Add a Playwright MCP web QA agent example

Summary

Add an MCP example that turns a short QA brief into repeatable browser checks using Playwright. The goal is to teach safe, observable browser automation for a local demo application.

Proposed location

mcp_ai_agents/playwright_mcp_qa_agent/

What to build

Create a small local web page with a few intentional states, then build an MCP-enabled agent that can:

1. Read a QA brief such as "verify sign-in validation and the product search empty state".
2. Use Playwright tools exposed through MCP to inspect the page and interact with it.
3. Capture screenshots for failed checks.
4. Produce a structured QA report with pass or fail status, reproduction steps, and evidence paths.

The example should run against a local fixture app only. Do not require credentials or automate third-party websites.

Acceptance criteria

• Includes a README.md, dependency manifest, and .env.example if configuration is needed.
• Includes a local fixture app or static page that can be launched with documented commands.
• Uses MCP for the browser tool integration.
• Outputs a machine-readable JSON report and a human-readable Markdown summary.
• Includes at least two documented QA scenarios, including one intentional failure.
• Adds one catalog entry to the root README.md under MCP Agents.

Contribution notes

Please submit one self-contained project in one pull request and link it with Closes #<issue-number>.