WizardHeHeJun

WizardHeHeJun/AutoPlayQA

Command Claude Code

A command for watching a background agent run, checking its progress, handing work to another agent, continuing until completion, and combining findings from multiple parts.

WizardHeHeJun/AutoPlayQA

Command Claude Code

A smoke-test reporting command that sorts replay findings into defects and produces reports in a fixed HTML and Markdown format. A smoke test is a quick check that the main parts of a system still work.

WizardHeHeJun/AutoPlayQA

Skill Claude CodeCodex

A skill for writing task-flow JSON as a recognition-and-action state machine, with checks that confirm each step and help recover from unexpected screens.

WizardHeHeJun/AutoPlayQA

Skill Claude CodeCodex

A process for supervising long unattended tasks that include handoffs between agents. It starts a run, checks progress, completes required manual steps, archives them, and continues until the work finishes.

WizardHeHeJun/AutoPlayQA

Skill Claude CodeCodex

A skill for safely editing existing task definitions stored as JSON.

WizardHeHeJun/AutoPlayQA

Skill Claude CodeCodex

A workflow that records a person or agent operating a device, including the interface elements they use, and turns the recording into a draft task.

WizardHeHeJun/AutoPlayQA

Skill Claude CodeCodex

A way to record a workflow while a person manually performs it on a real device. The agent watches each step through connected tools and creates a task JSON file based on what it recognises.

WizardHeHeJun/AutoPlayQA

Skill Claude CodeCodex

A project-level code checker for AutoPlayQA that finds problems ordinary tools such as ruff and mypy may not detect.

WizardHeHeJun/AutoPlayQA

Skill Claude CodeCodex

A reporting skill for sorting findings from repeated automated playback runs into product defects and producing HTML and Markdown reports.