mesop

Warn

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The skill references and encourages the use of external scripts and stylesheets from CDNs. Evidence: Documentation in web-components.md and styling-and-layouts.md for loading JS from cdn.jsdelivr.net and external CSS. Risk: This is often combined with the recommendation to use dangerously_disable_trusted_types=True, which bypasses critical browser-level XSS protections.
  • [COMMAND_EXECUTION] (LOW): Provides standard instructions for running and deploying applications via CLI. Evidence: mesop main.py and Docker build commands in deployment-and-config.md.
  • [INDIRECT_PROMPT_INJECTION] (MEDIUM): The framework inherently handles untrusted data which can influence agent behavior. Ingestion points: me.input, me.textarea, and me.uploader components in components-reference.md. Boundary markers: Not mentioned or enforced in the guidance. Capability inventory: The skill handles event logic that can involve API calls or file operations. Sanitization: Not explicitly addressed; components like me.markdown can render malicious content from untrusted sources.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 16, 2026, 08:57 AM