Playwright MCP lets your coding agent drive a browser in the moment. Wopee.io is a managed AI tester: saved plain-language tests, verdicts, visual baselines, schedules and evidence. Most teams use both: Playwright MCP in the editor, Wopee.io as the regression safety net.
Short answer
Playwright gives you browser automation. Wopee.io gives you an AI tester. Playwright MCP is Microsoft's open-source MCP server that lets your own LLM drive a browser in the moment; it saves no tests, and Playwright's separate Test Agents can plan, generate and heal Playwright tests with your own AI tool. Wopee.io is the managed version: saved plain-language test cases run by its agent on a schedule or on demand, visual baselines, and a pass or fail with evidence for each test. Use Playwright MCP while you build and Wopee.io for regression.
Last checked: 9 November 2026
Browser automation your LLM drives, next to a managed AI tester that is still there after the conversation ends.
What it is
Browser automation: an open-source MCP server that lets your own AI agent drive a browser
An AI tester: a managed platform with agents, saved tests, verdicts, visual baselines and schedules, plus its own MCP server
Tests
Ad hoc: your LLM decides and executes in the moment; browser profiles persist, tests do not
Persistent: plain-language test cases saved, versioned in Git and re-runnable by the agent
Regression
You prompt again and hope the agent repeats itself
Scheduled and on-demand runs of saved test cases
Visual testing
Not built in
Visual baselines with branching, ignore areas, and cross-browser configs
Maintenance
A fresh improvisation each run; Playwright's own Test Agents (planner, generator, healer) can repair Playwright code, driven by your own AI tool
Adaptive by default: the agent re-reads your UI on every run, and test cases stay editable at every pipeline stage
Evidence
A chat transcript, plus opt-in video and trace recording
Step screenshots, run reports, logs, and run history
Verdicts
Whatever the LLM reports in the chat; saving tests requires the separate Test Agents workflow
A pass or fail for each saved test case, with a run report and evidence
Quick one-off task
One config line in your editor, and the agent is browsing
Built for suites, so a one-off click-through is more setup than it is worth
Cost
Open source, plus your own LLM usage and whatever you build around it
Free to start; paid plans from 19 € per user per month, LLM included
Saved test cases. Plain-language tests your team can read, edit and rerun.
A verdict per test. Pass or fail with a run report and evidence attached.
Visual baselines. Regressions caught across releases, not only in the moment.
Schedules. The same suite tonight, on demand and in CI.
This is not an either-or choice. Keep Playwright MCP for interactive browser work in your editor. Add the Wopee.io MCP server and the same coding agent can dispatch the AI tester, generate test suites, fetch run results, and file the bugs it finds, without leaving the conversation.
Let your coding agent set it up
Set me up for Wopee.io: follow https://wopee.io/agent-setup/prompt.md, show me the plan first, and ask before changing anything.
Paste it into your agent’s chat. It shows you the plan before it changes anything.What the prompt does
No coding agent? Start for free in the browser → · No credit card required

Watch Playwright's planner, generator and healer next to a coding agent, the Playwright CLI and Wopee.io with MCP, on one app and one task.
Watch the 22:56 video →Playwright MCP is open source. The cost is everything you build around it yourself: schedulers, baselines, reporting, and the maintenance of tests that only exist in a context window.
Wopee.io's pricing is on the page, not behind a demo form: start free, then 19 to 179 € per user per month, Enterprise custom. The LLM behind every agent run is included in your plan, so there is no separate model bill to manage.
Yes. The wopee-mcp server connects your AI coding agent (Claude Code, Cursor, VS Code, or Windsurf) to Wopee.io. Your agent dispatches testing agents, generates test artifacts, and fetches results, while Wopee.io keeps the persistent suites, visual baselines, and scheduled regression runs.
Playwright MCP executes what your LLM decides in the moment. It can persist browser profiles and session state, but not tests. Regression testing needs the same tests to run repeatedly: saved suites, visual baselines, scheduled runs, and an evidence trail. That system is what Wopee.io adds on top of the browser automation.
Since Playwright 1.56, Microsoft ships agent definitions that plan, generate, and repair Playwright tests, driven by your own AI tool and LLM account. They are a strong DIY building block. Wopee.io is the managed version of that loop: hosted agents, visual baselines, scheduling, and team review, with no orchestration to build. See our Wopee.io vs Playwright CLI comparison for that story.
Your test cases are plain-language files in a Git repository your team can access, and you can download them as JSON. Running them with the AI agent is a Wopee.io service. Passing tests can also be generated as Playwright code, which uses the Wopee.io SDK for visual checks.
You start free with no credit card. Paid plans are published: Starter 19 € per user per month, Basic 79 €, Premium 179 €, and Enterprise custom with on-premise deployment.
Last checked: August 2026. Playwright MCP and Playwright Test Agents are open-source projects by Microsoft; their capabilities are described from their public documentation.