ghl-page

Warn

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill instructs the agent to read sensitive environment files at ~/.claude/ad-profiles/.env to retrieve the GoHighLevel Private Integration Token. Accessing files containing credentials introduces a risk of sensitive data exposure within the agent's context.
  • [REMOTE_CODE_EXECUTION]: The skill recommends the execution of external code via the command npx impeccable install. This involves downloading and running a package from a public registry that is not associated with a trusted organization or well-known service.
  • [EXTERNAL_DOWNLOADS]: The skill refers to an optional plugin frontend-design@claude-plugins-official from a trusted organization for design assistance.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes brand profile markdown files without delimiters or sanitization, creating a surface for indirect prompt injection. Ingestion points: Brand profile markdown files. Boundary markers: None. Capability inventory: Generates HTML/JS code and reads filesystem. Sanitization: None.
  • [DYNAMIC_EXECUTION]: The generated landing page code uses JavaScript to dynamically inject CSS styles at runtime to bypass platform limitations that strip static style tags.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 14, 2026, 08:24 AM
Security Audit — agent-trust-hub — ghl-page