The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a variety of platform-specific commands to audit host security. This includes network inspection via lsof and ss, firewall status checks using socketfilterfw, pfctl, ufw, and firewall-cmd, and system encryption verification through fdesetup and Get-BitLockerVolume. It also utilizes vendor-specific commands like openclaw security audit, openclaw gateway status, and openclaw doctor to perform deeper audits of the OpenClaw environment.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon the output of local system commands and security reports, which serves as a potential attack surface if the underlying host or tools were to produce malicious content.
Ingestion points: The agent parses the standard output from system diagnostic tools (e.g., lsof, systemctl) and vendor audit tools (openclaw security audit).
Boundary markers: While the skill provides a specific "Report shape" for the agent to follow, it does not explicitly define character-level delimiters for the command output it processes.
Capability inventory: The skill has the capability to execute shell commands for both inspection (read-only) and host hardening (state-changing write actions), though the latter is gated by user confirmation.
Sanitization: The instructions do not specify sanitization or escaping protocols for the tool output before it is processed by the agent.

healthcheck