> ## Documentation Index
> Fetch the complete documentation index at: https://docs.agent-loadout.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Usage Limits, Resource Caps, and Overages – Agent Loadout

> How Agent Loadout pools limits across your organization, what happens when you reach a cap, and how to add capacity or monitor consumption.

All limits in Agent Loadout are pooled across your entire organization — not per agent. Your five agents share one bucket of inbound messages, one bucket of machine time, and so on. This makes it easy to concentrate resources on the agents that need them most without redistributing allowances manually.

## Pooled Resources

The following resources are tracked at the org level and shared across all agents:

| Resource                    | Free Sandbox       | Pro (5 agents) |
| --------------------------- | ------------------ | -------------- |
| Outbound recipients / month | 25 (verified only) | 1,000          |
| Inbound messages / month    | 100                | 5,000          |
| Mailbox storage             | 50 MB              | 5 GB           |
| Machine time / month        | 0.5 h              | 50 h           |
| Processed mail data / month | 50 MB              | 2 GB           |
| Machines at once            | 1                  | 3              |

<Note>
  Each additional Pro agent seat adds 200 outbound recipients, 1,000 inbound messages, 1 GB mailbox storage, and 10 h of machine time to the shared pool.
</Note>

## What Happens at a Limit

When your organization reaches a resource cap:

* **Outbound sending** pauses. Queued messages are not delivered until capacity is restored.
* **Mailbox imports** pause. New inbound messages cannot be written to mailboxes until the inbound limit resets or additional capacity is added.
* **Incoming overflow** has a limited recovery window. Agent Loadout does not drop incoming mail immediately, but you should not rely on overflow as a long-term storage buffer. Monitor your inbound usage and add capacity before you approach the limit.

<Warning>
  Do not rely on the incoming mail overflow window as a substitute for adequate inbound message capacity. Messages held in overflow are not guaranteed to be retained indefinitely.
</Warning>

The REST API returns HTTP `402` when access is paused or a quota has been reached:

```json title="Quota error response" theme={null}
{
  "error": {
    "code": "quota_exceeded",
    "message": "Monthly inbound message limit reached.",
    "hint": "Add agent seats or enable pay-as-you-go in Billing & usage."
  }
}
```

## Adding Capacity

You have two ways to add capacity without upgrading your entire plan:

<CardGroup cols={2}>
  <Card title="Add Agent Seats" icon="user-plus">
    Each new Pro seat expands the shared pool: +200 outbound, +1,000 inbound, +1 GB storage, +10 h machine time.

    Seats are added in **Settings → Agents**.
  </Card>

  <Card title="Pay-as-You-Go Overages" icon="credit-card">
    Enable overages by setting a spend cap in **Billing & usage**. Overages are billed at the end of the billing period.

    | Resource                  | Overage rate         |
    | ------------------------- | -------------------- |
    | 1,000 outbound recipients | €2                   |
    | 1,000 inbound messages    | €1                   |
    | Storage                   | €1 / GB-month        |
    | Machine time              | €0.06 / machine-hour |
  </Card>
</CardGroup>

<Tip>
  Pay-as-you-go is **disabled by default**. An owner must set a spend cap before any overage charges can apply. Until then, sending and imports pause at the hard limit rather than accumulating charges.
</Tip>

## Machine Time Billing

Machine time is billed per second and weighted by machine size — a larger machine consumes the allowance faster than a standard one.

Key details to keep in mind:

* **The full TTL window is reserved** from your allowance when a machine starts, even if you stop it early.
* **Unused time is returned** to the pool when you stop a machine before its window expires.
* Stopping a machine snapshots the disk; `resume_machine` restarts it from that snapshot.
* The Compute Scale add-on (€15 / month) unlocks large machines, 24-hour run windows, and 10 additional concurrent machines.

## Sender Trust Limits

Sender trust limits are separate from your plan allowances and budget. They apply independently and govern production sending to unverified recipients. A successful payment method and meeting sender trust requirements are both required before production sending is enabled.

<Note>
  On the Free Sandbox, outbound mail is limited to verified recipients only — at most 25 per month and 5 per day. Sending attachments requires a paid plan.
</Note>

## Monitoring Usage

Open **Billing & usage** in the dashboard to see real-time consumption across your organization:

* Delivery counts and bounce/complaint rates for each sending domain
* Remaining inbound and outbound allowances for the current billing period
* Machine time consumed vs. available
* Vault entry count and paired device count
* A complete list of overage charges accrued (if pay-as-you-go is enabled)

Check this section before starting a high-volume automation to make sure you have sufficient headroom, or set a spend cap and enable pay-as-you-go so sending continues uninterrupted.