novnc-webapp-control

Pass

Audited by Gen Agent Trust Hub on Oct 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external web application data, including DOM state and chat content, creating a potential surface for indirect prompt injection.
  • Ingestion points: The agent is instructed to read DOM structures, chat transcripts, and status markers from the target web application (SKILL.md, references/direct-browser-contract.md).
  • Boundary markers: The instructions recommend stable selectors but do not include explicit instructions to disregard commands found within the webapp content.
  • Capability inventory: The skill possesses the ability to execute local shell scripts and Node.js tools and perform file system writes for evidence capture.
  • Sanitization: No explicit sanitization or escaping of data retrieved from the browser is described.
  • [COMMAND_EXECUTION]: The skill facilitates the execution of local scripts to manage the browser environment and application workflow.
  • Evidence: The skill instructions involve running scripts/launch_studio_novnc.sh and node tools/lala-studio-browser.mjs from the project directory to initiate and control the automation process.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 19, 2026, 06:48 PM