Acceptable Use Policy

1. Scope

This policy applies to everyone who sends mail through Agent Herald, and to every message sent — whether a human composed it or a software agent did. It forms part of the Terms of Service. Where this policy is stricter than the Terms, this policy governs.

You are responsible for all mail sent using your API keys and credentials, including mail sent by agents, scripts or third parties acting on your behalf.

2. What the service is for

Agent Herald exists to send transactional and operational email: messages sent to a specific recipient as a direct result of something that recipient or the account holder did. Notifications, confirmations, receipts, alerts, replies, reports, verification codes, and the ordinary back-and-forth of an agent doing a job on someone's behalf.

It is not a bulk email platform, a marketing automation tool, a newsletter service, or an outbound sales tool. We do not offer list management, campaign scheduling, or audience building, and we do not intend to. If your use case is broadcasting to an audience, this is the wrong product and we would rather tell you now.

3. Consent

You may only send to recipients who have a pre-existing relationship with you and a reasonable expectation of hearing from you — because they are your customer, your user, your correspondent, or because they contacted you first.

You must be able to demonstrate, on request, how a given recipient came to be on the receiving end of your mail. "We found the address" is not consent.

4. Prohibited sending practices

You may not:

• Send unsolicited bulk email, of any volume, under any justification
• Send to purchased, rented, scraped, harvested, appended or otherwise acquired address lists
• Send cold outreach, prospecting, or unsolicited sales mail
• Send marketing or promotional campaigns
• Send to addresses collected without the recipient's knowledge
• Continue sending to a recipient who has asked you to stop
• Spread sending across multiple accounts, domains or credentials to evade limits, thresholds or enforcement
• Resell or sublicense sending capacity to third parties without our written agreement

5. Prohibited content

You may not send messages that:

• Are phishing, credential harvesting, or otherwise designed to deceive a recipient into surrendering information or money
• Impersonate any person, company or institution, or misrepresent your affiliation with one
• Distribute malware, ransomware, or links to it
• Constitute harassment, threats, stalking, doxxing or targeted abuse
• Contain child sexual abuse material, or any content sexualising minors
• Promote violence, terrorism, or incite hatred against a protected group
• Advertise illegal goods or services, unlicensed pharmaceuticals, or fraudulent financial schemes
• Infringe intellectual property rights
• Violate the law applicable to you, to us, or to the recipient

6. Sender identity

Every message must be honestly attributable:

• You may only send from a domain you control and have verified with us by publishing the DNS records we generate. Sending from a domain you do not control is prohibited and technically prevented.
• Headers, envelope addresses and display names must not be forged or misleading.
• A recipient must be able to reply and reach a monitored destination.
• Where a message is sent by an automated agent, the recipient must not be deceived into believing that a human wrote it if they ask.

7. Autonomous agents

Most mail on this platform is sent by software acting without a human watching. That does not dilute responsibility — it concentrates it:

• Every account has an accountable human or legal person behind it. Anonymous operation is not available.
• An agent misbehaving is treated exactly as a human misbehaving. "The model decided to" is not a defence, and we will not treat it as one.
• You are expected to bound what your agent can send — recipients, volume, and content — before you give it a key, not after an incident.
• Prompt injection is a foreseeable risk of connecting an agent to email. If untrusted input can cause your agent to send mail, that is your risk to control.

8. Deliverability thresholds

Sending reputation is shared. One account sending badly degrades delivery for everyone else, so we enforce limits rather than merely publishing them.

These are the thresholds our upstream providers apply to us. Enforcing them per account is how we keep one customer's list hygiene from becoming everyone's delivery problem. Your current standing is visible on your dashboard before it becomes an issue.

9. Suppression

When a recipient hard bounces or files a complaint, that address is added to your account's suppression list and further mail to it is blocked. You may not attempt to circumvent this — by re-adding the address, by varying its formatting, by routing around it through another account, or otherwise.

You may add addresses to your own suppression list at any time. You may remove an address you suppressed manually. Addresses suppressed because of a bounce or complaint are permanent.

10. Security and integrity

You may not probe, scan, overload or attempt to gain unauthorised access to the service or to other accounts; attempt to circumvent rate limits, quotas or authentication; or use the service to attack any third-party system.

Keep your API keys secret. Report a suspected compromise to [email protected] immediately.

11. Enforcement

Where we believe this policy has been breached, we may — depending on severity, and without prior notice where the harm is ongoing — rate limit the account, block specific messages or recipients, suspend sending, or terminate the account.

Serious breaches, particularly phishing, malware and content harming children, result in immediate termination and, where the law requires or the harm warrants it, a report to the relevant authorities. Fees are not refunded on termination for breach.

We would rather fix a problem than end an account. If your sending trips a limit because of a genuine mistake, tell us what happened and what you changed.

12. Reporting abuse

If you have received mail sent through Agent Herald that breaches this policy, tell us: