The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a variety of Linux system diagnostic commands to investigate performance, processes, and network state.
Evidence: Commands listed include uptime, hostnamectl, dmesg, top, free, df, iostat, ps aux, pstree, lsof, strace, journalctl, ip addr, ss, curl, dig, and systemctl.
[PROMPT_INJECTION]: The workflow establishes a surface for indirect prompt injection by having the agent ingest and analyze system logs that could contain malicious strings or instructions.
Evidence: Phase 4 ("Log Analysis") directs the agent to read output from journalctl, /var/log/syslog, and other files in /var/log/, and then apply the @error-detective skill for analysis.
Ingestion points: SKILL.md (Phase 4: Log Analysis).
Boundary markers: No delimiters or instructions to ignore embedded commands are present in the log analysis phase.
Capability inventory: The skill has extensive capabilities including shell command execution (@bash-linux, @bash-pro) and service management (@server-management).
Sanitization: No sanitization or validation of log content is specified before the agent processes it.

linux-troubleshooting