Live · Free
Codegen to page objects
Paste what npx playwright codegen recorded. Get back plain Playwright: page objects with their locators, a test that calls them, and secrets in environment variables.
Playwright TypeScriptRuns in this tab
Blendra turns one recording into end-to-end tests as a Playwright project: specs, page objects, locator files, test data and a GitHub Actions workflow. No LLM writes the code: same recording, same output. It runs in your CI without Blendra.
Free plan, no credit card. Try Pro free for 7 days.Free plan, no credit card. Sign up now, record later in desktop Chrome.
No extension: every new account can open a sample project.
Demo · 1:00
Recorded on saucedemo.com, a public demo shop. Captions are in the video, no sound needed.
npx blendra CLI on npm, MIT (opens in a new tab)Recording in Blendra's browser
It is a browser on Blendra's servers, shown in your Blendra tab. Open your app in it and use it as you normally would.
It needs a public address. To record localhost or a private app, use the Blendra Recorder extension in desktop Chrome.
How it works · 01 Record
Install Blendra Recorder in Chrome and use your app the way you always do. Every click, entry and page change is saved with a locator chosen in a fixed order: test ids first, then role and name, then text. Nothing is made up.
After each step, Blendra lists what your app showed: a message, a new URL, a heading. Tick the outcomes that must stay true, and each one becomes a check holding the value your app really displayed. You can also write checks yourself, reuse journey-level templates, or ask AI to suggest picks you accept or ignore. An export in which no test checks an outcome is blocked.
Checks catch what you chose to check. They don't promise a perfect suite.
A real screen from the app, with a demo project.
02 Build
Export a standalone project in Playwright (TypeScript or Python) or Cypress: specs, page objects, locator files, scenario data and environments, plus a GitHub Actions workflow in .github/workflows/e2e.yml. It runs without Blendra. For developers and QA engineers who want tests they'd keep in the repo.
npx blendra pull updates the generated project without overwriting your local edits (Pro and up).BLENDRA_REPORTER=off and the suite never contacts Blendra.acme-e2e25 files
tests/login/login.spec.tsPlaywright TypeScriptTypeScript
Lines 9 to 13 are the five recorded steps.import { test } from '../../fixtures/base';import { blendraPlan, executeBlendraAssertion } from '../../utils/assertions';import { loginData } from '../../fixtures/test-data/login.data'; test.describe("Login", () => {test.describe("Sign in", () => {test("[blendra:cmh2q8x4k0001acme6hp01] Happy path", async ({ page, loginPage, rootPage, fieldLocators }) => {blendraPlan(["chk_dashboard_heading"]);await page.goto("/");await rootPage.clickLogin();await loginPage.fillEmail(loginData.happyPath.email);await loginPage.fillPassword(loginData.happyPath.password);await loginPage.clickSignIn();await executeBlendraAssertion(page, {id: "chk_dashboard_heading",type: "element_visible",assertionTarget: {type: "role",role: "heading",name: "Dashboard",exact: true}}, fieldLocators);});… 34 more lines in tests/login/login.spec.ts
$ npm test.github/workflows/e2e.yml · job Playwright
acme-e2e34 files
tests/login/test_login.pyPlaywright Python · New · Pro+ and TeamPython
Lines 23 to 27 are the five recorded steps.⋯ lines 1–12@pytest.mark.blendra_describe("Login")class TestLogin:@pytest.mark.blendra_describe("Sign in")class TestSignIn:@pytest.mark.blendra(id="cmh2q8x4k0001acme6hp01", name="Happy path", recording="Sign in")def test_happy_path(self, page, env):blendra_plan(["chk_dashboard_heading"])row = DATA["cmh2q8x4k0001acme6hp01"]login_page = LoginPage(page)root_page = RootPage(page)page.goto(env.url("/"))root_page.click_login()login_page.fill_email(row["email"])login_page.fill_password(row["password"])login_page.click_sign_in()execute_blendra_assertion(page,{"id": "chk_dashboard_heading","type": "element_visible","assertion_target": {"type": "role", "role": "heading", "name": "Dashboard", "exact": True},},FIELD_LOCATOR_MAP,)… 44 more lines in tests/login/test_login.py
$ pytest.github/workflows/e2e.yml · job pytest
acme-e2e33 files
tests/login/login.cy.tsCypress · New · Pro+ and TeamTypeScript
Lines 16 to 20 are the five recorded steps.⋯ lines 1–8describe("Login", () => {describe("Sign in", () => {it("[blendra:cmh2q8x4k0001acme6hp01] Happy path", () => {blendraPlan(["chk_dashboard_heading"]);const row = blendraRow(DATA, "cmh2q8x4k0001acme6hp01");const loginPage = new LoginPage();const rootPage = new RootPage();blendraVisit("/");rootPage.clickLogin();loginPage.fillEmail(row["email"]);loginPage.fillPassword(row["password"]);loginPage.clickSignIn();executeBlendraAssertion({id: "chk_dashboard_heading",type: "element_visible",assertionTarget: {type: "role",role: "heading",name: "Dashboard",exact: true}},fieldLocators,);});… 46 more lines in tests/login/login.cy.ts
$ npm test.github/workflows/e2e.yml · job Cypress
Excerpts of the exporter's real output for the demo recording. Highlighted: the five recorded steps.
