The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and analyze source code files (Swift, Kotlin) and configuration files (.sxcu, .xsdc) from the repository. This represents an attack surface where malicious content embedded in the source code or configuration files could attempt to influence the agent's behavior.
Ingestion points: Files located in src/mobile/ios and src/mobile/android, as well as imported .sxcu and .xsdc files.
Boundary markers: The instructions do not define specific delimiters for separating data from instructions, but emphasize following existing module boundaries and domain models.
Capability inventory: The skill allows for file system read/write operations and the execution of build tools (./gradlew) and Android Debug Bridge (adb) commands.
Sanitization: No specific sanitization or filtering of external content is mentioned, which is standard for developer-oriented tools intended for internal repository use.
[COMMAND_EXECUTION]: The skill provides standard commands for building, testing, and verifying Android applications.
Evidence: Usage of ./gradlew assembleDebug, ./gradlew test, and adb commands for application installation and UI interaction. These are standard tools for Android development and are used here within their intended scope.

mobile-android-ios-parity