web-design-guidelines

Warn

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS] (LOW): The skill fetches instructions from a remote GitHub repository. * Evidence: The skill 'Fetch the latest guidelines' from a raw GitHub URL using WebFetch. * Source: https://raw.githubusercontent.com/vercel-labs/web-interface-guidelines/main/command.md. * Trust Check: The organization 'vercel-labs' is a Trusted Source, downgrading this finding to LOW per global rules.
  • [PROMPT_INJECTION] (MEDIUM): The skill is vulnerable to Indirect Prompt Injection due to its reliance on external instructions and untrusted data ingestion. * Ingestion points: Remote 'command.md' file and local UI code files (SKILL.md). * Boundary markers: Absent; there are no instructions to isolate the guidelines or the user code from the agent's executive function. * Capability inventory: The agent possesses 'read' access to local files and provides structured external output. * Sanitization: Absent; the skill accepts 'all rules and output format instructions' directly from the fetched file. * Reasoning: If the remote guidelines are compromised or the files being audited contain malicious prompts, the agent may obey those instructions instead of its original mandate.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 16, 2026, 06:03 AM