oma-mobile

Pass

Audited by Gen Agent Trust Hub on Oct 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external API responses and OpenAPI specifications which could potentially influence agent behavior if the source is compromised.
  • Ingestion points: API responses processed by Dio (Flutter) or the generated Client (Swift), and the openapi.yaml specification file.
  • Boundary markers: The instructions do not specify the use of delimiters or explicit "ignore embedded instructions" warnings for data fetched from remote endpoints.
  • Capability inventory: The skill possesses the capability to modify source code (WRITE) and execute build/verification tools (CALL_TOOL).
  • Sanitization: No explicit sanitization or validation of the remote content is described before it is integrated into the application logic.
  • [DYNAMIC_EXECUTION]: The skill utilizes a build-time plugin to generate executable Swift source code from a remote specification file.
  • Evidence: The workflow involves fetching openapi.yaml from a remote backend via curl and the subsequent generation of Client, Operations, and Components types during the swift build process.
  • [COMMAND_EXECUTION]: The skill is authorized to invoke a wide range of mobile development, build, and verification tools.
  • Evidence: The operational flow explicitly mentions calling flutter build, swift build, flutter test, swift test, and Maestro E2E checks, as well as simulator and emulator management commands.
  • [EXTERNAL_DOWNLOADS]: The skill fetches necessary development dependencies and build plugins from the Apple GitHub organization.
  • Evidence: References to apple/swift-openapi-generator, apple/swift-openapi-runtime, and apple/swift-openapi-urlsession are included in the Swift iOS implementation guidelines.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 6, 2026, 04:52 PM