Record the scenario once. Ship the whole test stack.
QA-FORGE turns a described or imported user scenario into manual test cases, runnable Playwright / Selenium / Robot suites, k6 and API tests, and CI pipelines — saved per project, on your device. No account, no server, no sample data.
Name, base URL, frameworks, browsers, CI target, environment variables. Secrets stay as names only.
Record with the DevTools snippet, build steps in the editor, paste Playwright codegen output, or import a HAR to capture the API layer.
Test-case CSV, framework suites, k6, Newman collection, GitHub Actions or Jenkins pipeline.
Download the project bundle as a zip, commit it to your repo, add secrets in CI, run.
Drop JUnit XML or a Playwright results.json after each run. Pass rate, durations, failures — tracked per project.
Flakiness scores from your real runs, plus a rule-based smoke / nightly / weekly recommendation.
What this tool does — and honestly doesn't
Does: persists every project, scenario, artifact and run result in your browser's IndexedDB; generates real, runnable code from your steps; parses real HAR files and real CI result files; calls real AI providers with keys you supply. It also drafts steps from pasted user stories (opt-in AI), lints scenarios for brittle selectors and missing assertions, generates Page Objects and Gherkin features, exports Jira/TestRail-ready CSVs, diffs runs (what broke since yesterday), clusters failures by signature, and produces a self-contained stakeholder report you can email.
Doesn't: record other websites from this tab — browsers block cross-origin recording, so recording happens inside your app's own page via the paste-in-DevTools recorder in Scenario Studio (alongside codegen paste and HAR import). Playback happens through the generated suites — npx playwright test --headed or --ui gives you watchable, steppable replay — not inside this tab, and results come back here as files.
Your data: never leaves this device except when you press an AI assist button, which sends only the scenario you're editing to the provider you configured. Back up or move workspaces from Settings.
Five-minute start
1 Projects → New project → fill base URL, tick Playwright, tick chromium, choose GitHub Actions.
2 Scenario Studio → New scenario → add steps (goto → fill email → fill password → click → expect visible).
3 Generate → Download project bundle (.zip) → unzip into your repo → npm ci && npx playwright test.
4 Add TEST_USER_EMAIL / TEST_USER_PASSWORD as CI secrets. Push. Pipeline runs.
5 Runs & Health → import results/junit.xml from the run → watch the trend line.
Projects
New project
Scenario Studio
Generate artifacts
Runs & Health
Settings
AI assist providers
Optional. Keys are stored only in this browser (IndexedDB) and sent only to the provider you choose, directly from your device. Providers are tried top-to-bottom until one answers. Ollama and LM Studio run locally and need no key.
Workspace backup
Exports every project, scenario and run as one JSON file. Import merges by ID — safe to move between machines. AI keys are excluded from exports.
Storage
About
QA-FORGE workbench · local-first PWA · vanilla JS, zero external dependencies. All generation is deterministic templating from your inputs; AI assist is clearly labelled and opt-in. Works offline after first load.