fail2ban-swag

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/fail2ban-swag.sh performs remote command execution using SSH and docker exec. The docker_exec function in the script uses $* to pass arguments, which allows for shell metacharacter injection if an argument (such as a jail name) is maliciously crafted.\n- [PROMPT_INJECTION]:\n
  • The skill provides an indirect prompt injection surface by reading remote log files that contain untrusted data from network requests.\n
  • Ingestion points: Remote logs from Nginx and Fail2Ban are read and displayed using the logs, nginx-access-log, and search-ip commands in scripts/fail2ban-swag.sh.\n
  • Boundary markers: The script uses simple text headers but lacks explicit delimiters or instructions to prevent the agent from following commands embedded in the log entries.\n
  • Capability inventory: The skill can modify firewall rules, edit remote configuration files, and execute arbitrary commands via Docker.\n
  • Sanitization: No sanitization is performed on log content before it is processed by the agent.\n
  • SKILL.md includes instructions attempting to override agent behavior with 'MANDATORY' invocation requirements.\n- [DATA_EXFILTRATION]: The backup command in scripts/fail2ban-swag.sh transfers configuration archives from the remote host to the local environment using scp.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 05:33 PM