Playwright reporter

@brittlehq/playwright-reporter plugs into Playwright’s reporter slot. It runs in-process with the test runner, posts one result per test, and forwards the framework’s native artifacts (trace, video, HAR) unchanged.

Install

pnpm add -D @brittlehq/playwright-reporter

Configure

Add the reporter to playwright.config.ts alongside whatever you already use:

playwright.config.ts

import { defineConfig } from '@playwright/test';

export default defineConfig({
  reporter: [
    ['list'],
    ['@brittlehq/playwright-reporter', {
      url: process.env.BRITTLE_URL,
      token: process.env.BRITTLE_TOKEN,
    }],
  ],
});

Brittle’s reporter runs in parallel with list, html, dot, and anything else. None of them interact. You don’t need to drop your existing reporters.

Run

export BRITTLE_URL=https://brittle.your-domain.com
export BRITTLE_TOKEN=brt_svc_xxxxxxxxxxxxxxxx
pnpm exec playwright test

Open the dashboard while the run is in flight; sessions arrive as each test finishes.

What gets captured

• Test result (pass / fail / skipped / timed-out / flaky).
• Full retry history. Every attempt is its own row.
• Playwright trace (when trace: 'on' or 'retain-on-failure').
• Video (when video: 'on' or 'retain-on-failure').
• HAR (when configured via contextOptions).
• All test.info().annotations (owner, severity, issue links).
• test.tag() values, attached to each session.
• Per-attempt screenshots from expect(...).toHaveScreenshot().

The reporter doesn’t enable any of these for you. They stay controlled by playwright.config.ts. Whatever Playwright would have written to test-results/ on a local run is what arrives on the hub.

Playwright-specific options

In addition to the shared options, the Playwright reporter accepts:

Option	Type	Default	Notes
runMetadata	RunUpsertRequest	{}	Free-form run-level metadata (branch, commit, target). Overrides git auto-detection.
browser	'chromium' | 'firefox' | 'webkit'	auto-detected per project	The browser identifier stamped on each session. Defaults to the project’s use.browserName.

['@brittlehq/playwright-reporter', {
  url: process.env.BRITTLE_URL,
  token: process.env.BRITTLE_TOKEN,
  runMetadata: {
    branch: process.env.GITHUB_REF_NAME,
    commitSha: process.env.GITHUB_SHA,
    target: 'staging',
  },
  tags: ['nightly', 'smoke'],
}],

Sharding & parallel workers

With --workers=N or --shard=1/4, each process has its own reporter. To converge them onto one Run row:

1. Auto-detect. In CI, the reporter picks up your CI’s build ID automatically. No config needed.
2. Explicit. Set runName in your reporter config (or BRITTLE_RUN_NAME in env) to a value that’s stable across all shards. E.g. BRITTLE_RUN_NAME=${GITHUB_RUN_ID}-pw.

Set runMetadata (branch, commit, tags) consistently across workers so they don’t disagree.

Common stumbles

• testInfo.attach() for arbitrary blobs ships as session artifacts. They land on the Session Detail page’s Artifacts tab, alphabetically by name.
• No per-test logs by default. Set BRITTLE_DEBUG=1 to see every post the reporter makes; otherwise only warnings reach stderr.

Next steps

• Configuration reference for every option, type, and default.
• Reporter overview for concepts that apply to every framework.