When should this skill be triggered?

It should be used whenever Claude is about to commit code, create a pull request, or declare a feature or bug fix as 'done' to ensure no shortcuts are taken.

Why is evidence-based verification important for AI coding?

AI models can sometimes provide overconfident reports of success. This skill requires empirical proof (command outputs and exit codes) to ensure honesty, accuracy, and production-readiness.

Does this skill help with regression testing?

Yes, it specifically enforces Red-Green TDD patterns to verify that a fix actually addresses the symptom and hasn't introduced new regressions.

What is the Verification Before Completion skill?

It is a specialized guidance system for Claude Code that mandates running actual shell commands—such as tests, builds, and linters—before claiming any task is complete.

Which development tools does this skill support?

The skill includes specific patterns for popular ecosystems including Bun, NPM, and TypeScript, covering unit testing, linting, type-checking, and build processes.

Verification Before Completion

Name: Verification Before Completion
Author: secondsky

bysecondsky

•

セキュリティとテスト

Enforces rigorous evidence-based verification protocols to ensure code accuracy and quality before any task is finalized.

The Verification Before Completion skill acts as a critical quality gate for Claude Code, preventing the common pitfall of premature success claims. It mandates a 'command-first' approach where Claude must identify, execute, and interpret fresh verification evidence—such as test results, build logs, and linter outputs—before declaring a task complete or committing code. By eliminating 'hallucinated success' and overconfidence, this skill ensures that every bug fix, feature, and refactor is backed by empirical proof, maintaining high standards for production-ready code.

主な機能

01Systematic Red-Green-Refactor validation for regression testing and bug fixes

02Mandatory gate function requiring fresh command output before any completion claim

03Support for verifying agent-delegated tasks through VCS diffs and independent checks

04Identification of red-flag language to prevent optimistic or unverified assumptions

05Pre-defined verification patterns for Bun, NPM, and TypeScript environments

0621 GitHub stars

ユースケース

01Ensuring all unit and integration tests pass before creating a pull request

02Confirming bug fixes by executing specific failure-case reproduction scripts

03Validating full project builds and type-safety after making architectural changes

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add secondsky/claude-skills verification-before-completion

For use in Claude.ai and ChatGPT

Download Skill