Playwright Report MCP

An MCP server that exposes Playwright JSON test reports as tools for LLM agents.

Requirements

• Python >= 3.10

• uv (brew install uv or curl -LsSf https://astral.sh/uv/install.sh | sh)

• Node.js + npx — only for mcp dev, which launches the MCP Inspector

Related MCP server: Tesults MCP

1. Install dependencies

From the repository root:

uv sync

This creates .venv/ and installs the project in editable mode. Every command below is run through uv run, so you never need to activate the venv manually.

2. Produce a Playwright JSON report

The server reads a Playwright report produced by the json reporter. In your Playwright project:

npx playwright test --reporter=json > results.json

Or configure it permanently in playwright.config.ts:

export default defineConfig({
  reporter: [['json', { outputFile: 'results.json' }]],
});

The file must contain the standard top-level keys config, suites, errors, stats.

A sample report is bundled at results.json if you just want to try the server out.

3. Configure the report path

Create a .env file in the repository root pointing at that report (absolute path recommended):

REPORT_PATH=/absolute/path/to/results.json

To use the bundled sample instead:

REPORT_PATH=/absolute/path/to/playwright-test-analysis-mcp/src/playwright_report_mcp/results.json

REPORT_PATH is read at every tool call, so you can regenerate the report without restarting the server. If it is unset, tool calls fail with REPORT_PATH environment variable is not set.

4. Run the server

Interactive development (MCP Inspector)

uv run mcp dev src/playwright_report_mcp/server.py

Opens the MCP Inspector in your browser, where you can list and invoke the tools by hand. Requires npx.

Plain stdio server

uv run playwright-report-mcp

or, equivalently:

uv run mcp run src/playwright_report_mcp/server.py

Speaks the MCP protocol over stdin/stdout. Useful for wiring into an MCP client, or for a quick smoke test:

printf '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"t","version":"1"}}}\n' \
  | uv run playwright-report-mcp

You should get back a JSON-RPC result with "serverInfo":{"name":"Playwright Report",...}.

Connect it to a client

Register the server with Claude Desktop:

uv run mcp install src/playwright_report_mcp/server.py

Or add it manually to an MCP client config (e.g. claude_desktop_config.json):

{
  "mcpServers": {
    "playwright-report": {
      "command": "uv",
      "args": [
        "--directory",
        "/absolute/path/to/playwright-test-analysis-mcp",
        "run",
        "playwright-report-mcp"
      ]
    }
  }
}

--directory matters: it makes uv resolve this project's venv and lets load_dotenv() find the .env file.

Tools

Running the tests

pytest and pytest-asyncio are installed, but the repository has no test files yet, so this currently collects 0 tests:

uv run pytest

Troubleshooting