Skip to main content

A safety net for your AI agent

Your AI builds.
Playrunner protects what works.

Give your agent the confidence to keep building.

Connect your agent to saved tests, ready-to-run environments and evidence it can return to. Keep checking the payments, bookings and customer journeys people depend on as your application grows.

For your AI agent

Start with a prompt.

Paste this into Codex, Claude, Cursor, Gemini CLI, Antigravity or Grok Build. Your agent will guide you through installing the Playrunner CLI, connecting Playrunner and preparing a first test run.

Copy the prompt, then paste it into your agent’s chat.

First time? Your client may ask you to add the MCP server and authorise access in your browser.

Connection guide
See how it works

Repeatable tests. Detailed reports. A lasting test history.

Works with Codex Claude Cursor Gemini CLI Antigravity Grok+ command line
CodexPlayrunner
Connected
The taskAdd billing. Update the pricing page.
  1. Codex

Illustrative conversation · Product direction.
Built around your existing stackPlaywright Streamline Icon: https://streamlinehq.com Playwright GitHub SlackJira Streamline Icon: https://streamlinehq.com JiraExplore integrations

The vision · Protection that grows

A new feature today.
A safety net for tomorrow.​

You’ve built something people depend on. Now you want to improve it without losing customer updates, stopping payments or breaking the booking flow.

We’re building towards a growing set of agreed checks your agent carries into every change. Protection that lasts beyond one task, one conversation or one developer.

For your AI agent

Start with a prompt.

Paste this into Codex, Claude, Cursor, Gemini CLI, Antigravity or Grok Build. Your agent will guide you through installing the Playrunner CLI, connecting Playrunner and preparing a first test run.

Copy the prompt, then paste it into your agent’s chat.

First time? Your client may ask you to add the MCP server and authorise access in your browser.

Connection guide
Run your first test

Remember what must keep working

Turn important journeys into saved tests. Reuse the same steps and pass/fail criteria as the application changes.

Give your agent evidence to act on

Run the checks through Playrunner. Return the failing step, expected result and captured evidence so your agent can investigate and verify its fix.

Carry the history forward

Keep earlier failures and subsequent runs available. Your next conversation—and the next person on your team—can pick up with the evidence intact.

Built on Playwright. Ready for your agent.

Your agent’s testing setup.
Ready for the next conversation.​

Playwright executes the tests. Playrunner brings the saved workflows, environments, authentication and run history together so your agent can operate them from one place.

Your AI agent

Build. Request checks. Investigate. Fix.

Playrunner

Saved workflowsEnvironmentsAuthenticationRun history
Playwright executes your tests

Results, logs and reports come back to your agent.

From your first customer to enterprise

Protect the journeys
your business runs on.​

A booking flow. A customer payment. Two systems keeping thousands of records in sync. Give your agent repeatable checks for the behaviour your business relies on.

For teams, the evidence is the handover. Developers can investigate failures, QA can review coverage, and release owners can see what passed, what failed and what wasn’t tested.

Explore reports and history
Example · Two-way customer sync
01Your CRM
02Policy platform

Both systems. Same customer.

  • A customer update reaches the other system.
  • Conflicting edits preserve the right information.
  • A retried sync creates no duplicate records.

Agreed checks. Inspectable results. A record of every run.

Make your agent’s token budget go further

Less repeated setup.
More room for the fix.​

Every instruction and tool response your agent processes adds to its context. Playrunner can reduce repeated work by keeping test setup, execution history and reports ready to reuse.

Saved in Playrunner

The testing context carries over.

WorkflowEnvironmentAuthentication ProfileReports & traces

Set it up once. Reuse it in the next run or conversation.

Your agent’s next runIllustrative handoff
  1. 01 Start the saved workflow.
  2. 02 Read new run events.
  3. 03 Open the evidence to investigate.
01 / Reuse the setup

Replay established checks.

Start a saved workflow by its ID. Reuse the setup and run established tests without asking the model to reason through every browser step again.

02 / Read new information

Avoid reading the same log twice.

Cloud run updates let your agent pick up where its last check ended, without fetching earlier events again. Choose when to check back; every poll still has a cost.

03 / Keep evidence available

Open reports when they help.

Reports and captured traces stay attached to the run. Your agent can follow those links when investigating, rather than copying whole report files into every conversation.

Actual token savings depend on your agent and workflow. Run updates can include logs. Test execution and any AI analysis steps have their own costs.

How to measure the saving

Compare the same task, tests and failure using the same model: your current testing approach versus Playrunner. Record actual input, output and cached-token usage, including setup, polling and diagnosis. Compare first-run and repeat-run costs separately, and include Playrunner compute and any AI-step costs in the total.

For you. For your agent. For your terminal.

A few commands.
A lot more confidence.​

See how the Playrunner CLI connects browser sessions, test workflows, and your own compute.

A real sign-in. A reusable session.

Pair your computer from the terminal, create a profile with your agent or in Cloud, then capture your browser session for authenticated tests.

