Agents are
first-class citizens

# Install, then set up an app in five commands. Every command takes --json and has stable exit codes.
$ curl -fsSL https://eml.dev/install.sh | sh
$ eml login                       # paste a key, or set EML_API_KEY
$ eml domains add acme.com        # prints the DNS records to publish
$ eml domains check acme.com      # looks them up in public DNS
$ eml doctor                      # key, credits, domains, DNS, SMTP: what's left, and the fix
$ eml test --to you@acme.com      # sends a test email and waits for delivery

# Then wire it into the app, and watch it work.
$ eml env --reveal >> .env        # EML_API_KEY plus SMTP settings
$ eml snippet python              # or curl, node, ruby, go, rails, nodemailer, django
$ eml emails tail                 # live feed of sends, deferrals, deliveries and bounces

curl -X POST https://eml.dev/api/v1/emails \
  -H "Authorization: Bearer $EML_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "from": "agent@acme.com",
    "to": "sam@example.com",
    "subject": "Your refund is on its way",
    "body": "Hi Sam, I have issued the refund for order #1234."
  }'

# 202 Accepted
{
  "message_id": "0199c1…@eml.dev",
  "status": "queued",
  "sandbox": false,
  "recipients": ["sam@example.com"]
}

// Hand this to any model that supports tool use; the handler is one POST.
{
  "name": "send_email",
  "description": "Send an email to one recipient via eml.dev",
  "input_schema": {
    "type": "object",
    "properties": {
      "to":        { "type": "string" },
      "subject":   { "type": "string" },
      "body":      { "type": "string" },
      "html_body": { "type": "string" }
    },
    "required": ["to", "subject"]
  }
}

# Same key, any SMTP library. Works where the agent already sends mail.
Host:     smtp.eml.dev
Port:     587
Security: STARTTLS
Username: apikey
Password: $EML_API_KEY

# eml doctor checks that the port is reachable from where you run;
# eml env prints all of this as env vars.