blueprint
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill allows fetching fonts from Google Fonts specifically when the user requests a 'handwritten' style. This uses a well-known service for legitimate stylistic purposes.
- [REMOTE_CODE_EXECUTION]: The skill includes a validation script (
scripts/validate.py) that employs theplaywrightlibrary to launch a Chromium browser for rendering checks. This is used as a quality assurance step for the generated HTML blueprints. - [DYNAMIC_EXECUTION]: The instructions suggest that the agent generate SVG markup using computed coordinates, which may involve the creation and execution of small scripts by the agent to perform mathematical calculations.
- [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for instructions embedded in user-supplied drawing topics.
- Ingestion points: User-provided topics and descriptions for the blueprint are processed by the agent.
- Boundary markers: The skill instructions emphasize a structured planning phase and the use of a fixed template.
- Capability inventory: The skill has the capability to write HTML files and execute the local validation script.
- Sanitization: The
validate.pyscript automatically checks the final output for unauthorized external network requests and runtime errors.
Audit Metadata