Animated walkthrough · Sample data
Terminalplayrunner CLI
Commands you can use. Sample output.
Browser · Playrunner Cloud
STEP 1 / 4

Approve your device

Check the matching pairing code in the browser and approve this computer.

Device paired
0s / 28s

Start with the CLI installed, Chrome available, and a Playrunner Cloud account.

Read the full guide
Read the walkthrough transcript
  1. Pair your computer
    playrunner login

    Check the matching pairing code in the browser and approve this computer.

    Pairing code: DEMO-1234
Paired demo-laptop with Playrunner Cloud. Run "playrunner auth connect" to accept authentication requests.
  2. Create the profile
    playrunner auth connect

    Set the login URL and success condition. Select your online device, then request authentication. Keep this terminal running.

    Playrunner authentication companion is connected.
  3. Sign in with Chrome
    playrunner auth connect

    Chrome opens on your paired computer. Complete sign-in and MFA in the browser, then press Enter in the companion terminal.

    Playrunner authentication companion is connected.
  4. Reuse the captured session
    playrunner auth connect

    Attach the authenticated profile to a Playwright step. Run the workflow to verify the session against your application.

    Playrunner authentication companion is connected.
Authentication Profile capture completed.
npm install --global playrunner@latestExplore the CLI

One MCP connection. Your choice of agent.

Your favourite agent.
A testing workflow it can reuse.​

Bring testing to Codex, Claude, Cursor, Gemini CLI, Antigravity, Grok, or another compatible MCP client. The same saved workflows. The same evidence.

MCP + skill

Codex

Go from a coding task to a real test run. Create workflows, run your suite, and inspect the outcome from the same conversation.

Connect Codex
MCP connection

Claude

Bring test execution into your Claude workflow. Connect Playrunner’s remote MCP server to run tests and review the evidence.

Connect Claude
Remote MCP

Cursor

Bring testing into your editor. Add Playrunner’s remote MCP server to Cursor and authorise access to your saved workflows and test reports.

Connect Cursor
Remote MCP

Gemini CLI

Bring Playrunner’s testing tools to Google’s terminal agent. Configure the remote server and browser sign-in to get started.

Set up Gemini CLI
Remote MCP

Antigravity

Add Playrunner as a custom MCP server in Google Antigravity. Give your agent access to saved workflows and test evidence.

Set up Antigravity
Custom MCP

Grok

Add Playrunner as a custom connector in Grok, or a remote MCP server in Grok Build. Use the connection to request test runs and inspect their results.

Connect Grok
An open protocol

More MCP clients

Use a client with remote Streamable HTTP and OAuth support. Connect once and give it the same Playrunner tools.

See connection requirements
Terminal & automation

Playrunner CLI

Give shell-capable agents a command-line route to Playrunner. Start workflows, check execution status, and retrieve reports.

Explore the CLI

Serious testing. Less setup.

The execution layer
your agent was missing.

Scale the run

Automatic sharding distributes your suite across runners and brings the results back into one report.

Explore sharding

Run where it matters

Use managed Cloud compute or connect your own machine. Keep execution history and reports together.

Choose your compute

Test beyond the login

Reuse Authentication Profiles to test signed-in journeys, with browser sign-in handled on your device.

Explore authentication

Need more control? The visual workflow editor is there when you want to connect steps, inspect a run, or refine the flow.

Explore the platform

Cross-project discovery · Product direction

One shared component.
Every project that uses it.​

Working across several projects? We’re building towards Playrunner discovering where a shared component is used and selecting its relevant tests. Change a country selector, discover five projects that use it, and check that control in each one.

01

Discover the affected projects

Your agent would use Playrunner to find which connected projects use the component it changed.

02

Test that component across them

Run the component’s relevant checks across those projects while your agent continues another task. Keep the run focused on the change.

03

Return the evidence to your agent

Identify the affected project and failing step, with reports and traces to help your agent investigate, fix and check again.

Your tools. One connected workflow.

Great tests don’t stop at a result.
Connect what happens next.​

Bring your tests, AI and team tools together. Let your agent build the flow, then inspect every connection in Playrunner.

Explore all integrations
Workflow previewConnected with Playrunner
Run checkout testsPlaywrightSummarise failuresOpenAI · structured outputNotify the teamSlack · #qa-alertsCreate a bugJira · engineering

When a test fails, summarise the evidence with OpenAI. Send the summary to Slack and create a Jira bug so the team can investigate.

Build this workflow

Keep building on what works

Give your AI agent
a safety net.

Start with one journey your customers depend on. Keep the checks and evidence ready for every change that follows.

For your AI agent

Start with a prompt.

Paste this into Codex, Claude, Cursor, Gemini CLI, Antigravity or Grok Build. Your agent will guide you through installing the Playrunner CLI, connecting Playrunner and preparing a first test run.

Copy the prompt, then paste it into your agent’s chat.

First time? Your client may ask you to add the MCP server and authorise access in your browser.

Connection guide
Explore Playrunner Cloud
Prefer to self-host? Start locally →