omh-localization-review
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external, user-provided text for localization and cultural-fit review, which introduces a potential surface for indirect prompt injection.
- Ingestion points: Untrusted text strings for review enter the agent context through the localization workflow inputs described in SKILL.md.
- Boundary markers: Absent; the skill instructions do not require the use of delimiters to separate user content from system instructions.
- Capability inventory: The skill utilizes the
omhcommand-line tool for runtime recording and state management. - Sanitization: No explicit sanitization, filtering, or validation of user-provided strings is defined in the instructions.\n- [COMMAND_EXECUTION]: The skill includes instructions to execute the
omhcommand-line utility (e.g.,omh runtime record --skill localization-review) to manage workflow state. This appears to be a vendor-native tool for the Hermes framework.
Audit Metadata