detecting-sql-injection-vulnerabilities

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The skill has a vulnerability surface for indirect prompt injection because it processes untrusted source code that may contain malicious instructions.
  • Ingestion points: The agent reads application source code from the local filesystem using tools like Read, Grep, and Glob.
  • Boundary markers: There are no instructions to use delimiters or ignore-previous-instructions markers when reading external files.
  • Capability inventory: The skill is permitted to write report files and execute system commands via the Bash tool.
  • Sanitization: No sanitization steps are defined for the input source code before it is interpreted by the agent.
  • [COMMAND_EXECUTION]: The skill is configured with permissions to execute bash commands, specifically for scanning and security testing purposes.
  • Evidence: The configuration defines 'allowed-tools' including 'Bash(code-scan:)' and 'Bash(security-test:)'.
  • Context: This functionality is intended to run local scripts to automate vulnerability detection.
  • [NO_CODE]: Several scripts described as bundled resources are missing from the provided skill files.
  • Evidence: The files 'sqli_scan.py', 'sqli_exploit.py', and 'sqli_remediate.py' are referenced in 'scripts/README.md' but the script contents are not included in the package.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 11:20 AM