query-builder

Pass

Audited by Gen Agent Trust Hub on Jul 3, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: No patterns of instruction override, safety filter bypass, or system prompt extraction were detected in the skill instructions or metadata.
  • [DATA_EXFILTRATION]: The skill does not access sensitive local files (like .env, .ssh, or cloud credentials) and performs no unauthorized network operations. Use of database tools is restricted to standard query building.
  • [REMOTE_CODE_EXECUTION]: No remote scripts are downloaded or executed. The skill does not involve package managers (npm, pip) or dynamic code execution patterns.
  • [COMMAND_EXECUTION]: There are no instances of shell command execution, privilege escalation (sudo), or persistence mechanisms (cron, startup scripts).
  • [CREDENTIALS_UNSAFE]: No hardcoded secrets, API keys, or private keys are present. The skill correctly recommends managing database connections via framework-standard services.
  • [SAFE]: The skill is purely educational, providing best practices for secure database interactions and contrastive examples to help developers avoid common SQL injection vulnerabilities.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 3, 2026, 04:16 PM
Security Audit — agent-trust-hub — query-builder