How do AI agents execute these tests?

The agents parse your YAML scenarios and use specialized drivers to perform actions like clicking, typing, or waiting, then validate the resulting state against your expected outcomes.

What is outside-in testing?

Outside-in testing is a methodology that verifies application behavior from an external user's perspective, focusing on the interface and outputs rather than the internal code implementation.

Which platforms can I test with this skill?

This skill supports testing for Command Line Interfaces (CLI), Terminal User Interfaces (TUI), Web applications, and Electron desktop apps.

Will my tests break if I refactor my code?

No. Because these tests focus on the 'What' (behavior) rather than the 'How' (implementation), they remain valid as long as the user-facing interface stays the same.

Does this skill provide evidence of test results?

Yes, the underlying framework generates comprehensive reports including screenshots, terminal output captures, and logs for every step of the execution.

Outside-In Agentic Testing

Name: Outside-In Agentic Testing
Author: rysweet

byrysweet

•

Seguridad y Pruebas

Generates and executes behavior-driven test scenarios for CLI, TUI, Web, and Electron applications using AI agents.

The Outside-In Testing skill enables developers to create declarative, implementation-agnostic test scenarios that focus on user-facing behavior rather than internal code logic. By leveraging the gadugi-agentic-test framework, users can define YAML-based tests for diverse platforms including command-line interfaces, terminal UIs, web apps, and Electron desktop applications. AI agents interpret these scenarios to execute actions, observe outcomes, and provide detailed evidence reports, making it an ideal solution for smoke testing, refactoring safety, and cross-platform quality assurance.

Características Principales

01Implementation-agnostic testing that survives internal code refactors

02Cross-platform support for CLI, TUI, Web, and Electron applications

03AI-powered execution and observation for complex user interactions

04Declarative YAML-based test scenario definitions that serve as documentation

05Automatic evidence collection including screenshots, logs, and execution reports

0616 GitHub stars

Casos de Uso

01Ensuring refactoring safety by verifying that external behavior remains unchanged

02Creating smoke tests to validate critical user flows across different interfaces

03Writing readable acceptance criteria that can be automatically verified by AI agents

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add rysweet/amplihack outside-in-testing

For use in Claude.ai and ChatGPT

Download Skill