AI Prompt for Mocking Dependencies

Create clean test mocks with this AI prompt: fakes, stubs and spies for your dependencies, with verification that your code used them correctly.

The prompt

Role: You are a testing expert who creates mocks that are faithful, minimal, and maintainable.

Context: Language: [LANGUAGE], framework: [FRAMEWORK]. Code under test: [CODE_UNDER_TEST]. Dependencies to mock: [DEPENDENCIES with what they do]. Mock style: [STYLE: manual mocks / framework mocks / fake implementations].

Task: 1) For each dependency, create the mock/fake: only the methods the code under test actually uses, with realistic return values. 2) Show how to inject it (constructor, parameter, DI container — match the codebase). 3) Write the test demonstrating the mock in use, INCLUDING verification that the code called the dependency correctly (arguments, call count). 4) Warn about over-mocking: what should stay real in this test.

Format: Mocks first, then injection, then the test with verifications, then the over-mocking warning.

Constraints: Mock at the boundary (the dependency's public interface), not internals. Mocks must reflect the real dependency's contract — note any assumption you made about it. Do not mock the thing you are testing.

Variables to fill in

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

[LANGUAGE] [FRAMEWORK] [CODE_UNDER_TEST] [DEPENDENCIES] [STYLE]

Example input

LANGUAGE=TypeScript, DEPENDENCY=Stripe API client + Postgres via Prisma, TEST=checkout function, FRAMEWORK=Vitest, MOCK_STYLE=manual mocks

How to use this prompt

List exactly what the dependency does in your test (returns what, throws when) — mocks must mimic the contract. State your mock style; framework idioms differ.

Tips for the best results

Verify mock interactions (called with what, how often) — that is where mock tests earn their keep.
Ask it to list the contract assumptions so to check them against the real dependency.
Prefer fakes over mocks for stable interfaces; they survive refactoring better.