Using Loadrigo

Loadrigo can call everything on this page. It uses the same MCP tool catalog as any connected agent, limited to an allowlist of actions that fit a chat turn. If an action is not listed here, ask Loadrigo and it tells you it can’t do it. It does not make up a tool that doesn’t exist.

Loadrigo’s own controls are on one console page: Settings → Loadrigo (/settings/loadrigo). There you set the custom instructions it reads on every turn, and you see how many messages your account has left today.

Settings → Loadrigo. Loadrigo reads these custom instructions on every turn. The page also shows your account's daily message allowance.

Account and workspace

• Tell you who you’re signed in as and which workspace is active (“what account am I in?”).
• List the workspaces your account can see, and switch the active one for the rest of the conversation.

Projects and tests

• List and create projects.
• List tests, open one test’s detail (including its last validation status), and show a test’s run-history overview.
• Create a test and upload its entrypoint script or supporting files. Paste code inline, or point Loadrigo at a file in your repo if it runs with local file access.
• List and download a test’s uploaded files.

Runs

• Start a run, poll its status, list recent runs, and re-run a previous run. A re-run either repeats the run exactly or rebuilds it from the test as it is today.
• Cancel a run. Cancelling is destructive, so Loadrigo always asks you to confirm first.

Results and diagnostics

• Pull a run’s KPI summary (throughput, latency, error rate, duration) and query specific metric time series.
• List grouped errors. For one error, show its sampled request/response bodies and status codes.
• Pull a run’s error-level engine and system logs.
• Check runner health: heartbeat, CPU/memory trend, and whether the run reached its target virtual-user count (VU shortfall).
• Scan a finished run for anomalies and explain the outliers in plain language.

Dashboards

• List and open dashboards, browse the metrics catalog, and create or update a dashboard.

Browser fleets

• Create a managed browser fleet (chrome/firefox/edge, by count and location), check on it, list your fleets, scale one up, and delete it. Browser fleets explains what a fleet is and when to use one instead of a scripted browser test.

Reverse tunnels

• Create a reverse tunnel to expose something that runs locally or isn’t deployed yet. List and inspect your tunnels, stop one, and pull its traffic stats. See Tunnels.

Virtual services

• List, inspect, create, deploy, stop, and delete a virtual service. A virtual service is a mock backend that stands in for a dependency you don’t want to call for real. See Virtual services.

Execution locations and on-demand browser inspection

• List where a fleet, virtual service, or test run can physically run: MaxoPerf’s Shared Sandbox outpost, your own connected outposts, and managed cloud regions.
• Start a real browser that renders JavaScript and look at a URL. Loadrigo uses this when it needs a single-page app’s rendered content, not only the static HTML, to plan a browser test’s selectors.

The escape hatch

• For anything no named tool covers (billing, secrets, notifications, schedules, and other less common Platform API endpoints), Loadrigo can search the public API reference and call the endpoint directly. It never shows you a secret’s value. It only confirms that it created the binding.

Docs and the skill

• Search MaxoPerf’s documentation to answer “how do I…” questions based on what has shipped.
• Read the MaxoPerf agent skill bundle and its sections. External coding agents get the same instructions, so Loadrigo builds tests the same way they do.

Driving the console for you

Loadrigo also has a few local actions that work only in the console. It can navigate to a page, scroll to or highlight an element, read what’s on screen, click, fill a field, select an option, and refresh the data on the page you’re viewing. These are not MCP tools. Loadrigo uses them when a request needs a visible action in the console, and you always see each one happen live in your own tab.

When you ask Loadrigo to remember a preference, it can save it to your custom instructions during the conversation. This is the same field as on Settings → Loadrigo.

The assistant on the website

This page describes Loadrigo in the console, where it acts with your own account’s authorization. The assistant on the public MaxoPerf website is a separate, deliberately small surface. It answers questions about the product and does exactly one thing: it sets up a free MaxoPerf account for the email address you give it, with a demo load test against a URL you name. It can’t reach any account data and creates nothing else.

After it sends you into the console, you talk to the full Loadrigo described on this page. There is no reduced tier and nothing to upgrade. The new account is an ordinary free account. The only step left is to set your password from the link MaxoPerf emails you.

What Loadrigo doesn’t do

• It won’t scale a fleet down. It only scales up. To shrink a fleet, delete it and create a smaller one. The tool itself works the same way.
• It won’t take a destructive action without asking. Cancelling a run, deleting a fleet, stopping a tunnel, and deleting a virtual service each need your confirmation first.
• It follows read-only sessions like any MCP client. In read-only mode, write actions return an error instead of silently doing nothing. MCP overview explains read-only mode.
• It doesn’t make up capabilities. If you ask for something outside this list, it says so instead of guessing at an action that sounds right.