web-to-design-md

Warn

Audited by Gen Agent Trust Hub on Jul 18, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill directs the user to install a companion tool from an external GitHub repository: https://github.com/Paidax01/web-to-design-md. This source is not from a recognized trusted organization.
  • [COMMAND_EXECUTION]: The instructions include a shell command (npx skills add) to install the external dependency.
  • [REMOTE_CODE_EXECUTION]: Installing and executing third-party code from an unverified repository is a remote code execution risk, as the external code is not audited and can be changed at any time by its maintainer.
  • [PROMPT_INJECTION]: The skill has an indirect prompt injection attack surface because it processes content from external websites via user-provided URLs.
  • Ingestion points: External URLs provided for design token extraction (referenced in SKILL.md).
  • Boundary markers: Absent; there are no instructions to ignore embedded commands or delimiters used for the scraped content.
  • Capability inventory: The installed tool scrapes website content and writes output to a local file (design.md).
  • Sanitization: Absent; the skill does not implement any validation or filtering of the ingested website content.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 18, 2026, 08:43 AM
Security Audit — agent-trust-hub — web-to-design-md