Evenfire is an intelligence layer platform for running LLM agents that take real actions, on your infrastructure
TypeScript
import { defineConfig } from '@playwright/test'
import path from 'node:path'
import { loadQaRecorderEnv } from '../scripts/qa-recorder/loadEnv'
const repoRoot = path.resolve(__dirname, '..')
loadQaRecorderEnv(repoRoot)
const recorderRoot = process.env.QA_RECORDER_ROOT
? path.resolve(process.env.QA_RECORDER_ROOT)
: path.join(repoRoot, '.local-notes', 'qa-recorder')
const browserChannel = process.env.QA_RECORDER_BROWSER_CHANNEL?.trim()
const slowMo = Number(process.env.QA_RECORDER_SLOW_MO_MS || 75)
if (!Number.isFinite(slowMo) || slowMo < 0) {
throw new Error('QA_RECORDER_SLOW_MO_MS must be a non-negative number.')
}
export default defineConfig({
testDir: './e2e',
// All Control UI recorder journeys share this config. Each journey lives in
// its own qa-recorder-<journey>.spec.ts and is run via a namespaced
// qa:recorder:<journey> command (see package.json). Stays isolated from the
// normal Playwright/CI suite — only recorder specs match.
testMatch: /qa-recorder-.*\.spec\.ts/,
timeout: 300_000,
expect: { timeout: 30_000 },
reporter: [['list']],
outputDir: path.join(recorderRoot, 'runs', 'control-ui'),
preserveOutput: 'always',
retries: 0,
workers: 1,
use: {
baseURL: process.env.CONTROL_UI_URL || 'http://127.0.0.1:3000',
headless: process.env.QA_RECORDER_HEADLESS === '1',
screenshot: 'on',
trace: 'retain-on-failure',
video:
process.env.QA_RECORDER_VIDEO === '0'
? 'off'
: {
mode: 'on',
size: { width: 1280, height: 720 },
},
viewport: { width: 1280, height: 720 },
},
projects: [
{
name: 'qa-recorder-chromium',
use: {
browserName: 'chromium',
...(browserChannel ? { channel: browserChannel } : {}),
launchOptions: { slowMo },
},
},
],
})