paper-design

Pass

Audited by Gen Agent Trust Hub on Sep 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses local scripts for setup and diagnostics. Specifically, scripts/setup.sh executes claude mcp add to register the MCP server, and scripts/health-check.py uses subprocess.run to call pgrep and claude mcp list for environment verification.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted HTML/CSS data, creating a potential surface for indirect prompt injection. Ingestion points include the write_html and get_jsx tools which process external design structures. The skill has the capability to interact with the file system and network via the registered MCP server. No explicit sanitization, filtering, or boundary markers for the ingested DOM content are described.
  • [EXTERNAL_DOWNLOADS]: The skill refers to external software and libraries associated with the tool, such as the @paper-design/shaders-react Node.js package and the Paper Desktop application available at paper.design/downloads.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 16, 2026, 05:48 AM
Security Audit — agent-trust-hub — paper-design