qa-web-test

Pass

Audited by Gen Agent Trust Hub on Mar 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes content from external URLs provided as TARGET_URL. Malicious content on a target page could attempt to subvert the agent's instructions.\n
  • Ingestion points: Web content from TARGET_URL is read during navigation and inspection steps (SKILL.md, steps/step-01-connect.md).\n
  • Boundary markers: Absent; there are no specific delimiters or instructions to ignore embedded commands found within the web content.\n
  • Capability inventory: The skill can execute Bash commands for directory creation and use the Write tool for generating reports and saving screenshots (steps/step-05-report.md).\n
  • Sanitization: No explicit sanitization or validation of the ingested web content or resulting identifiers is performed before use in file operations or reports.\n- [COMMAND_EXECUTION]: The skill uses the Bash tool to execute mkdir -p commands. These commands use the OUT_DIR variable, which is derived from user-provided arguments. This could potentially lead to directory traversal or unintended command execution if the input is not strictly validated by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 18, 2026, 09:15 AM