The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection by processing untrusted data from project configuration files to determine migration tracks and resolve parameters.
Ingestion points: The skill logic in SKILL.md and references/quick-start.md requires the agent to inspect package.json and app.json for classification. Additionally, references/bare-android-aar-generation.md directs the agent to resolve placeholders like <android_module_name> from project files.
Boundary markers: Absent. The instructions provide no delimiters or specific warnings to ignore instructions embedded within the processed data.
Capability inventory: The skill allows the execution of shell commands such as npm install, npx brownfield, pod install, and npx create-expo-app across several files (e.g., references/bare-quick-start.md, references/expo-create-app.md).
Sanitization: Absent. There is no described process for filtering or validating content read from files before it is used in agent operations.
[COMMAND_EXECUTION]: The skill instructs the agent to perform multiple shell-based operations essential for the migration workflow.
Evidence: Use of npm install, npx brownfield, and pod install for setup and artifact packaging.
Evidence: Recommendation to disable Xcode User Script Sandboxing (ENABLE_USER_SCRIPT_SANDBOXING=NO) in references/bare-ios-xcframework-generation.md to facilitate React Native build scripts.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to fetch external packages and documentation.
Evidence: Installation of @callstack/react-native-brownfield and create-expo-app from the npm registry.
Evidence: References to documentation hosted on oss.callstack.com.

react-native-brownfield-migration