Watch 3 AI agents turn one ticket into a verified Playwright test.
We build an agentic QA workflow live with Claude Code, Playwright and Playwright MCP. One agent writes the test, one checks it against the ticket, one refactors it into Page Objects. You see every prompt, browser action and line of code.
- 🛠Test Automation Agentexplores the app via Playwright MCP → writes the test
- 🔎Validator Agentchecks coverage, runs it → reports gaps
- 🧹Page Object Refactorerpulls interactions into reusable Page Objects
- ✓Run again & verifysame test, cleaner code, checked result
The workflow
One ticket. Three agents. Each with one clear job.
We start with a ticket and its test steps. Each agent does its part, and we check the result before handing it to the next one.
Save my free seat →- 01🛠 Test Automation AgentReads the ticket, explores the app through Playwright MCP and writes a Playwright test.OutputA first working test
- 02🔎 Validator AgentChecks whether the test actually covers the ticket's expected results, runs it and reports any gaps or failures.OutputA coverage and failure report
- 03🧹 Page Object RefactorerOrganizes the validated test's page interactions into reusable code.OutputClean, reusable Page Objects
- 04✓ Run it again and verifyWe run the refactored test one more time and confirm the result still holds.OutputA verified test you can trust
Nothing hidden
You see every step, not just the final result.
The setup and the prompts
How each agent is set up in Claude Code and exactly what it's told, so you can rebuild the workflow yourself.
Browser actions and code
Playwright MCP exploring the app live, and the test code each agent generates as it goes.
When things go wrong
What should happen when an agent misses a requirement or a test fails, and how to catch it before it ships.
Which part should we spend more time on?
Pick the part you care about most and save your seat. We'll shape the session around your answers.
Getting AI to write a test is easy. Knowing it actually tests the ticket is the real job.
An AI agent can write a Playwright test in seconds. The hard part is everything around it: does the test cover every expected result, does it fail when it should, and can your team still maintain it next month?
That's what this workshop is about. Not one clever prompt, but a workflow where each agent has a clear job and every result gets checked.
No 60-slide presentation. We build it live.
01Start with a ticket
A ticket with its test steps and expected results. That's the only input the workflow gets.
02Give each agent a job
Three agents with clear roles: write the test, validate it, refactor it. You see how each one is set up and exactly what it's told.
03Watch it in the browser
Playwright MCP drives the app live, so you see what the agent explores before it writes a single line of test code.
04Catch the gaps
The Validator checks the test against the ticket and runs it. We talk through what should happen when an agent misses a requirement or a test fails.
05Refactor and re-run
Page interactions move into reusable Page Objects. Then we run the test again and verify the result still holds.
06Open floor
Your questions about agents, Playwright MCP, or bringing this workflow into your own team.
You'll leave knowing how to make AI agents work for your tests.
You're a junior or mid-level QA engineer and you keep wondering: "how do I actually use AI agents in my automation?"
No AI agent experience needed. We explain every step as we build it.
Stay to the end and you'll get a free one-pager of the entire workflow you watched, yours whether you ever buy anything from Codemify or not.
