The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various system-level discovery commands (e.g., ss, lsof, ufw status, pfctl, uname) to assess network exposure and host configuration.
[COMMAND_EXECUTION]: Utilizes the vendor-owned openclaw CLI to perform security audits, apply fixes, and manage scheduled tasks via openclaw cron.
[CREDENTIALS_UNSAFE]: Accesses sensitive host security configurations, specifically evaluating SSH access methods and disk encryption status (LUKS, FileVault, BitLocker).
[EXTERNAL_DOWNLOADS]: Checks for the latest version of the openclaw package using npm view, which involves a network request to the npm registry.
[PROMPT_INJECTION]: The skill exhibits vulnerability to indirect prompt injection by processing external system data and command outputs without sanitization.
Ingestion points: Command outputs from network tools and openclaw security audits are used to drive the remediation workflow.
Boundary markers: Lacks explicit delimiters or instructions to treat ingested tool output as untrusted data.
Capability inventory: Possesses significant capabilities including modifying firewall rules, changing SSH configurations, and establishing persistence via cron jobs.
Sanitization: No evidence of sanitization or strict validation for data ingested from the host environment before it influences subsequent actions.

healthcheck