift findings). - Boundary markers: Not explicitly defined for ingested content, though output is structured via a specific template.
- Capability inventory: Outbound chat delivery and local file writing to
.context/reports-YYYY-MM-DD.jsonl. - Sanitization: Instructions explicitly mandate redacting sensitive data and summarizing content rather than pasting raw output.
Audit Metadata