The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the automated processing of incoming emails, creating a surface for indirect prompt injection where an external sender could attempt to influence the agent's behavior through email content.
Ingestion points: The email() handler in references/routing.md ingests untrusted data via the message object, including headers, subject, and raw body content.
Boundary markers: The provided code examples do not include explicit boundary markers or instructions to the agent to ignore potentially malicious content within the email bodies.
Capability inventory: The skill documentation describes capabilities including sending emails (env.EMAIL.send), forwarding messages (message.forward), and storing content in Durable Object SQLite databases (ctx.storage.sql.exec) in references/routing.md.
Sanitization: No explicit sanitization or validation of the email content is present in the reference implementation snippets.
[EXTERNAL_DOWNLOADS]: The skill fetches documentation, API specifications, and type definitions from Cloudflare's official domains and GitHub repositories (e.g., developers.cloudflare.com, github.com/cloudflare/mcp). These are well-known and official sources for the service being implemented.
[COMMAND_EXECUTION]: The documentation provides instructions for using the wrangler CLI tool to manage email domain configurations, DNS records, and sending operations via npx wrangler email.

cloudflare-email-service