Question 1

What is Testver, in one sentence?

Accepted Answer

Testver is a self-hosted, AI-native control plane for every major test framework — recording, code generation, test management, dashboards, and a polished Git workflow, all in one tool.

Question 2

How do I install Testver?

Accepted Answer

Set up Testver on a developer laptop, internal server, or VM in minutes. Point it at a folder containing your existing test code and the UI lights up — no database to provision, no agents to deploy across your team, no cloud account to create.

Full installation steps live in the documentation at https://www.docs.testver.com/.

Question 3

Do I have to rewrite my existing tests?

Accepted Answer

No. Testver wraps your existing project rather than replacing it. Tests stay in the same files, run on the same framework runner, and live in the same repo — whether that's Playwright, Cypress, Selenium, WebdriverIO, Cucumber, REST Assured, PyTest, Puppeteer, or Robot Framework. Testver adds a UI, AI assistance, and lifecycle tooling on top.

Question 4

What does Testver detect about my project automatically?

Accepted Answer

Framework type, test file conventions, page-object patterns, fixture locations, environment-variable usage, locator strategies, and existing tags / metadata. This drives style-aware code generation so AI-written tests look like the rest of your codebase.

Question 5

Which test frameworks are supported?

Accepted Answer

Nine frameworks are first-class: Playwright, Selenium, Cypress, WebdriverIO, Cucumber, REST Assured, PyTest, Puppeteer, and Robot Framework. Each has a dedicated adapter that translates Testver actions into native framework calls — recording, code generation, and runtime instrumentation all use the framework's own primitives, not a shim.

Framework support is pluggable; additional adapters can be added without changing the core platform.

Question 6

Which programming languages can the AI generate tests in?

Accepted Answer

TypeScript and JavaScript are primary across the JS-based runners (Playwright, Cypress, WebdriverIO, Puppeteer, Cucumber-JS). Python is supported for PyTest and Robot Framework. Java is the native language for REST Assured and is supported for Selenium. C# is supported for Selenium.

The generator follows your project's existing conventions — naming, file layout, helper patterns — rather than imposing a fixed style.

Question 7

Does Testver support BDD / Cucumber?

Accepted Answer

Yes. Cucumber is one of the nine first-class frameworks. Feature files are recognized and editable through the same UI; the AI can generate step definitions from feature scenarios, and recordings can be exported as Cucumber steps when the project is configured that way.

Question 8

Can different projects in the same Testver instance use different frameworks?

Accepted Answer

Yes. Framework choice is per-project. One Testver instance can host a Playwright project, a PyTest project, and a Robot Framework project side-by-side — or any other mix of the supported frameworks. The AI assistant, dashboards, scheduler, and reporting all work uniformly across them.

Question 9

What about mobile testing?

Accepted Answer

Mobile-web testing via Playwright (mobile emulation) and Selenium Grid (real-device clouds) is supported today. Native mobile (Appium / Detox) is on the roadmap.

Question 10

Which LLM providers are supported?

Accepted Answer

OpenAI, Anthropic, Google Gemini, xAI Grok, and any OpenAI-compatible endpoint (Groq, Together, OpenRouter, vLLM, etc.) are supported out of the box. For local models, both Ollama and LM Studio are supported — pick whichever local runner you already use.

Question 11

Can I run Testver fully offline?

Accepted Answer

Yes — point the LLM router at a local Ollama or LM Studio instance (Llama 3, Qwen, Mistral, DeepSeek, etc.) and the entire AI surface works without an internet connection. Recording, code generation, and chat all run against the local model.

Some features that depend on frontier-class quality (large-context refactors, very long failure analysis) read better with hosted models like GPT, Claude, Gemini, or Grok — but nothing is gated.

Question 12

What data is sent to the LLM provider?

Accepted Answer

Only what the current task needs: the prompt, the relevant code or test artifacts, and the conversation history for that session. There is no telemetry channel — Testver does not phone home with usage data.

When using a hosted provider, that provider's privacy policy applies to the prompts you send. Use Ollama, LM Studio, or a self-hosted OpenAI-compatible endpoint (vLLM, Text Generation Inference, etc.) if you need that data to stay on your infrastructure.

Question 13

Can different projects use different LLM providers?

Accepted Answer

Yes. The provider, model, temperature, and tool-execution policy can all be configured per-project. One team might use Claude for code generation, another might use Gemini for large-context refactors, and a third might run Llama via Ollama for cheap, fast log summarization — all on the same Testver instance.

