The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documents an attack surface where instructions embedded in the mobile application UI could influence the agent's behavior during test execution.
Ingestion points: The copyTextFrom command (detailed in examples/core.md and examples/device-interactions.md) allows extracting text from UI elements (accessibility identifiers or text labels) into the maestro.copiedText variable.
Boundary markers: No delimiters or safety warnings are provided for handling content extracted from the UI.
Capability inventory: The skill possesses powerful capabilities including JavaScript execution via evalScript and runScript, network operations via the http object, and the ability to input text back into the application using inputText.
Sanitization: There are no patterns or instructions for sanitizing or validating extracted UI content before it is processed by JavaScript or used in subsequent commands.
[DYNAMIC_EXECUTION]: The skill extensively uses dynamic execution features inherent to the Maestro framework.
Evidence: The use of evalScript for inline JavaScript evaluation and runScript for executing external logic (e.g., scripts/generate-test-data.js) is documented in SKILL.md and examples/flow-control.md. While these are standard features for the tool, they allow for arbitrary logic execution at runtime.

mobile-testing-maestro