remote control
Octomind from your IDE

Stay in your workflow. Connect your AI agent (Cursor, Claude, etc.) to Octomind via MCP to generate, run, and debug tests without ever leaving your editor.

{ }~/.claude/config.json
1{
2  "mcpServers": {
3    "octomind": {
4      "command": "npx",
5      "args": ["@octomind/octomind-mcp"],
6      "env": {
7        "APIKEY": "<YOUR API KEY>"← Add your key
8      }
9    }
10  }
11}

Trusted by hundreds of teams around the world

BRM
Deriv
Fixify
Wingrep
Hadrian
Lennar

watch the agent at work

From Jira ticket to passing test in one prompt

Daniel Roedler demonstrates how to chain Jira, and GitHub and Octomind into a single prompt using Cursor.

chain Octomind with Jira & GitHub

Don't just chat. Execute. MCP allows agents to pipe data between tools

INPUT

Jira MCP

"Fetch Ticket-102"
Octomind
PROCESS

Octomind MCP

"Generate Test Case"
OUTPUT

Local Filesystem

"Login flow test generated"

deploy a QA team
inside your IDE

Octomind's MCP server lets you run a specialized 'Tester Agent' alongside your 'Coder Agent', keeping your focus on the feature while we break it.

Never Leave VS Code

Your editor is your command center. Trigger test runs, fetch traces, and generate new specs directly from your chat window. No more context switching to CI dashboards.

The Adversarial Agent

Don't let the agent that wrote the code write the test. Use Octomind to spin up a separate 'QA Agent' that impartially verifies your work, ensuring unbiased coverage.

Connect the Dots

Chain tools together. Ask your agent to 'Read Jira Ticket #123, generate an Octomind test for it, and open a PR in GitHub.' One prompt, three systems connected.

Real Users, Real Browsers

We don't mock the DOM. Your agent triggers real browsers in our scalable cloud grid that click, type, and interact like actual humans—catching bugs that API tests miss.

frequently asked questions

Everything you need to know about the MCP server

Which AI editors are supported?
The Octomind MCP server follows the open Model Context Protocol standard. It works natively with Cursor, Windsurf, and Claude Desktop. Any future client that implements MCP will also work out of the box.
Can I test localhost apps from my IDE?
Yes. Since the MCP server runs locally on your machine, it can access your localhost server directly. You can trigger runs against localhost:3000 to verify changes before you even push to staging.
Why shouldn't I just let the agent "drive" the browser?
Agents acting as "drivers" (clicking coordinates) are flaky and slow. Octomind treats the agent as an Engineer, not a user. Your agent generates the code, and our deterministic engine handles the execution, ensuring your tests are stable and repeatable.
Can I use this to verify code my agent just wrote?
Absolutely. We recommend the "Adversarial Agent" pattern: use one agent instance to write the feature code, and use the Octomind MCP tool to spin up a separate "QA Agent" to independently verify that code with a fresh test case.
How do I debug a failed test?
You don't need to open a dashboard. Just ask the agent to "Fetch the trace for the last failure." It will pull the screenshots, error logs, and DOM snapshots directly into your chat window so you can fix the issue without ever leaving your editor.

release confidence starts here

Try Octomind and instantly get stable runs, visual debugging, and self-healing tests.