building-mcp-server-on-cloudflare
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides templates for building tools that ingest data from external APIs and user-supplied parameters, establishing a vulnerability surface for indirect prompt injection.
- Ingestion points: Tool parameters defined in
SKILL.mdand responses from external sources viafetchcalls. - Boundary markers: The provided examples do not explicitly demonstrate the use of delimiters to isolate untrusted content within the agent's context.
- Capability inventory: The generated servers can perform network requests, database operations (D1), and key-value store access (KV).
- Sanitization: The skill heavily promotes input validation using Zod and provides specific sanitization functions for HTML and URLs in
references/oauth-setup.mdto mitigate cross-site scripting and other injection attacks. - [EXTERNAL_DOWNLOADS]: The skill provides instructions for downloading project templates and configuration from Cloudflare's official GitHub repositories.
- Source: Official Cloudflare templates and repositories including
github.com/cloudflare/ai/demos/*andgithub.com/cloudflare/mcp-server-cloudflare.
Audit Metadata