How does the Gate Function work?

The Gate Function requires a five-step process: Identify the proof command, Run the full command, Read the complete output, Verify the output confirms the claim, and only then make the success claim.

What is the 'Iron Law' of this skill?

The Iron Law states that no completion claims can be made without fresh verification evidence; if you haven't run the specific verification command in the current context, you cannot claim the task passes.

Can I use previous test runs as evidence?

No. This skill specifically identifies 'previous runs' as insufficient. It requires fresh, current command output to ensure no regressions occurred during the final stages of work.

How does it handle delegated agent reports?

It prevents blind trust in sub-agents. The skill requires checking the VCS diff and independently verifying changes rather than simply accepting a 'success' report from another agent.

Verification Before Completion

Name: Verification Before Completion
Author: noeljackson

bynoeljackson

0•

Security & Testing

Enforces a strict evidence-based workflow requiring command execution and output verification before any task is claimed as complete.

The Verification Before Completion skill establishes a rigorous 'Iron Law' for AI-driven development, mandating that no success claims be made without fresh, verifiable evidence. It implements a structured gate function that requires identifying proof, executing commands, and reading full outputs before committing code or creating pull requests. By systematically eliminating assumptions and 'should work' statements, this skill ensures higher code quality, prevents regressions, and builds a foundation of honesty and reliability in agentic workflows.

Key Features

01Implements a multi-step gate function for systematic verification

020 GitHub stars

03Mandates fresh command execution before asserting task completion

04Identifies common failure patterns and rationalization excuses

05Provides specific verification patterns for tests, builds, and requirements

06Ensures independent verification of delegated agent tasks

Use Cases

01Verifying complex feature requirements against a line-by-line checklist

02Ensuring full build and linter compliance prior to opening a pull request

03Validating bug fixes through red-green regression testing cycles before submission

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add noeljackson/agentic verification-before-completion

For use in Claude.ai and ChatGPT

Key Features

01Implements a multi-step gate function for systematic verification

020 GitHub stars

03Mandates fresh command execution before asserting task completion

04Identifies common failure patterns and rationalization excuses

05Provides specific verification patterns for tests, builds, and requirements

06Ensures independent verification of delegated agent tasks

Use Cases

01Verifying complex feature requirements against a line-by-line checklist

02Ensuring full build and linter compliance prior to opening a pull request

03Validating bug fixes through red-green regression testing cycles before submission

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add noeljackson/agentic verification-before-completion

For use in Claude.ai and ChatGPT