The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill describes how to ingest data from external system surfaces (Siri, Shortcuts) and pass it to the application's web view. This creates a potential surface for indirect prompt injection or malicious navigation. The skill includes explicit security guidance to treat this input as untrusted and validate it against a whitelist of known routes before use. \n
Ingestion points: references/web-bridge.md (via the PendingRouteStore class and AppIntentsBridgePlugin methods). \n
Boundary markers: The architecture uses a dedicated native bridge to pass intent data to the web layer. \n
Capability inventory: The skill involves file creation and modification within the iOS project and execution of build tools. \n
Sanitization: The skill provides clear instructions in references/web-bridge.md: 'Validate route against known routes before you navigate; treat it as untrusted input like any deep link.' \n- [COMMAND_EXECUTION]: The skill instructs the user to execute standard Capacitor and iOS build commands for project synchronization and verification. \n
Evidence: Commands such as npx cap sync ios and xcodebuild are documented in SKILL.md and references/project-setup.md.

capacitor-app-intents