MCP OAuth Cloudflare

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill interacts with official and trusted Google OAuth 2.0 endpoints (accounts.google.com and googleapis.com) for authentication and user profile retrieval, which is the primary purpose of the skill.- [SAFE]: Sensitive credentials such as Google Client IDs and Secrets are managed through Cloudflare's secret management system rather than being hardcoded in the source code.- [SAFE]: The provided templates implement industry-standard security controls, including HttpOnly cookies with the __Host- prefix for CSRF protection, session binding via SHA-256 hashing to prevent token theft, and one-time-use state management stored in Cloudflare KV.- [SAFE]: Input sanitization is applied to external data retrieved from OAuth providers (e.g., client names and logos) before rendering them in the HTML-based approval dialogs to mitigate potential injection or XSS risks.- [SAFE]: The skill relies on established, reputable dependencies such as the official Model Context Protocol SDK, Cloudflare's OAuth provider package, and the Hono web framework.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 03:49 AM