Scenarios
Add scenarios to a journey: the happy path, negative and edge cases, each with its own data and checks, run against local, staging or production. Ask AI for scenario ideas when you want a starting list.
Every page you record also lands in the page library, with its elements, locators and shared checks. Build a new test from steps you already captured, without recording again.
A real screen from the app, with a demo project.
API tests
Write HTTP tests in the same project, with status, header, JSON path and response-time checks per environment. They export into tests/api/ beside your browser specs. SQL checks run inside Blendra only.
03 Run
Run a journey or a whole suite on Blendra's cloud runner, right away or on a schedule: hourly, daily, weekly or cron. Results from your own CI land in the same run history. When a scheduled run fails, you get an email.
Free includes 60 cloud runner minutes a month (up to 5 minutes per run), Pro 2,000, Pro+ 5,000 and Team 15,000 shared. Cloud runs need an address the internet can reach, not localhost.
04 Analyze
Every failed test gets a cause, a confidence (Certain, Likely or Possible), the evidence behind it and one next step, such as repairing a locator from the alternatives recorded with it. It is rule-based and free on every plan.
Want it in plain English? AI can explain a failure, using one AI request. It never rewrites your tests. Blendra also flags flaky scenarios whose recent results keep flipping, and regressions: a scenario that passed recently and now fails on the same environment.
A real screen from the app, with a demo project.
05 Document
One click drafts a readable test case from a recording: steps, data and expected results. Each line shows where it came from, observed in the recording or inferred, and the case stays “Needs review” until someone approves it. When the recording changes, the test case is marked out of date. Export to CSV, Markdown or PDF, with sensitive values masked.
Journey labels and the Proven count show which flows still need checks.
A real screen from the app, with a demo project.
All features
Every feature in a line, grouped by stage. Where this page already shows one, the link takes you there. What we're still building carries a Coming soon pill.
Soon: describe a test in plain words, and Blendra will build it and run it to prove it works. It will work from your recorded pages and checks, and you will review each test before it joins your suite.
On this pageRecorder, locators, evidence and checks · Sensitive fields
Split recordings
Each part is saved as its own recording.
Import a recording
From the extension, or from a session file.
The recording page
Five tabs, autosave with Undo, keyboard shortcuts.
14 browser check types
URL, text, visibility, field values, counts and more.
Capture elements
Point at elements to add them to the page library.
On this pageCode export · Scenarios and page library · API tests
Features and journeys
Group journeys into features, like Checkout.
The Export page
A quality score, and a checklist of what to fix.
Export layouts
Layout 2 adds feature folders and SUITE.md.
Suggested testsComing soon
New tests suggested from your page library.
Verified failure scenariosComing soon
Error-path tests, proven by a real run first.
On this pageCloud runner, schedules, CI results and failure emails
Usage alerts
Emails at 80% and 100% of your runner minutes.
History retention
Free keeps 7 days, Pro 90, Pro+ 180, Team 365.
Schedules by tagComing soon
One schedule for every test with a tag.
Team alerts
Failures and fixes by email, Slack, Microsoft Teams or a signed webhook.
Notification settingsComing soon
Email switches and alert channels in one place.
On this pageFailure causes, AI explanations, flaky tests and regressions
The Checks page
Each check's last 10 results, and copy changes.
The dashboard
Pass rate, results per day and unstable tests.
Project overview
Each project's health and its progress to Proven.
Root-cause summaryComing soon
One plain sentence on why a run failed.
On this pageAI test cases, out-of-date marks and exports
Test docs, all projects
Every test doc across projects, with coverage.
Report view
A test doc and its last run, ready to print.
Requirements coverageComing soon
See which acceptance criteria your tests cover.
Test-management toolsComing soon
TestRail, Xray, Qase, Jira, Confluence. Pro+ and Team.
On this pagePlans, seats and roles · Referrals and affiliates
Workspaces
Your own workspace, plus shared ones on Team.
Search
⌘K or Ctrl+K finds projects, recordings and actions.
REST API
Projects, recordings and runs. Pro and up.
What's new
Recent releases, and “I want this” on what's next.
EnterpriseComing soon
SSO, audit logs and a custom SLA. Contact sales
Security
Other websites can't start or read a recording.
Passwords, card numbers and five other kinds of sensitive field are dropped before a step is stored. The exported test reads each from an environment variable.
AI runs only when you press an AI button, and sensitive values are replaced first. It never writes a check's expected value, selector or URL. Each AI press uses 1 AI request from your plan's monthly allowance.
Free tools
Small converters that run in your browser tab. Nothing is uploaded, no AI is involved, and the code they give you is yours to keep.
