AI Prompt for Edge Case Tests

Uncover hidden edge cases with this AI prompt: adversarial inputs grouped by category, expected behavior for each, and ready-to-run test code.

The prompt

Role: You are an adversarial tester whose job is to break functions with inputs the developer never imagined.

Context: Language: [LANGUAGE], framework: [FRAMEWORK]. Function: [FUNCTION_CODE]. Its contract: [CONTRACT]. Already-known tricky inputs: [KNOWN_TRICKY].

Task: 1) Brainstorm edge cases by category: empty/null/undefined, boundaries (0, -1, max), type confusion, unicode/emoji, huge inputs, time/date quirks, locale differences, concurrent calls. 2) For each case: the input, what SHOULD happen per [CONTRACT] (or "undefined — you decide"), and why it is tricky. 3) Write the test code for the 10 most valuable cases. 4) Flag any case where the contract itself is ambiguous — those need a product decision, not just a test.

Format: Cases grouped by category, each with input | expected | why-tricky, then the test code, then the ambiguous-contract flags.

Constraints: Prioritize cases likely in REAL usage over exotic ones. If the contract is silent on a case, say so instead of inventing the expected behavior.

Variables to fill in

Replace each [variable] in the prompt with your own details:

[LANGUAGE] [FRAMEWORK] [FUNCTION_CODE] [CONTRACT] [KNOWN_TRICKY]

Example input

FUNCTION=parseDate(input) accepting multiple formats, LANGUAGE=JavaScript, FRAMEWORK=Jest, KNOWN_TRICKY=Feb 29, timezone offsets, 'tomorrow'

How to use this prompt

Name the tricky inputs you already know — the AI builds on them instead of repeating them. Specify which edge cases should be handled vs rejected.

Tips for the best results

Run the generated tests before fixing anything — failing edge tests are a roadmap.
For ambiguous cases, ask the AI to draft the product question for your team.
Re-run this prompt after every major refactor; new code has new edges.