The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documents methods for receiving and processing inbound emails, which is a common vector for untrusted data to enter an agent's context. The skill mitigates this risk by including explicit security guardrails (ASI05) that instruct developers to sanitize external content, treat HTML as a hostile surface, and implement human-in-the-loop patterns for reviewing automated replies. Evidence: Ingestion points in references/routing.md (email() handler) and references/sending.md (Agents SDK onEmail()).
[EXTERNAL_DOWNLOADS]: The skill references official Cloudflare developer documentation, well-known Node.js packages, and established GitHub repositories. These are recognized as trusted or well-known service providers and do not present a security risk. Evidence: developers.cloudflare.com, github.com/cloudflare/agents, and npmjs.com/package/@cloudflare/workers-types.

cloudflare-email-service