What are the 'red flag' words to avoid?

Phrases like 'probably,' 'seems to,' 'should work,' or expressing satisfaction ('Great!', 'Done!') before running verification are flagged as process violations.

Can this skill be used for build and lint checks?

Yes, it requires executing the full build or lint command and verifying an 'exit 0' status before allowing any claim that the code is clean or buildable.

Why is 'should pass' not considered sufficient?

Qualitative claims lack empirical evidence. This skill mandates running fresh commands to prove functionality through raw logs and exit codes rather than assumptions.

How does this skill handle regression testing?

It follows a strict red-green cycle: verifying a test fails when the fix is reverted and passes when the fix is applied, ensuring the test actually covers the bug.

What happens if an agent reports success independently?

The skill prevents trusting agent reports at face value, requiring a manual check of the VCS diff and a fresh run of verification commands to confirm the state.

PR Readiness & Verification

Name: PR Readiness & Verification
Author: dsifry

bydsifry

•

安全与测试

Enforces rigorous, evidence-based verification protocols to ensure code is fully tested and functional before making completion claims or creating pull requests.

This skill implements a deterministic 'trust but verify' framework for AI coding agents, ensuring that no task is marked as finished without empirical proof. It mandates a systematic approach—identifying necessary validation commands, executing fresh tests, and analyzing raw output—to prevent 'hallucinated' success. By removing qualitative assumptions and replacing them with deterministic evidence like exit codes and test pass counts, it guarantees high-quality contributions and prevents broken code from entering the version control history.

主要功能

01Detection of non-committal language (should, probably, seems)

02Automated red-green regression testing patterns

03Deterministic build and linter status validation

0429 GitHub stars

05Mandatory gate function for all completion claims

06Independent verification of delegated agent outputs

使用场景

01Ensuring zero-error linter and build status before PR creation

02Validating complex bug fixes through strict red-green TDD cycles

03Verifying all functional requirements against a checklist before task handoff

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add dsifry/goodtogo verification-before-completion

For use in Claude.ai and ChatGPT

Download Skill