Question 1

How does the AI unit test case generator work?

Accepted Answer

Paste a function (or a whole module) and pick a framework — Jest, Vitest, PyTest, or JUnit. The tool parses function signatures on-device, infers parameter kinds (number, string, array, object, boolean), and generates happy-path, edge-case, error-case, and parameterized/table-driven tests using Arrange-Act-Assert structure. It also scaffolds mocks for fetch and module imports, generates setup/teardown, and lists coverage-gap hints. Every test is labeled deterministic (signature-driven, reliable) or heuristic (best-effort — review the expected value).

Question 2

Which languages and frameworks are supported?

Accepted Answer

JavaScript and TypeScript with Jest or Vitest, Python with PyTest, and Java with JUnit 5. Each framework gets idiomatic assertions (expect().toBe(), assert ==, assertEquals()), parameterized syntax (it.each / @pytest.mark.parametrize / @ParameterizedTest), mock scaffolding (vi.mock, unittest.mock.patch, Mockito), and setup/teardown (beforeEach/afterEach, @pytest.fixture, @BeforeEach/@AfterEach). React components detected by capitalized names and JSX returns get snapshot tests.

Question 3

Are the generated tests always correct?

Accepted Answer

No — and the tool is honest about it. Each generated test is labeled 'deterministic' (signature-driven — e.g., a throw assertion that the function actually throws) or 'heuristic' (best-effort — e.g., an edge-case assertion that uses a placeholder like toBeDefined()). Happy-path expected values are inferred from the return type or function name; you must replace placeholders with the real expected value. Always run the suite and adjust expected values to match the actual behavior of your code.

Question 4

What extra features does this tool have?

Accepted Answer

(1) 4 languages (JS/TS/Python/Java). (2) 4 frameworks (Jest/Vitest/PyTest/JUnit). (3) Auto-detect source language. (4) Function-signature parser (params, types, async, exports, arrow, generators, throws). (5) Happy-path test per function. (6) Per-parameter edge-case tests. (7) Error-case tests (throw assertions + async rejection). (8) Parameterized/table-driven tests (it.each / parametrize / @ParameterizedTest). (9) Snapshot tests for React components. (10) Mock scaffolding for fetch and module imports. (11) Setup/teardown scaffolding per framework. (12) Coverage-gap hints per function. (13) Per-test 'explain what this covers' panel. (14) Sample snippets per language. (15) Stats (totals by category). (16) Local history (localStorage, last 20). (17) Shareable URL with code + framework + language. (18) Copy/download as .test file. (19) Optional BYO-key LLM explanation (OpenAI/Anthropic). (20) Honesty disclaimer about heuristic tests.

Question 5

Is my code sent anywhere?

Accepted Answer

No. All parsing, generation, mock scaffolding, and history run locally in your browser — your code never leaves this device. The only network path is if you explicitly paste your own LLM API key and click 'Explain with LLM' — that request goes directly to your chosen LLM provider (OpenAI or Anthropic) and never touches UnQTools servers. Skip the LLM step for 100% offline use.

AI Unit Test Case Generator

How to use