AC Generator
PublicPaste a user story or description, attach any supporting files (PDF, DOCX, images, code, JSON, plain text — anything readable) and get Gherkin-style acceptance criteria. No sign-in required.
Open →From a one-line user story all the way to a green CI build — the tools, the runners, and the dashboards that take you there.
Turn requirements into testable artifacts in seconds. Every output is editable, copy-pasteable, and downloadable — no lock-in.
Paste a user story or description, attach any supporting files (PDF, DOCX, images, code, JSON, plain text — anything readable) and get Gherkin-style acceptance criteria. No sign-in required.
Open →Convert acceptance criteria into a comprehensive set of manual test cases — happy path, negative, boundary, edge, validation, security, accessibility. Choose Steps or BDD (Cucumber) output. Export as JSON, CSV, or .feature.
Open →Generate production-ready test automation code for Playwright, Cypress, Selenium, WebdriverIO, Puppeteer, or Robot Framework — in TypeScript, JavaScript, Python, Java, C#, or Ruby (whichever each tool supports). Pick Standard or BDD/Cucumber style. Download as a runnable project zip.
Open →Run real Playwright tests against a real browser — locally, in CI, or on a fleet of agents that you control.
Hook generated tests into your CI pipeline (GitHub Actions, GitLab CI, Jenkins, CircleCI). Tests use the canonical runner for each framework (`npx playwright test`, `npx cypress run`, `mvn test`, etc.).
Spin up TestPerfect Agents on any machine — your laptop, a Docker container, a remote VM — and dispatch tests to them from the dashboard. Live progress, step-by-step screenshots, console logs streamed back over WebSocket.
A first-class test type with its own step vocabulary — `Make POST request to /api/sessions`, `Set bearer token`, `Assert status 201`, `Assert JSON path "$.data[0].status" equals "pending"`, `Assert response time under 500 ms`. Chain calls by extracting a token or id from one response into a variable the next request uses. Print steps (response, headers, status) for debugging without leaving the tool. API runs land in the same execution history as UI and manual runs.
Steps like `Fill "standard_user" in "Username"` resolve against label, placeholder, role, data-testid, aria-label, and naming-convention variants (kebab/snake/camel/Pascal) — so flaky locators stop being the bottleneck.
Every layer of generation, code emission, and execution strips wrapping quote characters so values like `"standard_user"` are never typed verbatim into form fields.
The resolution ladder was always a form of healing — it just did it silently, so a run that only passed on a later strategy looked like a clean pass. Every resolution is now recorded: which strategy won, what the step asked for, and a durable selector worth writing back. Where a step named an explicit selector that matched nothing, the proposal is actionable — accept it and the test case is rewritten so the next run uses the locator that actually worked; reject it and nothing changes. Resolutions that merely limped are recorded as informational, never auto-applied.
Per-test flake scores computed from run history. A test that fails every run is broken, not flaky — only a test that disagrees with itself scores. Same-commit flips count as proof; outcome sequences as inference, with confidence stated. Above the threshold, quarantine in one click: the test still runs and is still scored, but cannot fail the build. Release it when its history is clean.
Bring your team, your sources of truth, and your existing tools into one place. Sign-in unlocks these.
Group test cases, AC, agents, and runs by project. Owner / Admin / Member roles with per-release permissions.
Pull user stories from Jira or Azure Boards, convert them to AC + test cases in one click, and write results back as comments or linked work items.
Every test case is versioned. Compare diffs, roll back to a prior version, and see who changed what.
Per-user OpenAI / Claude / Gemini / HuggingFace / Grok keys, stored encrypted. Generation on your own key is uncapped and unmetered on every paid tier, and the active provider is recorded per release.
Centrally owned data sets with typed columns, scoped to a project or release and linked to the test cases that use them. Write `{{column}}` in a step once and the run executes per row. Data sets are not versioned yet — there is no history or rollback on them the way there is on a test case.
See what passed, what flaked, and where to invest next.
Every run is recorded with step-level results, screenshots, console output, and timing. Filter by release, agent, branch, or date range.
Pass rate, flaky-test detection, average duration, coverage trends. Export as PDF or share a public link.
Drop the test artefacts straight into your existing CI report system, or use the TestPerfect report viewer with deep links to failing steps.
Add a `Visual check "login page"` step and the first capture becomes its baseline; every later run is pixel-compared against it, with anti-aliasing tolerated and named regions (`ignoring .timestamp`) masked. Mismatches land in a review queue with baseline, candidate and diff side by side — approve to promote the candidate, or reject. The agent freezes animations, hides the caret and waits for fonts before every capture so unchanged pages diff clean.
For regulated teams, large QA orgs, and anyone who needs the keys to stay inside their own VPC.
Run the full TestPerfect stack (web + agent hub + MCP servers) inside your own infrastructure via Docker Compose or Helm.
SAML for Okta, OneLogin and Ping, plus SCIM 2.0 for automated joiner/mover/leaver handling. Scoped and set up in discussions with our Enterprise team — talk to sales if it gates a deployment.
Append-only log of sign-ins and sign-in denials, password and API-token changes, and organization deletion, restore and permanent purge. Credentials are redacted before write, and the log is queryable and CSV-exportable for review.
Pinning your data to a chosen region (UK, EU, US). Available today only as part of a self-hosted deployment, where you choose the infrastructure — talk to sales about managed residency.
Connect to your internal tools (incident tracker, test data service, observability) via the Model Context Protocol — we build the connector with you.