verification-discipline

Verify completed work with observable evidence beyond unit tests.

10|11|Updated Feb 22, 2026
One-click install
npx skills add https://github.com/microsoft/amplifier-bundle-skills --skill verification-discipline-microsoft
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: verification-discipline
Source: https://github.com/microsoft/amplifier-bundle-skills/tree/main/skills/verification-discipline
Command: npx skills add https://github.com/microsoft/amplifier-bundle-skills --skill verification-discipline-microsoft

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you avoid declaring work finished too early by checking whether the system actually works, not just whether the code looks correct. It is meant for verification moments where passing unit tests is not enough and you need real evidence that the intended outcome was achieved.

Core Features & Use Cases

  • Verification mindset: Distinguishes unit test success from real integration, smoke, and end-to-end evidence.
  • Completion checklist: Guides you through what “done” means, including observable proof and repo-specific gates.
  • Outcome-first testing: Encourages writing tests from the user-visible result backward so the right behavior is validated.
  • Use case: Before closing a task, use this Skill to confirm that the change works in a fresh environment and that you have logs, screenshots, or other proof to support the claim.

Quick Start

Use the verification-discipline skill to check whether this change is truly done and what evidence still needs to be collected.

Frequently Asked Questions about verification-discipline

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

FAQPage Schema
How do I verify my code works beyond passing unit tests?

To verify code beyond unit tests, you must collect real integration, smoke, or end-to-end evidence. This means confirming the system functions in a fresh environment and documenting observable proof like logs or screenshots before declaring the task done.

What evidence do I need for a pre-merge verification check?

Pre-merge verification requires observable proof from unit, integration, smoke, or production-equivalent testing. You must document these results to confirm real system behavior rather than just checking that the code looks correct.

What is the best way to define completion criteria for a software task?

Defining completion criteria involves distinguishing unit test success from real integration evidence. You should use an outcome-first approach, writing tests from the user-visible result backward to validate the right behavior and document observable proof.

Why does passing unit tests not guarantee my feature is done?

Passing unit tests does not guarantee completion because they only validate code correctness, not real system behavior. You need integration and smoke test evidence to prove the intended outcome works in a production-equivalent environment.

Can I use outcome-first testing for post-implementation review?

Yes, outcome-first testing fits post-implementation review by validating user-visible results backward. It guides you to confirm changes work in a fresh environment and require logs or screenshots as observable proof before closing the task.