skills/raddue/crucible/ui-verify/Gen Agent Trust Hub

ui-verify

Pass

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it ingests and prioritizes content from mockup files (e.g., docs/mockups/*.html) to guide its actions. An attacker who can modify these files could embed instructions that the agent might follow during the verification or 'fix' phases.
  • Ingestion points: Reads mockup HTML/CSS from docs/mockups/ and implementation source code (USS, C#).
  • Boundary markers: None present; the skill explicitly instructs the agent to 're-read' the mockup and follow it over prior context.
  • Capability inventory: The agent has the ability to take screenshots via mcp__UnityMCP__manage_scene, read arbitrary project files, and modify implementation files to 'fix' detected deltas.
  • Sanitization: No validation or sanitization of the mockup content is performed before the agent interprets it as a source of truth.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 10, 2026, 02:18 PM