omh-design-quality-gate
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a custom shell command
omh runtime recordto track and log the workflow progression. This command is a platform-specific utility for state management.\n- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted user content such as source content and design references while having access to shell commands and internal functions.\n - Ingestion points: User-provided descriptions, source materials, and external style references are ingested into the workflow to generate rubrics and quality matrices.\n
- Boundary markers: The skill refers to "evidence boundaries" but lacks explicit system-level delimiters or instructions to ignore commands potentially embedded in the ingested design content.\n
- Capability inventory: The skill has command-line access via the
omhtool and can load or view other internal skill files viaskill_view.\n - Sanitization: There is no explicit mechanism described to sanitize, escape, or validate the content of ingested external materials before they are processed by the agent.
Audit Metadata