bug-killer

Reproduce, isolate, and verify defects across frontend, gateway, and backend services.

Updated Aug 6, 2025
One-click install
npx skills add https://github.com/fabiano001/claude-config --skill bug-killer-fabiano001
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: bug-killer
Source: https://github.com/fabiano001/claude-config/tree/main/skills/bug-killer
Command: npx skills add https://github.com/fabiano001/claude-config --skill bug-killer-fabiano001

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Reproduce, isolate, and fix defects efficiently with a repeatable diagnostic loop that coordinates frontend, backend, and service interactions, minimizing manual investigation.

Core Features & Use Cases

  • Deterministic observe-hypothesize-test loop for end-to-end debugging
  • Reliable Git branch handling including a CURRENT mode and safe transitions
  • Guided diagnostic cycle spanning frontend, API gateway, services, and data stores
  • Interactive planning and iterative fixes with progress tracking

Quick Start

Command the agent to start a bug-hunting session on the current workspace or on a Jira-named branch, following the workflow described above.

Frequently Asked Questions about bug-killer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I debug issues across frontend, API gateway, and backend services systematically?

End-to-end debugging across frontend, gateway, and backend services uses a deterministic observe, hypothesize, and test loop to reproduce, isolate, and verify defects. This coordinated diagnostic cycle minimizes manual investigation across multi-service changes.

What's the best way to manage Git branches while fixing bugs across multiple services?

Git branch management during debugging uses a CURRENT mode with safe transitions to maintain a clean state. It integrates branch handling directly into the diagnostic workflow, supporting Jira-named branches for automated planning and progress tracking.

Do I need Claude Code permissions to run automated diagnostic loops?

Yes, executing the diagnostic loop requires Claude Code permissions along with shell command and Git access. These permissions allow the workflow to run diagnostic steps, interact with data stores, and record plan artifacts automatically.

How does the observe hypothesize test loop work for software defect isolation?

The observe hypothesize test loop reproduces and isolates defects by observing symptoms, forming a hypothesis about the root cause, and testing fixes iteratively. It spans UI, API, and data store interactions for guided, verifiable resolution.

Can I use this debugging workflow for full stack changes touching the UI, API, and data stores?

Yes, this workflow specifically targets multi-service debugging scenarios where changes touch the UI, API, and data stores. It coordinates frontend and backend service interactions through interactive planning and iterative fixes.

Why should I use a deterministic debugging loop instead of manual investigation?

A deterministic debugging loop provides a repeatable diagnostic cycle that replaces ad hoc manual investigation. It ensures defects are reproduced, isolated, and verified through structured progress tracking and automated planning artifacts.