| What it is | Persistent named agents with role context, background work and repeatable routines on a shared cloud computer. | Agent software you operate, with configurable models, execution backends, persistent memory and reusable skills. |
|---|
| Can I use it? · documented | Included on X Subscription Premium, Super and Ultra. Paid legacy Cursor/Teams and qualifying linked SuperGrok/X Premium+ accounts have other routes. Enterprise arrangements differ. | Open-source software with desktop and CLI options. Supported platforms and execution backends vary; this is an operated deployment, not a hosted-plan entitlement. |
|---|
| Where work runs · documented | Persistent cloud computer shared across the account's Bots. Optional Mac/Windows local command execution has its own permission policy; local execution does not imply local model processing. | Local machine, containers, SSH or supported cloud backends. Jobs need the configured backend and gateway running; a sleeping local host is not always-on hosting. |
|---|
| How usage is paid for · documented | Current X Subscription shares a weekly pool with Cursor and Grok, with prepaid top-ups. Legacy plans use Bot grants and optional Cursor-billed on-demand usage; linked grants do not stack. | MIT software is free. Budget for model/API subscriptions, tool services, hosting or hardware/electricity and maintenance. Optional Nous Portal bundles models and tools; some infrastructure costs remain separate. |
|---|
| What keeps running · documented | Background work, scheduled routines, supported event triggers and handoffs between Bots. Verify that a specific connector supports the action your routine needs. | Scheduled jobs, subagents and persistent Bot routines, with the appropriate running backend and gateway configuration. |
|---|
| Browser and computer tools · documented | Browser, files, terminal, connected apps and human takeover. Optional local execution defaults to asking; websites can still block automated access. | Configurable browser automation, terminal/files, plugins and MCP. Backend and model endpoint choices determine where execution and data processing happen. |
|---|
| What it remembers · documented | Per-Bot preferences, roles and conversation summaries. Working files, browser sessions and logins remain shared across the account. | Local memory/profile files, session search and reusable skills; optional external memory providers. Check that requested memory writes occurred. Learning here is persisted context, not model-weight training. |
|---|
| Approvals and data boundaries · documented | Auto Review and action rules; local policy defaults to asking. Requires storage, does not support Legacy Privacy Mode, and uses applicable Cursor training opt-out settings. | Command approvals, allowlists and isolation options depend on operator configuration. Command matching is not an OS sandbox. Cloud model, tool and memory services have separate data policies. |
|---|
| Decision-changing limits · documented | Shared account state is not a per-Bot security boundary. Some unattended approval requests expire after about ten minutes without running; an unanswered request is not completed work. | Deployment, upgrades, secrets and uptime are your responsibility. A small local model is not equivalent to a frontier hosted one; evaluate the exact stack rather than the agent name alone. |
|---|