Question 14

What can the AI assistant actually do — and is it safe?

Accepted Answer

The assistant has a curated tool palette: file reads, file writes, browser automation actions, test execution, and git operations. Every tool call goes through a permission gate by default — the user explicitly approves (or denies, or "always allow") each tool the assistant tries to invoke.

Permissions are per-user, per-tab, and audited. Nothing happens to your repo or your test runs without an explicit grant.

Question 15

Where does my data live?

Accepted Answer

Wherever you install Testver. Project metadata, run history, recordings, and reports are stored locally in SQLite databases under a configurable data directory. There is no required cloud service.

For multi-user deployments, point the data directory at shared storage (NFS, EFS, etc.) and front Testver with a load balancer.

Question 16

Does Testver send anything to the cloud by default?

Accepted Answer

No. Out of the box, the only network traffic is between the browser, the Testver server, and the LLM provider you configured. There is no telemetry endpoint, no analytics SDK, and no automatic update check.

Question 17

How does authentication work?

Accepted Answer

Local username/password with bcrypt + JWT sessions out of the box. SSO via OIDC and SAML is available for enterprise deployments — sessions are recorded so admins can revoke any active token. Pre-shared API tokens cover machine-to-machine integrations.

Question 18

Is there role-based access control?

Accepted Answer

Yes — admin, developer, and viewer roles ship by default, with per-resource permissions (projects, runs, reports, connectors). Role assignments are audited, as are sensitive actions (credential changes, RBAC edits, project deletion).

Question 19

Is there an audit log?

Accepted Answer

Yes. Every authentication event, every connector touch, every git mutation, every credential prompt resolution, and every privileged settings change is logged with timestamp, user, IP, and action context. The log is queryable from the admin UI and exportable as CSV / JSON.

Question 20

How do I run tests automatically — is there CI integration?

Accepted Answer

The built-in scheduler is the primary path today. Schedule any test, suite, or report on a cron — daily smoke, weekly regression, monthly reporting — and Testver handles dispatch, retries, notifications, and report distribution without an external orchestrator.

Webhook-triggered runs from external CI systems are on the roadmap; they aren't available today.

Question 21

Can Testver push defects to Jira?

Accepted Answer

Yes. The Jira connector is bidirectional — a failed test can auto-create a defect, link it to the run, and update its status when the test passes again. Defects opened externally can be linked to test cases for traceability.

Other defect trackers can be wired up via custom connector configuration; Jira is the only one that ships out of the box.

Question 22

Does it integrate with Slack and Microsoft Teams?

Accepted Answer

Yes — as outbound notification destinations. Run summaries, scheduled reports, and on-failure alerts can be dispatched to a Slack channel, a Microsoft Teams channel, or via email. Configure once and the destination shows up everywhere a notification rule applies.

Question 23

Which cloud test-runners are supported?

Accepted Answer

Testver works with BrowserStack, Sauce Labs, and LambdaTest. Configure credentials once, then dispatch any test or suite to run on real browsers and devices in the cloud — artifacts (video, traces, screenshots) flow back into the Testver report alongside locally-run tests.

Question 24

Can I trigger Testver runs from a webhook?

Accepted Answer

Not yet — webhook-triggered runs are on the roadmap but not available today. For now, use the built-in scheduler for cron-driven runs and the "Run now" override in the UI for ad-hoc dispatches.

Question 25

Is Testver free?

Accepted Answer

Yes — Testver is completely free. No paid tiers, no test-run limits, no project limits, no AI-credit gating, and no feature gating between editions. Bring your own LLM keys for any supported provider — OpenAI, Anthropic, Google Gemini, xAI Grok — or run a local model via Ollama or LM Studio. Either way, you get the full product.

Question 26

How do I get support?

Accepted Answer

Visit the support portal at https://www.support.testver.com/ to open a ticket, browse known issues, or talk to the team.

Question 27

Where is the documentation?

Accepted Answer

The full documentation lives at https://www.docs.testver.com/ — installation guides, configuration reference, framework adapters, API docs, and walkthroughs for the AI assistant, recorder, dashboards, and connectors.

Answers to the questions we hear most

Getting started

Frameworks & languages

AI & LLMs

Self-hosting & security

Integrations & ecosystem

Licensing & support

We're here to make adoption painless

Documentation

Support

Community

Bring AI to the framework your team already uses