pencil-to-code
Pass
Audited by Gen Agent Trust Hub on Mar 12, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes content from external .pen design files and interpolates it directly into generated React/TypeScript code. This creates a surface for indirect prompt injection where malicious instructions embedded in design labels or text could influence the agent's output or the resulting application.\n
- Ingestion points: Design data is ingested in
SKILL.mdusing themcp__pencil__batch_getandmcp__pencil__get_variablestools to read local .pen files.\n - Boundary markers: There are no boundary markers or instructions to ignore embedded commands when processing content from the design nodes.\n
- Capability inventory: The skill generates executable React/TypeScript code and Tailwind configurations which are presented to the user or downstream processes.\n
- Sanitization: There is no evidence of escaping or validation of the text content retrieved from the design file before it is placed into code templates.
Audit Metadata