appllama-usage

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data from the Appllama mobile app library, which includes screen metadata and flow definitions.
  • Ingestion points: Data is retrieved from the Appllama MCP using tools such as search_apps, list_app_screens, and search_screens.
  • Boundary markers: The instructions do not define specific delimiters or instructions for the agent to separate user-provided or external tool data from its primary logic.
  • Capability inventory: The agent is directed to generate app code and perform local testing using an iOS Simulator or Android Emulator.
  • Sanitization: There are no explicit requirements for the agent to sanitize or validate the external content before processing it.
  • [DYNAMIC_EXECUTION]: The skill describes an iterative development cycle where the agent generates code for app screens and executes them in a local simulator or emulator to verify design and motion fidelity.
  • [COMMAND_EXECUTION]: The playbooks mention the potential use of external command-line tools for image generation, such as the Higgsfield CLI, to support the building process.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 05:33 AM