pencil-to-code
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- PROMPT_INJECTION (LOW): Vulnerability to Indirect Prompt Injection (Category 8).
- Ingestion points: The skill ingests untrusted data from external .pen files using the
mcp__pencil__batch_gettool as defined in the workflow ofSKILL.md. - Boundary markers: There are no boundary markers or 'ignore' instructions specified in the prompt logic to differentiate between design data and potential malicious instructions embedded in the design's text nodes.
- Capability inventory: The skill allows the agent to generate and potentially write React component files (
components/[ComponentName].tsx) and CSS files (app/globals.css), which represents a filesystem write capability that could be exploited. - Sanitization: The mapping rules in
SKILL.mdandreferences/node-mapping.mddo not include any sanitization or validation logic to prevent the inclusion of executable scripts or injection payloads in the generated code.
Audit Metadata