Let your coding agent test what it just built
Connect Claude Code, Cursor, Codex or VS Code to Operio's hosted MCP server, or use the CLI. Your agent can write tests, start runs and read the results without leaving the editor.
Invite a teammate
Running- 1Sign in as @admin
- 2Open Settings, then Team
- 3Invite new.hire@acme.dev as an Editor
- 4The invite shows up as Pending
- Hosted MCP, no server to run
- JSON on stdout for every command
- A pass or fail your CI understands
Capabilities
End-to-end testing your agents can drive
The same tests, runs and results as the dashboard, exposed in the formats agents and pipelines work with.
Hosted MCP server
Add api.operio.run/mcp to your client and sign in through the browser. There's no API key to copy and no server to host.
Scoped access
Approve read access to inspect results, or read and write access to let the agent create tests and start runs.
A CLI made for agents
Install with npm install -g @operio/cli. Every command prints JSON, so agents and scripts can parse what happened.
A gate for CI
Run a test or a suite from CI. The command waits until the run finishes, then passes or fails the job, so your pipeline knows whether to ship.
An agent skill
operio skill install adds an Operio skill to your repo, so Claude Code knows how to write and run tests on its own.
REST API
Projects, test cases, suites, runs, environments and bugs are all available at api.operio.run/v1 with a personal API key.
How it works
From prompt to passing test
- 01
Connect your client
One command in Claude Code, or one entry in your Cursor, VS Code or Codex config. Approve access in the browser.
- 02
Ask for a test
“Write a test for the new invite flow and run it on staging.” The agent creates the test case and starts the run.
- 03
Close the loop
The agent reads the result and the failure reason, fixes the code, and runs the test again until it passes.
The Operio platform
One agent, every way your team ships
- Web TestingEnd-to-end tests for any web app, in plain EnglishExplore Web Testing
- iOS TestingNative iOS apps on cloud simulatorsExplore iOS Testing
- Android TestingNative Android apps on cloud emulatorsExplore Android Testing
- Exploratory TestingGive the agent a goal and get bugs backExplore Exploratory Testing
FAQ
MCP and CLI questions
Any client that supports MCP over Streamable HTTP with OAuth. The docs walk through Cursor, Claude Code, Claude, VS Code and Codex.
Not for MCP: you sign in through the browser and approve read or read-write access. The CLI uses a browser login too, and the REST API uses a personal API key.
No. Upload builds in the Operio web app or with the CLI, then reference the build from MCP to run tests against it.
Run your tests or a suite with the Operio CLI in a pipeline step. The step fails when the run fails, and the full result stays available for later steps. The docs have the details.
Give your coding agent a QA engineer
We'll connect Operio to your editor and run a test on your app in the demo.