omh-security-safety-review
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process untrusted data such as user prompts, code diffs, and configuration files for security reviews, which naturally exposes it to indirect prompt injection attempts.
- Ingestion points: It ingests target workflows, code changes, and prompts via user-provided evidence as described in the
Required inputssection ofSKILL.md. - Boundary markers: The skill uses structured artifact contracts (e.g.,
prompt_injection_risk_review/v1) and specific checklists to maintain boundaries between the untrusted data and the agent's instructions. - Capability inventory: The skill's instructions explicitly forbid it from running scanners, mutating dependencies, or executing destructive commands, limiting the impact of any potential injection.
- Sanitization: Instructions require the use of redacted evidence and concrete remediation handoffs to ensure sensitive data is not leaked or misused during the review process.
Audit Metadata