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

# Engines and Commerce MCP Servers

> Three interchangeable computer-use engines — Claude, OpenAI, and NSR — plus the Shopify, Mirakl, and Dsco MCP stack and OSWorld benchmarking.

# Engines and Commerce MCP Servers

The StateSet Computer Use Agent now runs on **three interchangeable engines** and can drive
commerce APIs and the GUI in a single loop.

## Engines

Select the engine per job — from the dashboard, or with `--engine {claude,openai,nsr}` on the
CLI.

| Engine     | What it is                                   | Notes                                           |
| ---------- | -------------------------------------------- | ----------------------------------------------- |
| **Claude** | Anthropic computer use                       | `--model` selects the Claude model              |
| **OpenAI** | `computer-use-preview` via the Responses API | Driven through the same platform `ComputerTool` |
| **NSR**    | The zero-API **local** engine                | No API calls, no X display required             |

All three share the cost and summary path, so switching engines does not change how a job is
accounted for or reported.

### NSR web capability

NSR drives a **headless Playwright browser** with no X display and no API calls. A bridge
implements the two methods NSR calls — `execute` and `take_screenshot` — feeding the
`a11y_first` perception real DOM element boxes rather than pixels.

On the realistic vertical benchmark, action-task accuracy went from **0% to 80%** (12/15). The
remaining misses are the three hardest multi-step purchase and transfer flows.

NSR also gained:

* **Real `<select>` dropdown support** (`select_option`).
* **Perceive-and-scroll** to reach below-the-fold elements.
* **Expanded money-action steps** so multi-step finance flows complete.

## Commerce MCP stack

One `MCPManager` aggregates **25 commerce tools** across three servers, so the agent uses APIs
where they exist and falls back to the GUI where they don't — API-first, GUI-fallback, in one
loop.

| Server            | Tools | Covers                                                                                                                                                |
| ----------------- | ----- | ----------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Shopify**       | 6     | Store and order operations                                                                                                                            |
| **Mirakl**        | 11    | MMP (orders, offers, messaging, including accept-order writes), Catalog Manager (products, hierarchies, attributes), and MPS (service orders)         |
| **Dsco (Rithum)** | 8     | Distributed inventory and drop-ship: ping, generic request escape hatch, inventory single and batch writes, list and create orders, shipment, invoice |

Each is a FastMCP stdio server with an agent preset. The Dsco server implements a faithful
OAuth2 client-credentials lifecycle per the v3 Standard API guide — mint, cache to `expires_in`,
bearer, and refresh on 401.

<Note>
  Multi-server aggregation required an `MCPManager` handshake fix; the multi-server plus
  computer-use loop is verified live.
</Note>

## OSWorld benchmarking

A real OSWorld score is runnable locally for all three engines, with a Docker/KVM run path.

```bash theme={null}
# all three engines are live
--engine nsr | openai | claude
```

The runner ships a `docker.yaml` config and an `osworld_os_smoke.jsonl` of 8 real os-domain
tasks drawn from OSWorld's evaluation examples — terminal and file operations, settings, and
one infeasible task.

<Warning>
  The OSWorld Docker provider runs with `NET_ADMIN` but no tun device, which costs QEMU its
  port forwarding and leaves the VM unreachable. The runner injects `/dev/net/tun` and
  `privileged` for the OSWorld image — **this requires `/dev/net/tun` on the host**. With it the
  VM boots in about 24 seconds; without it, readiness times out.
</Warning>

NSR has been validated end to end on the real Dockerized OSWorld VM, scored by OSWorld's own
evaluator.

There is also a cross-domain benchmark grader for `computer-use-preview`.

## Related

* [Computer Use Agent](/computer-use-agent) — architecture and agent catalog
* [MCP Integration Guide](/guides/mcp-integration-guide)