What types of evidence does it capture during testing?

It captures comprehensive data including stdout/stderr console output, platform-specific screenshots, log file contents, and execution duration.

Does it support subjective UI testing?

Yes. Tasks involving UI actions or qualitative descriptions are classified as SUBJECTIVE or GUI, which defaults to a mandatory human checkpoint for visual verification.

Can I skip the human approval step for simple tests?

No. To prevent production incidents, this skill enforces strict quality gates that require explicit human approval, ensuring evidence is reviewed even for 'obvious' passes.

What triggers the testing-end-user skill?

The skill is automatically invoked when Claude encounters markers such as 'TEST:', 'TEST:VERIFY', 'TEST:CONTRACT', or phrases like 'execute verification' in task descriptions.

How does it handle background processes like web servers?

It supports a '(background)' modifier for actions, tracking PIDs to ensure processes run asynchronously and are properly cleaned up after verification.

Human-in-Loop End-User Testing

Name: Human-in-Loop End-User Testing
Author: deepeshBodh

bydeepeshBodh

•

보안 및 테스팅

Executes structured Setup/Action/Assert verification sequences with mandatory human checkpoints to validate real infrastructure behavior.

This skill enables Claude Code to perform robust end-to-end verification by transforming testing markers into executable sequences. It automates environment setup, action execution, and assertion checking across CLI, GUI, and subjective tasks while maintaining strict quality gates. By enforcing a human-in-the-loop approval process and capturing detailed evidence—including logs, screenshots, and console output—it ensures that production-critical infrastructure and workflows are validated thoroughly before deployment, preventing regressions and silent failures.

주요 기능

01Multi-channel evidence capture including console logs, screenshots, and file states

02Structured Setup/Action/Assert execution for reproducible verification workflows

03Support for async background processes and customizable execution timeouts

04Mandatory human-in-the-loop checkpoints for final approval and audit trails

05Automated task classification (CLI, GUI, or Subjective) for tailored review

0619 GitHub stars

사용 사례

01Verifying contract compliance and CLI tool outputs against strict technical specifications

02Validating real-world infrastructure behavior after complex system deployments

03Running end-to-end GUI tests that require qualitative human judgment and visual review

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add deepeshbodh/human-in-loop testing-end-user

For use in Claude.ai and ChatGPT

Download Skill