sqlmap-database-pentesting
Fail
Audited by Gen Agent Trust Hub on Mar 20, 2026
Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides numerous command-line patterns for the SQLMap tool, a utility used for the automated detection and exploitation of SQL injection vulnerabilities.
- [REMOTE_CODE_EXECUTION]: Documentation includes instructions for high-risk flags such as
--os-shelland--os-cmd, which allow for the execution of arbitrary operating system commands on the remote database server. - [DATA_EXFILTRATION]: The skill facilitates the bulk extraction of database tables and the reading of sensitive local system files, such as
/etc/passwd, from the target server using the--file-readflag. - [CREDENTIALS_UNSAFE]: The skill explicitly focuses on harvesting usernames, passwords, and cryptographic hashes from target systems, including instructions for automated password cracking.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from URLs, HTTP request files, and logs without visible sanitization or boundary markers. 1. Ingestion points: Target URLs (-u), HTTP request files (-r), and log files (-l). 2. Boundary markers: Absent. 3. Capability inventory: High-impact operations including OS shell access, file read/write, and database dumping. 4. Sanitization: No input validation or escaping procedures are defined in the instructions.
Recommendations
- AI detected serious security threats
Audit Metadata