The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides an installation command for the Maestro CLI that pipes a remote script directly to the bash shell: curl -Ls "https://get.maestro.mobile.dev" | bash. This executes code from an external server without local verification. Evidence found in SKILL.md and references/detox-and-maestro.md.
[EXTERNAL_DOWNLOADS]: The skill downloads the Maestro CLI installer from a remote URL (https://get.maestro.mobile.dev) as part of the setup instructions provided in SKILL.md and references/detox-and-maestro.md.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources, specifically deep link URLs and push notification payloads, which can be manipulated to influence agent behavior.
Ingestion points: Data enters through the mobile: deepLink and device.sendUserNotification APIs as demonstrated in references/mobile-patterns.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are used when processing these external strings.
Capability inventory: The skill utilizes the mobile: shell command to execute system-level operations on the device, as seen in references/appium-patterns.md.
Sanitization: The instructions do not specify any validation or sanitization steps for the data passed into these testing APIs.
[COMMAND_EXECUTION]: The Appium mobile: shell command is utilized to execute shell commands directly on the mobile operating system, such as toggling airplane mode via cmd connectivity airplane-mode. This is a high-privilege capability that could be misused if provided with malicious input. Evidence found in references/appium-patterns.md and references/mobile-patterns.md.

mobile-testing