Why is this skill necessary for AI development?

It prevents 'hallucinated' success by requiring the AI to prove code works with actual command output before claiming completion, preserving developer trust.

Does it work with any test runner?

Yes, it is designed to run any verification command including Jest, Pytest, linters, and custom build scripts, checking for explicit exit codes and failure counts.

How does it handle 'should work' or 'probably fixed' claims?

It explicitly flags and blocks phrases like 'should pass' or 'seems correct,' forcing a fresh execution of the verification suite instead.

Is this useful for Test-Driven Development (TDD)?

Absolutely. It enforces a strict Red-Green pattern, ensuring regression tests are verified to fail before they are verified to pass with the fix.

Can it verify tasks delegated to other agents?

Yes, it requires the AI to check the Version Control System (VCS) diffs and actual file states instead of blindly trusting success reports from delegated agents.

Verification Before Completion

Name: Verification Before Completion
Author: Kingly-Agency

byKingly-Agency

0•

安全与测试

Enforces a rigorous verification-first workflow by requiring fresh command evidence before any completion or success claims are made.

This skill acts as a critical integrity gatekeeper for Claude Code, preventing premature or false claims of task completion. It mandates a 'Proof Before Assertions' approach, requiring the AI to identify, execute, and analyze relevant verification commands—such as unit tests, linters, or build scripts—before committing code or signaling success. By eliminating guesswork and 'confidence-based' reporting, it ensures high-integrity development, maintains trust between human and AI, and prevents broken code from entering the codebase.

主要功能

01Comprehensive exit code and failure count analysis for builds and linters

02Mandatory verification command execution before any status claims

03Standardized Red-Green-Refactor cycle validation for regression tests

04Automated rationalization prevention for 'should work' assumptions

05Independent verification of VCS diffs against agent success reports

060 GitHub stars

使用场景

01Confirming bug fixes with verified regression tests before closing tickets

02Ensuring all unit tests pass before creating a pull request or committing code

03Verifying build success and linter cleanliness after complex refactoring

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add kingly-agency/kingly-claude-adapter verification-before-completion

For use in Claude.ai and ChatGPT

Download Skill