deterministic-design

Pass

Audited by Gen Agent Trust Hub on Aug 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to use standard local server tools such as 'npx serve' or 'python3 -m http.server' to host UI files for evaluation.
  • [PROMPT_INJECTION]: The skill analyzes external web content, which presents a surface for indirect prompt injection.
  • Ingestion points: The agent processes DOM snapshots and visual data via 'layout-audit.js' and browser screenshots.
  • Boundary markers: No explicit markers are used to separate layout data from potential malicious text content.
  • Capability inventory: The skill utilizes browser interaction capabilities to gather audit data.
  • Sanitization: No sanitization of ingested page content is performed before analysis.
  • [SAFE]: The 'layout-audit.js' script is a helper utility for calculating spatial metrics. It uses standard DOM APIs to measure element geometry and provides visual annotations via a browser overlay. No suspicious network operations or data exfiltration logic were found.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 23, 2026, 04:33 PM
Security Audit — agent-trust-hub — deterministic-design