Open the side panel on any page. Or paste a user story in plain text. TestLens returns happy, edge, and negative cases in Playwright, Cypress, Gherkin, or Markdown.
Point TestLens at any rendered web page and it reads the forms, buttons, links, and ARIA roles in the DOM. Most AI test tools restrict you to a fixed catalogue of sites or a single flow like signup. TestLens works on whatever is on screen — a checkout, a dashboard filter, a search page, or something your team shipped yesterday.
Skip the page entirely and paste a user story in plain text — TestLens generates the same happy, edge, and negative cases. Most page-scanning tools only work when you can already load the app. This mode is useful before the UI exists, on internal tools you cannot expose, or when a ticket is the only artifact you have.
Pick Gherkin, Playwright, Cypress, or Markdown and download a file that is ready to commit. No massaging the output, no stripping prompt fragments or JSON. The generated .feature, .spec.ts, .cy.ts, or .md file preserves preconditions, steps, and expected results — you paste it into your repo and run it.
Bring your own Anthropic or OpenAI key — it is stored locally in the browser and used for direct calls to the provider. There is no TestLens server in the path, so your page content and keys never transit through us. A typical Pro user runs about 20 generations per month for roughly 0.36 USD in provider cost, billed by Anthropic or OpenAI — not by us.
Every run returns three kinds of cases: the nominal path, the boundaries, and the error routes. Coverage is enforced by the prompt contract and the zod schema, not a preference you remember to toggle. Ten cases per run is the default; you can edit or delete any of them in the side panel before export.
One price, kept forever. No seat fee, no auto-renewal, no price tier that creeps upward once you depend on the tool. Payment and license keys are handled by Gumroad; the extension itself calls Anthropic or OpenAI directly with your own key, so your recurring cost is whatever your provider charges — usually under a euro a month on typical volume.
TestLens is a Bring-Your-Own-API-key tool. You paste your own Anthropic (Claude) or OpenAI key in the Options page; it is stored in chrome.storage.local on your machine and used for direct calls to the provider. TestLens Labs does not run an API key of its own, does not hold yours, and does not relay your requests through any server we operate.
Source: SPEC-testlens.md §4, FAQ Q1, privacy-policy.md §2.
The extracted page structure (forms, buttons, links, ARIA roles) or the user story text you type is sent directly from your browser to your chosen LLM provider — never to a TestLens server. Your license key is sent to Gumroad for verification. Nothing else leaves the browser. Page text content is off by default and only included if you toggle “Include text content” on; when it is, a regex filter scrubs email and phone patterns before the request is built.
Source: privacy-policy.md §4, SPEC-testlens.md §4.
At launch (v0.1), Pro users export to:
.feature) — Cucumber / SpecFlow-compatible.spec.ts).cy.ts)Each exporter preserves preconditions, steps, and expected results. Jest, Selenium, Robot Framework, and Postman collections are not yet supported; you can request one by email or by opening an issue on the GitHub repo.
Source: src/exporters/index.ts, FAQ Q2, SPEC-testlens.md §3.
Pro is a one-time 9€ purchase. Within 30 days of purchase, if TestLens does not fit your workflow, email support@testlens.dev and we will process a full refund through Gumroad — no questions asked. After 30 days we still read every email but cannot guarantee a refund. Your provider bills (Anthropic or OpenAI) are separate and out of scope.
Source: FAQ Q6, refund-policy.md §2.
The extension itself runs locally, and exporting previously generated cases from your history is fully offline. Generating new cases requires an internet call to your chosen LLM provider (api.anthropic.com or api.openai.com). License verification contacts api.gumroad.com on cold start, with a 7-day grace period if Gumroad is unreachable — short outages will not lock you out.
Source: FAQ Q5, SPEC-testlens.md §4.
TestLens Labs does not store your page content, user stories, generated cases, or API keys — there is no backend to store them on. Support emails are retained under the periods listed in the privacy policy. Gumroad retains purchase and license data under its own terms as merchant of record.
Source: privacy-policy.md §4 and §5.
Either works. Claude Sonnet 4.6 is the default. OpenAI is available as a fallback provider. You can switch providers in the Options page at any time; the export formats and case taxonomy do not change.
Source: SPEC-testlens.md §4 LLM client row, src/generator/generate.ts.
Source: FAQ Q1, SPEC-testlens.md §4 data flow.
The generation request fails with the provider's error message — typically insufficient_quota or a 429 status. TestLens surfaces the error in the side panel and does not consume a Free-tier daily credit when the provider call fails at that stage. Top up your Anthropic or OpenAI credits and retry; nothing on the TestLens side needs to change.
Source: src/generator/generate.ts:131-136 (GenerateErrorCode = 'llm_error'), license gating logic in src/license/gating.ts.
Yes. The content script reads the rendered DOM of the active tab, whether that tab is served from localhost:3000, a .internal host, or a staging URL behind SSO. The only network calls TestLens itself makes are from your browser to your chosen LLM provider and — for licensed users — to Gumroad for verification. If your LLM provider is reachable from the machine, generation works.
Source: SPEC-testlens.md §4 permissions (activeTab only), src/content/extract.ts.
Yes, subject to your employer's policies on browser extensions and on sending request context to third-party LLM providers. TestLens Labs does not receive or store your company's data. The BYO-key model means your employer — not us — decides which provider sees the traffic; if your employer already has a vetted Anthropic or OpenAI account, you can use that key. If you need a single-tenant deployment or a team license, the Team tier is planned but not at launch.
Source: privacy-policy.md §4, SPEC-testlens.md §5 Team tier row, FAQ Q7.