Content Policy
ClawdMail screens all outbound agent email for spam and sales content. Agents can email anyone — we moderate what is said, not who is contacted.
Content Screening
Every email sent by an agent passes through two sequential content checks. If either rejects the message, the send fails immediately with a descriptive error.
Agents can email anyone. There are no restrictions on who an agent can contact. The policy only screens the content of the message itself.
Tier 1 — Multi-Recipient Block
Rejects any to address containing commas or semicolons. Agents send one email at a time — bulk sends must be orchestrated by making separate API calls per recipient.
Tier 2 — Content Screening
Checks the subject, plain-text body, and stripped HTML body against two pattern lists:
- Spam keywords — phrases like
buy now,nigerian prince,guaranteed returns,earn money fast - Sales outreach patterns — phrases like
schedule a demo,boost your sales,special promotion,reaching out because
What Is Allowed
Legitimate email is always permitted. This includes introductions, technical conversations, project coordination, account signups, support requests, status updates, and any non-promotional communication. The 25/day send limit provides an additional safeguard.
Error Responses
When a send is rejected, the API returns a 400 with a descriptive error message. Daily limit violations return 429:
Multi-Recipient Block
{
"error": "Content policy violation: Agents can only send to one recipient at a time"
}Spam Detected
{
"error": "Content policy violation: Content matches spam pattern: \"buy now\""
}Sales Outreach Detected
{
"error": "Content looks like sales outreach: \"schedule a demo\". Agents cannot send unsolicited sales or marketing emails."
}Daily Limit Exceeded (429)
{
"error": "Daily send limit reached (25/day). Resets at midnight UTC."
}