The Agent Skills Directory

[PROMPT_INJECTION]: No prompt injection patterns were detected. The instructions are focused on providing technical guidance for watchOS development without attempting to override agent safety protocols or system behavior.
[DATA_EXPOSURE]: No hardcoded credentials, sensitive file paths, or data exfiltration patterns were found. The skill explicitly promotes security best practices, such as using the .privacySensitive() modifier in SwiftUI to redact sensitive information (e.g., account balances) during Always On display states.
[OBFUSCATION]: No obfuscation techniques, such as Base64 encoding, zero-width characters, or homoglyphs, were identified in the text or code snippets.
[REMOTE_CODE_EXECUTION]: The skill does not contain instructions to download or execute code from untrusted remote sources. It references standard Apple developer tools like devicectl and official Apple documentation (Tech Notes, WWDC sessions).
[PRIVILEGE_ESCALATION]: No commands for privilege escalation (e.g., sudo) or excessive permission modifications were found.
[PERSISTENCE]: The skill does not attempt to establish persistence through shell profiles, cron jobs, or system services.
[METADATA_POISONING]: The skill's metadata (name, description, author) is accurate and consistent with its contents.
[INDIRECT_PROMPT_INJECTION]: The skill does not ingest or process untrusted external data in a manner that would expose the agent to indirect prompt injection. It primarily serves as a static reference for developers.
[DYNAMIC_CONTEXT_INJECTION]: The skill does not utilize the !command syntax or any other mechanism for executing shell commands at load time.
[SAFE_PRACTICES]: The skill demonstrates high awareness of platform-specific security and privacy features, including guidance on Private Cloud Compute (PCC) for AI features on watchOS 27, ensuring network-dependent AI tasks are handled securely.

axiom-watchos