Bring your own AI agent

You may already work inside an AI assistant you like, in your editor, on your desktop, wired into your own scripts. You should not have to leave it to drive your tests. AxonQA runs a server that speaks the Model Context Protocol, the open standard that lets an assistant discover and call external tools, so an MCP-compatible client can operate AxonQA directly, end to end, from wherever you already are.

One protocol, many assistants

MCP is a shared language between an AI client and a tool. Because AxonQA exposes its capabilities over that protocol, any assistant that speaks it can connect: desktop assistants like Claude Desktop, coding assistants inside editors like VS Code, and other MCP-compatible clients, including ones your own team builds. You are not tied to a single vendor or a single window. The assistant you already trust becomes a front end for your test platform.

What your assistant can actually do

Connected over MCP, your assistant is not just chatting about testing, it is doing it. The server exposes real AxonQA actions as tools the assistant can call, and it hands back results the assistant can reason about.

• List your projects, suites, and tests, and read what each one covers.
• Generate test cases from a story, grounded in your project context.
• Start a run on the browsers you name, and follow it to completion.
• Read the verdicts, including any healed steps, and fetch the run report.

It fits the way you already work

The value of driving AxonQA from your own assistant is that testing stops being a separate destination. Ask the assistant in your editor to generate cases for the story you are implementing, run the smoke suite before you open a pull request, and tell you whether anything broke, all without switching tabs. The same actions Axon, your AI assistant, performs inside AxonQA are available to the assistant on your side, so the workflow meets you where you are instead of pulling you out of it.

The same rules apply, whoever is driving

Bringing your own agent does not bypass anything. A client connects with your credentials and stays inside your organization's permissions, so it can only see and do what you can. Runs still execute on the real platform, results are still recorded, and every action is still auditable. The MCP server is a new front door onto AxonQA, not a side door around its guarantees: whoever is driving, the tests, the data, and the boundaries are exactly the same.

An AI agent is most useful when it can reach the tools you actually use. By speaking MCP, AxonQA turns your test platform into one of those tools, callable from the assistant you already have open. Connect your client once, and generating cases, running suites, and reading results become things you can do in the same conversation where you write the code, with no context switch required.