cloudflare-email-service
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill describes an architecture for agents to process incoming emails, which represents a potential indirect prompt injection surface.
- Ingestion points: The
email()handler defined inreferences/routing.mdandonEmail()inreferences/sending.mdingest untrusted MIME content from external senders. - Capability inventory: The skill documentation includes instructions for data persistence via Durable Objects SQLite (
references/routing.md) and network operations via email sending bindings (references/sending.md). - Boundary markers: The provided code snippets for parsing and processing emails do not implement specific delimiters or instructions to ignore embedded commands within the email body.
- Sanitization: The skill does not provide explicit examples of sanitizing or escaping email content before it is processed by an AI agent.
- [EXTERNAL_DOWNLOADS]: The documentation references official Cloudflare repositories and resources for configuration and guidelines.
- Evidence: It suggests fetching configuration from
github.com/cloudflare/agentsandgithub.com/cloudflare/mcp. - [SAFE]: The skill explicitly warns against hardcoding API tokens and recommends using environment variables or Cloudflare secrets for secure credential management.
Audit Metadata