The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill provides excellent guidance on protecting CloudKit management and user tokens. It explicitly warns against using the --token flag in commands to prevent exposure in logs or shell history, recommending secure alternatives like the macOS Keychain or environment variables. It also identifies standard configuration paths such as ~/.config/cktool without encouraging insecure access.
[COMMAND_EXECUTION]: The instructions enable the agent to perform CloudKit administrative tasks using the standard xcrun cktool utility. Security is maintained through a workflow that requires validation before mutation and mandatory dry-runs for destructive actions. It also mandates explicit user authorization for production writes or schema resets.
[INDIRECT_PROMPT_INJECTION]: The skill manages the attack surface where untrusted data (CloudKit record content or schema files) enters the agent context. It mitigates this by instructing the agent to use structured JSON encoding via jq and avoiding manual string interpolation of external content into prompts or CLI commands.
[EXTERNAL_DOWNLOADS]: The skill references only official Apple developer documentation and sample repositories on GitHub. These links target a well-known, trusted organization and do not introduce unverifiable dependencies or remote code execution risks.

apple-cktool