omh-support-operations
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data in the form of support cases, creating an inherent surface for indirect prompt injection.
- Ingestion points: The skill explicitly requires 'support case' as a primary input.
- Boundary markers: The skill provides structural guidance, such as the 'Completion Checklist' and 'Handoff policy', which instruct the agent to distinguish between case facts and unknowns.
- Capability inventory: The skill utilizes the
omhCLI tool (a vendor resource) to record runtime evidence. - Sanitization: No programmatic sanitization is present; the skill relies on instructional constraints to manage untrusted input.
Audit Metadata