The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes detailed playbooks (cleanup-tiers.md) for reclaiming disk space through commands like rm -rf and sudo. These operations are presented as tiered options with explicit instructions for dry-runs and user confirmation to prevent accidental data loss.
[PERSISTENCE_MECHANISMS]: To provide proactive monitoring, the skill implements a LaunchAgent (com.local.mac-health-check.plist) that executes a health-check script every 5 minutes. This persistence is a core, documented feature of the alerting system.
[DATA_EXPOSURE_AND_EXFILTRATION]: The monitoring script accesses disk usage data and diagnostic reports (/Library/Logs/DiagnosticReports/) to identify potential kernel panic signals. It can optionally send alerts to a user-provided ntfy.sh URL for remote notifications.
[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill recommends installing third-party tools mole and alerter from GitHub repositories. These dependencies are used for safety-validated cleanup and notification delivery in a macOS environment.
[INDIRECT_PROMPT_INJECTION]: The skill polls filenames in system diagnostic directories for specific failure codes. This represents an external data surface, but the risk is mitigated because the script only performs string matching and does not execute file contents.
[PROMPT_INJECTION]: No instructions were found that attempt to override agent behavior, bypass safety filters, or extract system prompts.

maintaining-macos-health