Acceptable use policy

A MacRun unit is for developing, building, testing, and operating your own software. On the agent Mac plan, it may also run an AI assistant that works for you on your own accounts, such as your email, calendar, and chats.

Claude Code, Codex, OpenClaw, and any other agent you install must follow this policy.

• ✕Sending spam or bulk unsolicited email. Running mail relays.
• ✕Bulk or unsolicited messages on any chat app, such as iMessage, SMS, WhatsApp, or Telegram. An assistant may send your own messages to people you know, the way you would.
• ✕Cryptocurrency mining or other proof-of-work workloads.
• ✕Hosting or distributing illegal content, malware, or stolen data.
• ✕Anonymizing or proxy traffic for others, such as Tor relays or open proxies.
• ✕Scanning, attacking, or trying to access systems you do not own.
• ✕Running a public service or reselling access to the unit to anyone else.
• ✕Any use that breaks Apple's macOS license or the law.

An agent on your unit can browse, log in, send, and buy without a person watching. These rules exist because of that.

• +Your agent acts as you. Whatever it does, you did.
• +Only use accounts you own or are authorized to use. Do not let an agent operate someone else's account.
• +Do not use an agent to get around a website's terms, rate limits, CAPTCHAs, or bot detection.
• +No scraping at scale, no credential stuffing, no automated account creation, no mass posting or mass messaging.
• +Do not share your Claude or ChatGPT account across people, and do not resell model access through the unit. Follow the provider's terms.
• +Do not point an agent at systems, networks, or people it has no permission to touch.
• +Personal assistant agents such as OpenClaw are allowed on the agent Mac plan, for your own accounts. Every message they send counts as a message you sent.

If a third party complains about traffic or actions from your unit, we will pass the complaint to you and may suspend the unit while you fix it.

You may install any software the license allows, including other agents. It must follow this policy. We support Claude Code and Codex. Anything else is yours to run and yours to keep within these rules.

We monitor aggregate network egress at the hub for abuse signatures such as bulk SMTP, mass-messaging patterns, anonymizing traffic, and scanning. Every unit reaches the internet through the same address. One unit that sends spam can get that address blocked for every customer. So we watch how much mail and how many messages a unit sends, and we suspend a unit that sends in bulk. We count them. We never read them. This protects the fleet and other customers. We do not inspect the contents of your builds, your source code, or your agent conversations. See the customer portal's data and privacy page for exactly what the operator can and cannot access.

Violations result in immediate suspension of the unit. We suspend first and ask questions after. Repeated or severe violations lead to termination and a data wipe under the terms of service. Report abuse to support@macrun.dev.