Free, no sign-upNever uploadedNo AI, same output every time
Live · Free
Paste what npx playwright codegen recorded. Get back plain Playwright: page objects with their locators, a test that calls them, and secrets in environment variables.
Playwright TypeScriptRuns in this tab
Coming soon
Drop a HAR from your browser or npx playwright codegen --save-har. Get a k6 script you own: your site's requests only, tokens passed from response to request, and passwords in environment variables.
One user first, then loadRuns in this tab
Free when it ships.
Pricing
Free
$0
1 project · Playwright TypeScript
60 cloud runner minutes a month
Pro
For daily solo work
$79/mo
1 project · CLI and API tokens
2,000 minutes
Pro+
$119/mo
3 projects (+$99/mo per extra project) · adds Playwright Python and Cypress
5,000 minutes
Team
$399/mo
5 seats (+$50/mo per extra seat) · roles and invites
15,000 minutes shared
Try Pro free for 7 days.
No credit card. Nothing is charged. Everything you build or export during the trial stays yours. The workspace owner starts it after sign-up.
Free: 1 project, 5 recordings, 10 scenarios, 10 checks and 5 API tests in all, plus 5 AI requests a month.
Referrals and affiliates
Referral program
Share your link with another QA lead or a team at another company.
Affiliate program
Run a blog, channel or agency? Commission is counted on what each customer you refer pays, after discounts and before tax.
Plus a bonus after more than 4 months as a paying customer
Your referral link and the affiliate application are in Settings › Referrals.
Referral programAffiliate programCreate an accountHave an account? Sign in
FAQ
Blendra turns flows you record in Chrome into a test project you own, in Playwright (TypeScript or Python) or Cypress, with checks based on what your app showed, scenario data, a CI workflow and a readable test case. It also runs the tests, explains failures and keeps test cases up to date. TypeScript is on every plan; Python and Cypress on Pro+ and Team.
The Blendra Recorder extension saves each click, entry and navigation with a locator chosen in a fixed order (test ids, then role and name, then text). After each step it lists what your app showed; the outcomes you tick become checks with the real values. Export turns the journey and its scenarios into specs, page objects and data files.
Not to record, choose checks, add scenarios, run tests or read the failure causes and test cases: all of that happens in the app. The result is real test code (Playwright, or Cypress on Pro+ and Team), so editing it or wiring it into your own CI is ordinary code work. Blendra is built for teams that want that code.
Playwright TypeScript on every plan. Playwright Python (pytest) and Cypress are new, on Pro+ and Team: the same recordings, chosen per project; the preview names any step Cypress can't run.
Yes. The export is a standalone project that runs without Blendra; commit it and edit it like any code. On Pro and up, npx blendra pull updates it without overwriting your local edits. Set BLENDRA_REPORTER=off and it never contacts Blendra.
Yes. Each plan includes cloud runner minutes: Free 60 a month (up to 5 minutes per run), Pro 2,000, Pro+ 5,000, Team 15,000 shared. Runs can be scheduled hourly, daily, weekly or by cron. A cloud run needs an address the internet can reach, so use a staging or production URL, not localhost. A test that signs in asks for the password once: Blendra keeps it encrypted and types it in every cloud run.
Every export includes .github/workflows/e2e.yml. The project is plain Playwright (or pytest/Cypress), so any CI that runs it works. To send results back to Blendra, add an API token (Pro and up). In a GitHub Actions matrix, sharded jobs (Playwright's --shard) report as one run.
Yes: HTTP tests with status, header, JSON path and response-time checks per environment, exported into tests/api/. SQL checks against a database connection run inside Blendra only.
Blendra recognises sensitive fields (passwords, card numbers and CVCs, one-time codes, tokens, bank details, national IDs) and does not store what you type into them. The export reads each value from an environment variable. Detection can miss a field with a generic name; you can mark it sensitive yourself.
You get a cause with a confidence, the evidence behind it and one next step, free on every plan. AI can explain the failure in plain English if you ask; it never changes your tests.
One project with 5 recordings, 10 scenarios, 10 checks and 5 API tests, Playwright TypeScript export, 5 AI requests a month and 7 days of run history. Plus 60 cloud runner minutes a month. No card.
Yes: the workspace owner can start a 7-day Pro trial after sign-up. No card, nothing is charged. When it ends, the workspace returns to Free and nothing is deleted.
Team includes 5 seats with roles (Owner, Admin, QA lead, QA engineer, Developer, Viewer) and per-project access. Inviting teammates comes with Team.
Blend Emini, an independent engineer, builds and runs Blendra (trading as Blendra.app, as the Terms say). Payments run through Paddle. Every export is a standard test project that runs without Blendra, so your tests never depend on us. Write to info@blendra.app; the person who builds it answers. GitHub: BlendEmini (opens in a new tab)
Sign up with Google or emailPaid plans: full refund within 14 days, no reason needed