atelier
Pass
Audited by Gen Agent Trust Hub on May 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill implements an orchestration pipeline that ingests data from external sources, creating a surface for indirect prompt injection (Category 8).
- Ingestion points: The skill scans the project codebase for Tailwind/CSS configurations and reads design direction documents (e.g.,
direction.mdfrom the Vision agent) or user-supplied briefs. - Boundary markers: It utilizes the structured
DESIGN_INTENT_HANDOFFschema to pass data between agents, which provides a level of separation between data and instructions. - Capability inventory: The orchestrator has the ability to delegate tasks to a wide range of sub-agents (Artisan, Forge, etc.) and read/write files within the
.agents/directory. - Sanitization: While the skill enforces schema validation for handoffs, it does not explicitly describe sanitization or filtering logic for the ingested design tokens or instructions.
Audit Metadata