paste-up
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from Paper design files and project documentation which could potentially contain malicious instructions.
- Ingestion points: Data is ingested from Paper URLs, project specifications, and tracker context via the Paper MCP connection (SKILL.md).
- Boundary markers: The instructions mandate recording and carrying boundaries for in-scope artboards and permitted edits throughout the session (SKILL.md).
- Capability inventory: The skill can modify design files using tools such as
write_html,update_styles,set_tokens, andrename_nodes(SKILL.md). - Sanitization: No specific sanitization logic is provided for external content, relying instead on agent-directed boundary enforcement.
Audit Metadata