How does it improve Pull Request quality?

It forces a line-by-line requirement check and a full build/test cycle, ensuring that only fully verified and passing code reaches the review stage.

Can it handle regression testing validation?

Yes, it mandates a red-green-restore cycle to prove that a test fails without the fix and passes with it, ensuring the test is valid.

Why is this skill necessary for AI coding?

It prevents 'hallucinated' success claims by requiring AI agents to provide empirical evidence, such as test logs or build status, before reporting a task as complete.

What is the 'Iron Law' of this skill?

The Iron Law dictates that no completion claims can be made without fresh verification evidence; past runs or assumptions are strictly prohibited.

What happens if an agent tries to skip verification?

The skill identifies 'red flag' language like 'should work' or 'seems to' and stops the workflow until actual verification is performed.

Verification Before Completion

Name: Verification Before Completion
Author: Krosebrook

byKrosebrook

•

Security & Testing

Enforces a strict evidence-first protocol requiring empirical verification and fresh command output before any task is marked as complete.

About

This skill establishes a rigorous 'Iron Law' for Claude Code, mandating that no status claims or success assertions are made without fresh, verifiable evidence. It implements a structured gate function—Identify, Run, Read, and Verify—to prevent false positives, 'hallucinated' successes, and broken builds. By forcing the agent to run tests, check exit codes, and verify requirements line-by-line before committing code or creating pull requests, it builds high-trust development workflows and eliminates the risks of 'should work' logic and unverified assumptions.

Key Features

Rationalization prevention for terms like 'should' or 'probably'
1 GitHub stars
Mandatory 'Gate Function' workflow for all status claims
Independent verification of agent-delegated tasks and VCS diffs
Evidence-based verification for tests, builds, and linting
TDD-style red-green cycle validation for regression tests

Use Cases

Validating bug fixes via empirical red-green cycles to prove resolution
Ensuring all unit and integration tests pass before pushing to a repository
Confirming full build integrity and requirement satisfaction before PR creation

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add krosebrook/source-of-truth-monorepo verification-before-completion

For use in Claude.ai and ChatGPT

Download Skill

GitHub