omh-mobile-release
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, potentially untrusted data such as lists of shipped SDKs, beta test results, and crash/ANR figures to generate release and hotfix plans. An attacker could potentially inject malicious instructions into these data sources to influence the generated plans.
- Ingestion points: The skill collects SDK lists, beta tester feedback, and crash/ANR figures from external consoles and logs (mentioned in SKILL.md).
- Boundary markers: The skill does not define specific delimiters or explicit instructions for the agent to ignore embedded commands within these data inputs.
- Capability inventory: While the skill is primarily a planner, it uses the
omhCLI tool for runtime recording (e.g.,omh runtime record), representing a subprocess capability that could be targeted by an injection chain. - Sanitization: The skill includes safety rules to verify that privacy declarations match the actual shipped SDKs, providing a basic form of consistency validation.
- [SAFE]: The skill references official documentation from trusted sources including Apple (developer.apple.com) and Google (developer.android.com) for toolchain requirements and store rules. These references are handled safely as static documentation links.
Audit Metadata