design-import

Pass

Audited by Gen Agent Trust Hub on Aug 4, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted content from external design bundles to guide component scaffolding, creating an indirect prompt injection surface.
  • Ingestion points: Processes README.md and chats/*.md files extracted from a user-supplied URL or local tarball.
  • Boundary markers: No delimiters or safety instructions are used when the subagent reads the bundle content.
  • Capability inventory: The skill and its subagents use Read, Write, Edit, Bash, WebFetch, and Agent tools.
  • Sanitization: No validation or sanitization of the text content within the bundle is performed before it is used to influence code generation and filesystem operations.
  • [EXTERNAL_DOWNLOADS]: The skill downloads data from arbitrary, user-provided URLs using the WebFetch tool. While documented for use with claude.ai, the implementation allows any HTTP/HTTPS source.
  • [COMMAND_EXECUTION]: The skill executes shell commands to process downloaded assets. It uses Bash to run tar -xzf on the downloaded file to extract its contents to a temporary directory.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 4, 2026, 01:09 AM
Security Audit — agent-trust-hub — design-import