auditing-game-screen-readability

Pass

Audited by Gen Agent Trust Hub on Oct 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill provides a legitimate framework for UI auditing. The Node.js script (check-screen-frames.mjs) is a local utility that performs geometric and contrast calculations on user-provided data without network access or privileged operations. The procedure is manual and transparent.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external game data which constitutes a potential injection surface.
  • Ingestion points: The skill ingests manifest.json files and captured game screenshots for analysis in steps 3 and 4.
  • Boundary markers: No explicit boundary markers or instructions to ignore data-embedded directives are specified for the ingested content.
  • Capability inventory: Capabilities are limited to local file reading (readFileSync) and visual reasoning by the agent. The Node.js script does not have network access.
  • Sanitization: Standard JSON parsing is used for the manifest file. No specific mitigations for adversarial visual patterns in frames are described.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 4, 2026, 01:22 PM