build-design-system

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [COMMAND_EXECUTION]: The skill documents the use of external tools like extract-page-data and clone-page to gather design tokens. However, the provided script scripts/generate.mjs only performs local file operations (reading JSON, copying images, writing text files) and does not execute shell commands or spawn subprocesses.
  • [DATA_EXPOSURE]: The script reads a local theme.json and meta.json. It does not access sensitive system paths, credentials, or environment variables. All data processing is confined to the input directory provided by the user.
  • [REMOTE_CODE_EXECUTION]: There is no evidence of remote script downloading (curl|bash) or dynamic execution (eval/exec) of untrusted external content. The script is a static generator using predefined templates.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from theme.json (extracted from external websites). While this is an attack surface, the risk is minimal because the output is a static Markdown file (DESIGN.md) and a CSS file (globals.css) intended for human/AI review. The skill does not have capabilities like network access or file-writing outside its specified output directory to facilitate an exploit chain.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 10:07 PM
Security Audit — agent-trust-hub — build-design-system