Why is this skill necessary for AI coding?

It prevents the AI from assuming code works just because it looks correct, mandating actual command execution and output analysis before reporting progress.

Does this work with non-testing tasks like builds or linting?

Yes, it applies to any status claim including build success, linter cleanliness, and requirement fulfillment through checklists.

What is the 'Gate Function' in this context?

The Gate Function is a mandatory sequence: Identify the proof command, Run it, Read the full output, Verify the results, and only then Claim success.

How does it handle confidence-based claims like 'should pass'?

The skill treats confidence without evidence as a 'Red Flag' and instructs the AI to stop and run the necessary verification commands instead.

Verified Work Completion

Name: Verified Work Completion
Author: bgrober

bybgrober

•

보안 및 테스팅

Enforces rigorous evidence-based verification before any work status is claimed, committed, or submitted for review.

This skill establishes an 'Iron Law' for AI coding tasks, mandating that no work be declared finished without fresh, verifiable evidence from test runs, builds, or linters. By implementing a structured five-step Gate Function—Identify, Run, Read, Verify, and Claim—it eliminates the risk of hallucinated success, premature PRs, and regressions. It is particularly valuable for developers who require high-integrity code cycles and want to ensure that every 'bug fix' or 'feature completion' is backed by actual execution data rather than mere confidence.

주요 기능

01Five-Step Gate Function: Structured workflow to identify, execute, and interpret proof commands.

02Requirement Checklist Validation: Mandates line-by-line verification against original project plans.

03The Iron Law: Mandatory fresh execution of verification commands before any success claim.

04Rationalization Prevention: Explicitly blocks phrases like 'should work' or 'seems correct' in favor of raw output.

05Red-Green TDD Support: Enforces regression testing by verifying failure before confirming the fix.

061 GitHub stars

사용 사례

01Eliminating false success reports during complex refactoring by requiring evidence of zero linter and compiler errors.

02Preventing premature Pull Requests by ensuring the build and test suite pass in the local environment first.

03Validating delegated agent tasks by cross-referencing VCS diffs with actual runtime results.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add bgrober/indie-stack verification-before-completion

For use in Claude.ai and ChatGPT

Download Skill