tui-lazy-load

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No security issues detected. The skill follows established patterns for defining a UI component's rendering and specification workflow. It operates by transforming input JSON models into text-based TUI representations and tool-specific design commands.
  • [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests untrusted JSON data (Input Model) and interpolates it into output commands for the Pencil MCP tool.
  • Ingestion points: Input model JSON containing properties like props, state, and style (SKILL.md).
  • Boundary markers: None present; the instructions do not specify delimiters to separate untrusted data from tool commands.
  • Capability inventory: Generates text-based commands for the Pencil MCP tool (e.g., PENCIL_BATCH_DESIGN blocks). It does not perform direct file writes, network operations, or shell execution.
  • Sanitization: No explicit sanitization or validation of input strings is mentioned. While this creates a vulnerability surface, the impact is limited to the drawing tool's environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 10:44 AM