design-pro

Pass

Audited by Gen Agent Trust Hub on Nov 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill allows users to provide a URL for an 'audit' of an existing design (Section 0 and 'Fix an existing design' section). This represents an attack surface where untrusted data from external websites is ingested into the agent's context during the audit process.
  • Ingestion points: The $ARGUMENTS parameter and user-provided URLs in the 'Fix an existing design' workflow, where the agent is instructed to read the site content and screenshots.
  • Boundary markers: The instructions do not define clear delimiters or specific warnings for the agent to ignore instructions embedded within the audited content (e.g., hidden HTML comments or text designed to hijack the agent).
  • Capability inventory: The skill has capabilities to write and modify project files (Steps 3 and 5), execute shell commands for browser automation and screenshot capture (described in references/screenshots.md), and perform network operations via browser tools.
  • Sanitization: There is no evidence of sanitization or filtering applied to the content retrieved from external URLs before it is processed by the agent's reasoning engine.
  • [EXTERNAL_DOWNLOADS]: The skill provides an installation command (npx skills add Melvynx/ai-design-pro) to fetch additional components from the author's repository. This is a standard mechanism for extending the skill's capabilities and points to the vendor's own resources.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 20, 2026, 02:18 PM