The Agent Skills Directory

[COMMAND_EXECUTION]: The documentation provides numerous shell commands for system management, including service controls (systemctl restart dnsmasq), network configuration (iptables, wg syncconf), and file cleanup operations (find ... -delete).
[DATA_EXFILTRATION]: The skill identifies the location of sensitive files, including the root API key at /opt/ludus/install/root-api-key and system configuration at /opt/ludus/config.yml.
[PRIVILEGE_ESCALATION]: Remediation steps involve high-privilege operations such as using chroot to access environments, modifying kernel debug controls, and editing Perl scripts in system directories.
[PERSISTENCE]: The skill includes instructions for creating an automated maintenance task using a cron job (0 2 * * 0 find ... -delete) to periodically clean up template cache files.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data in the form of user-provided error logs and status outputs.
Ingestion points: User-provided error messages, context gathering, and log review steps defined in SKILL.md.
Boundary markers: There are no explicit delimiters or instructions for the agent to treat diagnostic data as passive text rather than active commands.
Capability inventory: The troubleshooting reference contains numerous high-privilege commands and sensitive file access paths.
Sanitization: The skill does not provide instructions for validating or sanitizing diagnostic logs before they are processed by the agent.

troubleshooting