How does it handle bug fix verification?

It mandates a red-green verification cycle where the original symptom is proven to fail before the fix and proven to pass after the fix is applied.

Can the agent skip these verification steps?

No, the skill explicitly defines 'spirit over letter' rules that identify common rationalizations and prevent the agent from bypassing mandatory verification steps.

When does this skill activate?

It triggers whenever the agent is about to express satisfaction, claim work is done, or perform a final action like a commit, push, or pull request.

Why is this skill necessary for AI coding?

It prevents AI agents from making false claims about code status, ensuring that every 'success' message is backed by fresh, verified command output rather than assumptions.

What counts as valid verification evidence?

Evidence must be fresh command output—such as linter results, test logs, or exit codes—generated in the current session, not historical or assumed data.

Verification Before Completion

Name: Verification Before Completion
Author: jett1011586853

byjett1011586853

•

セキュリティとテスト

Enforces rigorous evidence-based verification and command execution before any task completion claims are made.

The Verification Before Completion skill provides a strict framework to prevent Claude from prematurely claiming success or reporting false positives. By implementing the 'Iron Law' of evidence-based reporting, it requires the agent to run fresh verification commands, verify exit codes, and cross-reference results with project requirements before committing code or creating pull requests. This ensures higher reliability, maintains developer trust, and eliminates 'should work' hallucinations by mandating a disciplined red-green verification cycle for all bug fixes and features.

主な機能

01Line-by-line requirement checklist validation

02Detection and prevention of common success rationalizations

03Prevention of premature PR and commit creation

041 GitHub stars

05The Iron Law: Mandatory fresh verification evidence for all claims

06Red-Green regression testing for verified bug fixes

ユースケース

01Cross-checking complex feature requirements against a manual checklist before task handover

02Ensuring unit and integration tests pass before pushing code to a repository

03Verifying bug fixes through documented symptom-to-solution cycles

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jett1011586853/go-agent verification-before-completion

For use in Claude.ai and ChatGPT

主な機能

01Line-by-line requirement checklist validation

02Detection and prevention of common success rationalizations

03Prevention of premature PR and commit creation

041 GitHub stars

05The Iron Law: Mandatory fresh verification evidence for all claims

06Red-Green regression testing for verified bug fixes

ユースケース

01Cross-checking complex feature requirements against a manual checklist before task handover

02Ensuring unit and integration tests pass before pushing code to a repository

03Verifying bug fixes through documented symptom-to-solution cycles

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jett1011586853/go-agent verification-before-completion

For use in Claude.ai and ChatGPT