What is a test harness in this methodology?

A test harness is the specific infrastructure, such as a Docker container or fixture file, required to run tests against a real dependency in a repeatable way.

How does this skill handle failing tests during development?

It uses a 'Progress Test' system where tests under development live in a specific 'specs' directory and are allowed to fail, while the main 'tests' suite must always pass.

Why does this skill discourage mocking?

Mocking is viewed as a sign of poor design or testing at the wrong level. This skill promotes using real implementations or dependency injection to ensure tests reflect real-world behavior.

Does this skill work for languages other than Python?

Yes, while many examples are in Python, the foundational logic of the Confidence Pyramid and the no-mocking rule applies to any software development project.

What are the three levels of the Confidence Pyramid?

The pyramid consists of Level 1 (Unit/Pure logic), Level 2 (Integration with real infrastructure), and Level 3 (System/E2E workflows in real environments).

No-Mock Test Strategy

Name: No-Mock Test Strategy
Author: simonheimlicher

bysimonheimlicher

•

Security & Testing

Implements a high-confidence testing methodology focused on real-world behavior and the elimination of brittle mocks.

This skill establishes a foundational test strategy based on Behavior-Driven Development (BDD) and a three-tiered 'Confidence Pyramid.' It guides Claude to prioritize real-world reliability over arbitrary coverage metrics by enforcing a strict 'no-mocking' policy. By advocating for dependency injection at the unit level and rigorous test harnesses at the integration level, this skill ensures that software actually works in production. It also introduces a disciplined workflow separating 'progress tests' (in-development) from 'regression tests' (production-ready), providing a structured roadmap for building resilient, well-architected systems.

Key Features

011 GitHub stars

02Three-level Confidence Pyramid for Unit, Integration, and System testing

03Isolated Level 1 testing using OS-provided temporary directories

04Clear separation between failing Progress tests and passing Regression tests

05Strict 'No-Mocking' enforcement via dependency injection patterns

06Formalized 'Test Harness' identification for external dependencies

Use Cases

01Establishing a robust testing architecture for new Python or CLI projects

02Building integration tests for real-world dependencies like databases and binaries

03Refactoring legacy code to replace fragile mocks with testable interfaces

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add simonheimlicher/spx-claude testing

For use in Claude.ai and ChatGPT

Download Skill