Acceptable Use Policy

This Policy sets out what you can and cannot do on Cool — including when an AI acts on your behalf (Sophia, Claude over MCP, or the command line). The rules exist for a practical reason: what one account does with bulk sending affects email deliverability for every other account.

AI agents follow the same rules you do

Cool lets you connect AI assistants to your community (Sophia, Claude via MCP, Codex via the command line). They can read, create, edit and publish on your behalf.

Every agent action is treated as yours: you are responsible for it, and it goes through the same limits. There is no "via AI" path that bypasses a rule in this Policy. If something cannot be done by hand, it cannot be done by command either.

Rate and volume limits

Every account has hourly and daily limits for automated actions: how many posts per hour, how many sends per day, and to how many people at a time.

These limits start lower on new accounts and rise as the account builds history — the same principle email providers use: sending reputation is built, not declared. An account created today does not blast an entire list today.

Drafts do not count. You may prepare as many articles, posts and campaigns as you like; what is limited is publishing and sending.

Contact lists and consent

You may only send marketing communication to people who consented to hear from YOU. Purchased, rented, traded or third-party scraped lists are prohibited on Cool.

Importing a contact is not the same as being allowed to market to them: the import records the contact; marketing depends on source and consent. Large or unusual imports may go through verification before sending is enabled.

Why: addresses without consent generate spam complaints, and spam complaints damage deliverability for every Cool customer — including those who did nothing wrong.

What is not allowed

You may not: (a) send bulk messages without consent; (b) publish repetitive, automated or low-value content at volume; (c) create multiple accounts or API keys to work around limits; (d) use the platform to deceive, defraud or impersonate a person or organisation; (e) automate social interaction beyond what the network's own policy allows; (f) attempt to access another community's data; (g) resell API or agent access without authorisation.

Item (c) is the most common and the most direct: limits are per community precisely so that creating ten keys does not multiply the ceiling by ten.

What we do when something looks off

When usage looks off, we act with the lightest measure that solves it, in this order: slow down sending, ask for confirmation before executing, pause the automation involved, temporarily suspend bulk sending and, in severe or repeated cases, suspend the account.

Whenever possible we warn first and explain why. Automated actions are logged and you can appeal through support.

Why the limits exist

This is not bureaucracy, and it is not distrust of you.

Cool's email delivery is shared: providers (Gmail, Outlook and others) judge the reputation of the domain and IP the message comes from. A single account blasting a non-consented list damages delivery for everyone sending through the same path.

Inside the community the logic is the same: five hundred posts published at once harm no provider, but they make a community look abandoned to a bot — and it is the owner who pays for that.