The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes system tools including openclaw, tailscale, launchctl, stat, and du to gather diagnostic information across 10 defined domains. These commands are limited to status checks and metadata retrieval.
[SAFE]: The skill strictly enforces a read-only protocol, instructing the agent to only recommend fixes rather than applying them directly, which minimizes the risk of accidental system modification.
[SAFE]: Security-related checks focus on verifying file permission metadata (e.g., ensuring configuration files are set to 600 or 700) to confirm that sensitive information is properly protected on the local filesystem.

openclaw-doctor