What is the main purpose of this skill?

It prevents the AI from claiming a task is finished until it has executed and verified specific commands that prove success with actual output evidence.

When is this skill triggered during the development cycle?

It applies before any success claim, satisfaction expression, code commit, PR creation, or task delegation.

Does this work with regression testing?

Yes, it specifically enforces the Red-Green cycle, ensuring a test fails before the fix and passes after, proving the fix actually addressed the issue.

How does this improve AI-human collaboration?

It builds trust by ensuring that when the AI says a task is 'done,' it is backed by verifiable logs and command results rather than a guess.

Why does this skill ban phrases like 'it should work'?

Using 'should' or 'probably' is a red flag for unverified assumptions; this skill requires hard evidence (like exit code 0) before any success claim can be made.

Verification Before Completion

Name: Verification Before Completion
Author: seanGSISG

byseanGSISG

0•

Seguridad y Pruebas

Enforces a strict evidence-based workflow by requiring verified command output before any task is marked as complete.

The Verification Before Completion skill implements a rigorous 'Iron Law' for Claude Code, preventing the AI from making premature or unverified claims about work status. By acting as a mandatory gate function, it ensures that every assertion—whether it's a bug fix, a build success, or a passed test—is backed by fresh, complete execution evidence rather than assumptions or previous runs. This skill is essential for maintaining high code quality and developer trust, eliminating 'hallucinated' successes and ensuring that code only moves forward once its functionality is proven via terminal output.

Características Principales

01Enforces the 'Iron Law' requiring fresh verification evidence for all success claims

02Implements a structured Gate Function process (Identify, Run, Read, Verify)

03Identifies 'Red Flag' language like 'should' or 'seems to' to prevent rationalization

04Requires independent verification of delegated agent tasks before reporting completion

050 GitHub stars

06Standardizes patterns for TDD Red-Green cycles and regression testing

Casos de Uso

01Verifying complex project requirements against a line-by-line checklist with evidence

02Ensuring full test suites and linters pass with zero failures before opening a PR

03Validating bug fixes through mandatory Red-Green test cycles before committing

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add seangsisg/crispy-claude verification-before-completion

For use in Claude.ai and ChatGPT

Download Skill