atdd

Enforce acceptance test driven development by specifying Gherkin behavior before implementation.

134|9|Updated Feb 16, 2026
One-click install
npx skills add https://github.com/swingerman/disciplined-agentic-engineering --skill atdd-swingerman
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: atdd
Source: https://github.com/swingerman/disciplined-agentic-engineering/tree/main/skills/atdd
Command: npx skills add https://github.com/swingerman/disciplined-agentic-engineering --skill atdd-swingerman

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It prevents AI-assisted coding from drifting by forcing feature work to start with acceptance specifications written in domain language, then validating both external behavior and internal correctness through separate test streams.

Core Features & Use Cases

  • Acceptance-first specifications: Write feature behavior in standard Gherkin (Given/When/Then) before implementation to lock in what “done” means.
  • Dual-stream verification: Generate and run acceptance tests plus unit tests so the system’s external behavior and internal structure are both constrained.
  • Deterministic guardrails for iteration: Enforce spec integrity, regenerate pipelines from spec.md, and handle failures with traceability back to the scenario and source spec.

Quick Start

Tell Claude Code: “Use the ATDD workflow to implement this feature by writing spec.md in Gherkin, generating the acceptance test pipeline, making acceptance tests fail first, then implementing with TDD until both acceptance and unit test streams pass.”

Frequently Asked Questions about atdd

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I enforce acceptance test driven development with Gherkin specifications?

Acceptance test driven development is enforced by writing feature behavior in Gherkin before implementation. The system generates executable acceptance tests from a portable spec IR, binds step handlers to internals, and runs tests until both acceptance and unit streams pass.

Why does AI-assisted coding drift from feature requirements during implementation?

AI-assisted coding drifts when implementation lacks locked specifications. Starting with domain-language Gherkin specs prevents drift by constraining external behavior and internal correctness through dual-stream acceptance and unit test validation.

What's the best way to prevent spec drift when regenerating test pipelines?

Prevent spec drift by applying permission-gated edits to spec.md and enforcing regeneration discipline. The workflow regenerates executable pipelines from the original spec and traces failures back to the source scenario to maintain spec integrity.

Can I run acceptance tests and unit tests in parallel during feature development?

Dual-stream verification generates and runs acceptance tests alongside unit tests. This parallel validation ensures the system's external behavior and internal structure are both constrained throughout the test-driven iteration cycle.

Does acceptance test driven development require writing spec.md before any code?

Acceptance-first discipline requires writing spec.md in domain language before implementation. This spec-first planning locks in what done means, generates the test pipeline, and drives TDD iteration until both test streams pass.

How do I bind Gherkin step handlers to internal application logic?

Step handlers bind to real internals after generating executable acceptance tests from the portable spec IR. This connects Gherkin scenarios to actual application logic, enabling traceability from test failures back to the source spec.