qt-form-and-input

Pass

Audited by Gen Agent Trust Hub on Apr 7, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill establishes an indirect prompt injection surface by reading external data from the clipboard and UI elements (via clipboard read, text, and tree commands). It does not include boundary markers or sanitization procedures for this ingested data. The skill possesses capabilities for writing to the file system (via screenshots) and executing arbitrary commands in a virtual machine environment.
  • [DATA_EXFILTRATION]: Functional instructions facilitate the reading of clipboard contents and the capture of screen images, which can be used to handle sensitive information during automation tasks.
  • [COMMAND_EXECUTION]: Uses the qt-ai-dev-tools CLI to perform UI automation and provides the vm run command for executing arbitrary shell commands within the host virtual machine.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 7, 2026, 10:19 PM
Security Audit — agent-trust-hub — qt-form-and-input