How does this skill prevent false completion claims?

It mandates the 'Iron Law,' which forbids using cached or expected results. The AI must execute a fresh command and analyze the full output to provide evidence of success.

What is the Five-Step Gate?

It is a structured protocol consisting of: Identifying the proof command, Running it freshly, Reading the complete output, Verifying against expectations, and only then making a Claim.

Which programming languages are supported?

The skill includes built-in verification patterns for Node.js, Python, Ruby, Go, and Rust, but can be adapted for any language that provides command-line verification tools.

What is the Verification Before Completion skill?

It is a specialized capability for Claude Code that enforces a 'proof-first' discipline, requiring the AI to run and verify commands before claiming any software task is finished.

Strict Verification Gate

Name: Strict Verification Gate
Author: bostonaholic

bybostonaholic

•

セキュリティとテスト

Enforces a rigorous evidence-before-claims discipline by requiring fresh verification evidence before any task is marked as complete.

The Verification Before Completion skill implements a critical safety gate within the Research-Plan-Implement (RPI) framework, ensuring that AI-driven development remains disciplined and reliable. By mandating a strict five-step process—Identify, Run, Read, Verify, and Claim—it prevents 'hallucinated' success claims and ensures that all code changes are backed by fresh, observable evidence from tests, builds, or linters. This skill is essential for maintaining high trust in automated workflows by eliminating assumptions and enforcing the 'Iron Law' that no work is finished until the output confirms it.

主な機能

01Pre-commit and Pre-PR validation checklists

02Rationalization detection to prevent skipping tests

03Standardized five-step verification gate process

042 GitHub stars

05Cross-language support for Node.js, Python, Go, Rust, and Ruby

06Mandatory fresh verification evidence requirement

ユースケース

01Ensuring clean builds and linting status before committing code to a repository

02Confirming bug fixes using specific reproduction steps before reporting success

03Validating delegated agent work through independent verification commands

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add bostonaholic/rpikit verification-before-completion

For use in Claude.ai and ChatGPT

Download Skill