Products /Test Case Generator

// education lab

Test Case Generator

This is an online, interactive test case generator for Python functions at education scale.
Write a small Python function — the test suite that reaches 100% coverage of it is generated live, as you type.
One suite for each of the four classic criteria: statement, branch, condition, and path coverage.

// how to use it

Three things to do.

Write code and annotate the function under test.

Write Python in the editor and put @software-under-test above exactly one function — it gets the graph and the suite, and helpers or constants beside it are analyzed too. The language is a compact Python subset — if/elif/else, while, for, helper calls, over ints, floats, bools, strings, lists and tuples; anything outside it (classes, dicts, imports, …) is refused with a line-anchored reason.

Pick a coverage criterion.

The dropdown in the TEST SUITE panel switches what the suite must achieve:

–

statement every executable line runs at least once.

–

branch every decision takes both its true and its false exit.

–

condition every atomic condition inside a decision evaluates to both true and false.

–

path every entry-to-exit route through the graph is executed (loops bounded to 2 iterations).

Explore the generated test cases.

Every case is one concrete call — inputs plus the returned value — and hovering it lights up its exact path through the control-flow graph (click to pin). The generation algorithm is a simple search: boundary values taken from your code's own constants plus a light random sweep, greedily trimmed to a minimal suite. If 100% is not achieved by our tool, you get the real score and the targets it couldn't reach.

// the lab

● loading…

This playground is interactive and needs JavaScript — the rest of the page explains the concepts without it.

01 · software under test Load an example function

@software-under-test
def foo(a: int, b: int):
    if a > b:
        return True
    else:
        return False

02 · control flow graph ◯ statements · ◇ decisions

03 · test suite Coverage criterion

// action needed

AI coding agents need the right tools for testing.

Not one AI call was made on this page for test case generation — every suite came from a deterministic, reproducible tool, for zero tokens.
Much of software testing is exactly this systematic, and dumping it wholesale on a generative model is expensive token waste.
Hand your agent the right tool instead: novetest, the testing intelligence layer that lets AI coding agents test systematically — and cheaply. Start with the novetest docs.

Hand your agent novetest → deterministic where it can be · intelligent where it must be