The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several standard shell tools including curl, sed, jq, ruby, gradle, xcodebuild, and dotnet. These tools are used for legitimate development tasks such as fetching SDK version metadata from remote APIs, verifying the local environment setup, and performing clean builds to validate successful SDK integration.
[EXTERNAL_DOWNLOADS]: Fetches SDK version tags and metadata from trusted external sources, specifically Google's Android Maven repository and official GitHub repositories belonging to the Google Ads organization. These operations are part of the documented process to ensure the latest stable version of the SDK is correctly identified.
[DYNAMIC_EXECUTION]: Instructs the agent to generate and execute a temporary Ruby script using the xcodeproj library. This is a common automation technique used to programmatically modify iOS project files to include Swift package dependencies, replacing manual Xcode configuration steps.
[INDIRECT_PROMPT_INJECTION]: The skill operates on user-provided project configuration files and environment data, which represents a potential surface for indirect injection if project files or external metadata were maliciously crafted.
Ingestion points: Reads local project files such as .xcodeproj directories, Info.plist, and Packages/manifest.json from the user's workspace.
Boundary markers: No explicit boundary markers or delimiters are defined; the skill relies on specific, task-oriented workflow steps.
Capability inventory: Contains capabilities to execute shell commands, invoke build systems (Gradle, xcodebuild, dotnet), and run scripts in the local environment.
Sanitization: No explicit validation or escaping is applied to the version strings fetched from remote APIs or project data read from the local filesystem before being processed by subsequent commands.

google-mobile-ads-get-started