Does it support Test-Driven Development?

Yes, it specifically includes patterns for Red-Green cycles, ensuring a regression test fails before the fix is applied and passes afterward to prove validity.

How does this prevent AI 'hallucinations' in code completion?

It forces the AI to execute real commands and read actual terminal output (exit codes, failure counts) rather than predicting that the code 'should' work based on its changes.

What is the core principle of this skill?

The core principle is 'Evidence before assertions.' It prohibits claiming work is complete without running fresh verification commands and confirming the output.

When is this skill most useful?

It is essential before committing code, creating pull requests, or moving to a new task to ensure the current work meets all requirements and passes all tests.

Rigorous Work Verification

Name: Rigorous Work Verification
Author: zhengxuyu

byzhengxuyu

0•

Security & Testing

Enforces a strict evidence-first protocol requiring fresh verification output before any task is marked as complete or successful.

This skill implements a robust 'Gate Function' for Claude Code, designed to eliminate premature or false claims of success. It mandates the execution of fresh verification commands—such as unit tests, build scripts, and linters—and requires a full analysis of the actual output before any completion assertions, commits, or pull requests are initiated. By prioritizing empirical evidence over assumptions or 'confidence,' this skill builds a high-trust development workflow and prevents common errors like shipping broken builds or missing requirements.

Key Features

01Strict identification of 'rationalization language' like 'should' or 'probably'

02Mandatory Gate Function requiring evidence before any success claims

03Independent verification of delegated agent tasks through VCS diffs

040 GitHub stars

05Red-green cycle verification for robust regression testing

06Systematic verification patterns for builds, tests, and linter checks

Use Cases

01Ensuring Test-Driven Development (TDD) compliance through verified failure and success

02Preventing broken or untested code from being committed to repositories

03Standardizing the 'Definition of Done' for AI-assisted development tasks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add zhengxuyu/claude-superpowers verification-before-completion

For use in Claude.ai and ChatGPT

Key Features

01Strict identification of 'rationalization language' like 'should' or 'probably'

02Mandatory Gate Function requiring evidence before any success claims

03Independent verification of delegated agent tasks through VCS diffs

040 GitHub stars

05Red-green cycle verification for robust regression testing

06Systematic verification patterns for builds, tests, and linter checks

Use Cases

01Ensuring Test-Driven Development (TDD) compliance through verified failure and success

02Preventing broken or untested code from being committed to repositories

03Standardizing the 'Definition of Done' for AI-assisted development tasks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add zhengxuyu/claude-superpowers verification-before-completion

For use in Claude.ai and ChatGPT