team-pipeline
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFE
Full Analysis
- Untrusted Content Handling: The skill identifies external data sources such as emails, chat transcripts, and external documents as untrusted. It includes specific instructions to treat this content as data rather than instructions, which serves as a mitigation against potential indirect prompt injection attacks.
- Human-in-the-Loop Safeguards: For actions that originate from message content (like naming a recipient or recipient target), the skill requires the agent to present the exact details for user approval before execution. This ensures that the user maintains control over operations triggered by external messages.
- URL and Link Safety: A specific security guideline is included to avoid rendering links found within untrusted content. Instead, the skill instructs the agent to link to records or threads using unique IDs, reducing the risk of accidental exposure to malicious URLs.
- Connector Permission Guidance: The instructions emphasize that the skill should operate within the permissions defined by the specific tool connectors, preventing the agent from overriding or silently widening the scope of access beyond what is intended by the system's authorization model.
Audit Metadata