omh-verification-gate
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (such as test logs, command outputs, and requirement specifications) to determine merge readiness. \n
- Ingestion points: Processes code diffs, observed command outputs, and requirements as described in
SKILL.mdandreferences/requirement-coverage-map.md. \n - Boundary markers: Implements a 'not-evidence boundary' to separate unverified claims from observed verification data. \n
- Capability inventory: Triggers the execution of local verification commands such as lint, test, and build. \n
- Sanitization: Does not define methods for sanitizing or escaping the content within processed logs or requirement documents that could potentially contain malicious instructions. \n- [COMMAND_EXECUTION]: The skill facilitates the execution of local tools (linting, building, testing) based on the project's own configuration. This presents a potential vector if a repository contains malicious configuration files meant to execute dangerous code during the verification phase.
Audit Metadata