Why is this skill necessary for Claude Code?

It prevents 'hallucinated' success by requiring actual command output evidence before the AI can claim a task is finished, ensuring honesty and accuracy.

Does it work with existing CI/CD pipelines?

Yes, it acts as a local pre-flight check to ensure your code will pass remote CI/CD builds, saving time and reducing pipeline failures.

What happens if a verification step is skipped?

The skill treats skipping any step as a violation of its core logic, requiring the agent to go back and run the full command before proceeding.

Can I use this for Test-Driven Development (TDD)?

Absolutely. It specifically includes patterns for red-green-refactor cycles to ensure that tests are actually testing the intended behavior.

Verification Before Completion

Name: Verification Before Completion
Author: fzymgc-house

byfzymgc-house

0•

보안 및 테스팅

Enforces rigorous verification steps and evidence collection before any work is declared complete or committed.

This skill implements a strict 'Iron Law' for Claude Code, preventing the agent from making premature or unverified claims of success. It mandates a systematic gate function where specific verification commands must be executed and their outputs analyzed before any commits, pull requests, or status updates occur. By requiring fresh evidence for every claim—whether it's passing tests, clean linting, or successful builds—it eliminates 'hallucinated' success and ensures that every code change is backed by actual verification data, building trust and maintaining code integrity.

주요 기능

01Comprehensive validation checklist for tests, builds, and linter status

02Automatic identification of 'red flag' language like 'should' or 'probably'

030 GitHub stars

04Mandatory verification command execution before any success claims

05Independent verification of delegated agent tasks through VCS diff analysis

06Red-green cycle enforcement for reliable regression testing

사용 사례

01Validating that all requirements and linter rules are satisfied before opening a PR

02Ensuring regression tests actually fail before they pass to maintain TDD integrity

03Preventing broken builds or unverified code from being committed to a repository

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add fzymgc-house/fzymgc-house-skills verification-before-completion

For use in Claude.ai and ChatGPT

주요 기능

01Comprehensive validation checklist for tests, builds, and linter status

02Automatic identification of 'red flag' language like 'should' or 'probably'

030 GitHub stars

04Mandatory verification command execution before any success claims

05Independent verification of delegated agent tasks through VCS diff analysis

06Red-green cycle enforcement for reliable regression testing

사용 사례

01Validating that all requirements and linter rules are satisfied before opening a PR

02Ensuring regression tests actually fail before they pass to maintain TDD integrity

03Preventing broken builds or unverified code from being committed to a repository

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add fzymgc-house/fzymgc-house-skills verification-before-completion

For use in Claude.ai and ChatGPT