verification-before-completion

Require observed build, test, and run outputs before declaring engineering tasks complete.

Updated May 6, 2026
One-click install
npx skills add https://github.com/korwlik23/skills --skill verification-before-completion-korwlik23
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: verification-before-completion
Source: https://github.com/korwlik23/skills/tree/main/v1.1/skills/verification-before-completion
Command: npx skills add https://github.com/korwlik23/skills --skill verification-before-completion-korwlik23

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It prevents false completion reports by forcing you to verify that a change truly works in observable results, not just by reasoning or expectations.

Core Features & Use Cases

  • Verification Before Completion HARD GATE: requires observed evidence for any claim that a task is done, fixed, working, or complete.
  • Evidence-first completion reporting: mandates build/test/run outputs and specific re-check confirmation of the originally reported problem.
  • Unverified-but-transparent outcomes: allows stating what remains unverified and exactly what would be needed to verify it, instead of fabricating results.
  • Red Flags anti-rationalization: blocks common “should/probably” completion phrases unless backed by outputs.

Quick Start

Use this skill whenever you are about to say the work is done and include only observed build/test/run evidence for the reported resolution.

Frequently Asked Questions about verification-before-completion

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prevent false completion reports when fixing code and closing engineering tasks?

To prevent false completion reports, enforce verification before completion by requiring observed evidence like concrete test outputs, rather than relying on reasoning or expectations alone.

What is a verification gate for build outputs and UI behavior changes?

A verification gate for build outputs is a hard requirement that mandates concrete command and test results to prove a fix works before claiming the task is complete.

How do I report unverified limitations when I cannot run tests to confirm a fix?

To report unverified limitations, state exactly what remains unverified and specify the concrete build or test outputs needed to confirm the resolution, instead of fabricating results.

Why does my debugging fix still fail the quality gate during completion claims?

Your debugging fix fails the quality gate because completion claims using should or probably rationalizations are blocked unless explicitly backed by observed test outputs and specific re-checks.

Can I close a readiness statement without concrete command outputs for the reported problem?

You cannot close a readiness statement without concrete command outputs, because the verification gate requires explicit observed evidence and specific re-check confirmation of the originally reported problem.