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

# Plans

> Free, Pro and Scale — what each one allows, what happens when you reach a limit, and what a downgrade does and does not destroy.

`/settings/billing` shows the plan cards and your [usage](/billing/usage) against them.

|                          | Free   | Pro        | Scale      |
| ------------------------ | ------ | ---------- | ---------- |
| Monthly                  | \$0    | \$49       | \$100      |
| Yearly                   | —      | \$490      | \$1,000    |
| History                  | 7 days | Everything | Everything |
| Calls a month            | 10,000 | 1M         | 10M        |
| MCPs                     | 1      | Unlimited  | Unlimited  |
| Team                     | 1      | Unlimited  | Unlimited  |
| Individual calls kept    | 7 days | 90 days    | 365 days   |
| Alerts and weekly digest | —      | ✓          | ✓          |
| All 16 metrics           | ✓      | ✓          | ✓          |
| API and MCP access       | ✓      | ✓          | ✓          |

**Yearly is ten months' money for twelve**, on both paid plans. That is worth more than it looks: our merchant of record charges a percentage **plus 50¢**, so the flat fee is what punishes a cheap plan, and a year is one 50¢ rather than twelve.

## MCP access is on every plan

Including Free. The [MCP connector](/mcp/overview) reads that account's own data through that account's own window, so a free account gets seven days of it inside Claude — which is the best demonstration this product has and costs nothing to give away.

## History and retention are the same lever

If an account keeps seven days, "last 30 days" is not a feature being withheld — it is a window with nothing in it. So the plan says how far back you can see, and the [date picker](/api/concepts/date-ranges) offers what is inside that.

But the **counters are never deleted**, only hidden. `tool_hours` is one row per hour per tool per client and costs almost nothing to keep, so:

* **Upgrading reveals history that was already there.** A better moment than starting a new clock.
* **Downgrading loses nothing.** The window narrows; the rows stay.

The one table genuinely worth deleting is `calls`, at one row per call, and that is what "individual calls kept" governs. Only the [live call feed](/metrics/live-calls) reads it.

## Where each limit is actually enforced

Never in the interface alone. The dashboard hides what your role and plan cannot reach, but that is a courtesy — the API is where it is decided.

| Limit              | What happens                                                                   |
| ------------------ | ------------------------------------------------------------------------------ |
| **Calls a month**  | Ingest answers `402` and stops recording                                       |
| **History**        | Every metric route **clamps** the range rather than refusing it                |
| **MCPs, seats**    | Refused at creation, with the plan named in the message                        |
| **Alerts**         | Creating a rule, re-enabling one, and subscribing to the digest are all closed |
| **Call retention** | The cleanup job deletes by your plan's window                                  |

Two of those are worth expanding.

**Recording stops rather than overflowing.** A banner appears above the overview, the meter goes red on the billing page, and the chart is flat from the moment it stopped. The alternative — recording past the cap and asking nicely — lost, because a cap that is not a cap is a line nobody can plan around. Your server is unaffected: the SDK treats any non-2xx as a failed batch and drops it, so this costs you data and never a tool call.

**Alerts have three doors and all three are shut.** Creating a rule, switching an existing one back on, and subscribing to the weekly digest. Only creating was gated at first, which is how a lapsed account could quietly keep the feature. *Disabling* and *unsubscribing* stay open, because somebody whose plan ended must be able to tidy up what they can no longer run — and the nightly evaluation skips free accounts, so a rule that outlives its subscription does not keep filling an inbox.

## Payment

**Polar**, as merchant of record. They are the seller, so VAT, GST and sales tax in every jurisdiction are handled by somebody who does that for a living — which for a product sold from one country to developers in forty is the difference between shipping and registering for tax in forty places.

Checkout is a redirect. Invoices, cards and cancellation are a signed link into Polar's customer portal. MCPulse never holds a card number, and no screen in this product is built to.

<Note>
  **`canceled` is not `revoked`.** Cancelling means "do not renew" — you keep the plan until the period you have paid for ends, and the billing page says which date that is. Access is removed only when the subscription is actually revoked.
</Note>

## What a free account can do

All sixteen metrics, the insight rules, the live call feed, the whole REST API and the MCP connector. Not a demo with the interesting half removed — a week of the product, on one server, for one person.

The two things it does not have are the two that only matter once a server is real: a date range longer than a week, and [alerts](/alerts/rules) that tell you when a tool's first-call rate falls without you going to look.

## Related

* [Usage](/billing/usage)
* [Date ranges](/api/concepts/date-ranges) — how the history window is applied
* [Alerts](/alerts/rules)