What is the primary purpose of the Verification Before Completion skill?

It ensures that the AI never claims a task is finished or fixed without first running a fresh verification command and providing the actual output as evidence.

Does this skill work with automated testing frameworks?

Yes, it is designed to work alongside test suites, linters, and build tools, requiring their successful execution before the AI reports completion.

What does 'The Iron Law' mean in this context?

The Iron Law states that no completion claims can be made without fresh verification evidence. If a command hasn't been run in the current session, the claim cannot be made.

How does this skill improve the development workflow?

It reduces errors, prevents the merging of broken code, and builds trust between the AI and human developers by providing empirical proof for every change.

Verification Before Completion

Name: Verification Before Completion
Author: carlopezzuto

bycarlopezzuto

0•

安全与测试

Enforces rigorous verification protocols to ensure all coding tasks are validated with empirical evidence before marking them as complete.

The Verification Before Completion skill acts as a critical gatekeeping function for Claude Code, preventing premature or dishonest success claims. It mandates a strict 'Evidence Before Assertions' workflow, requiring the AI to identify specific verification commands, execute them in real-time, and analyze the fresh output before making any claims of task completion or bug fixes. By eliminating assumptions and 'should-pass' logic, this skill ensures that every commit, pull request, and status update is backed by verifiable logs and exit codes, maintaining high standards of reliability and developer trust.

主要功能

01Identifies and blocks common rationalization failures like 'partial checks' and 'should-be' logic

02Enforces the 'Iron Law' requiring fresh command evidence for all success claims

030 GitHub stars

04Prevents blind trust in agent delegation reports through independent VCS diff verification

05Mandates Red-Green TDD verification for regression testing and bug fixes

06Implements a systematic 5-step gate function: Identify, Run, Read, Verify, and Claim

使用场景

01Confirming that build and lint commands return successful exit codes before committing changes

02Validating that a bug is actually fixed by verifying the original symptom fails and then passes

03Ensuring all unit and integration tests pass with zero failures before creating a pull request

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add carlopezzuto/agents verification-before-completion

For use in Claude.ai and ChatGPT

Download Skill