The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a Bash script to execute standard macOS system utilities including ps, launchctl, find, and security to collect status information about local processes and services.
[DATA_EXPOSURE]: The script reads local configuration and log files (health.json, chunker.log) to provide status updates to the user. This is consistent with its stated purpose as a status monitor.
[DATA_EXPOSURE]: It queries the macOS Keychain for specific generic passwords. The script uses the -w flag (which outputs the password) but correctly redirects all output (stdout and stderr) to /dev/null. This allows the script to verify the existence of the credentials for reporting purposes without exposing the actual secrets to the agent or the logs.
[PROMPT_INJECTION]: The skill processes untrusted data by reading the tail of a log file and file paths from the disk (Category 8). This presents a theoretical indirect prompt injection surface if an attacker could write malicious instructions into the logs or file names. However, since the skill's capability is limited to local status reporting and it lacks network exfiltration tools, the risk is minimal.

daemon-status