Which agents are utilized during the investigation phase?

The skill dispatches the internet-researcher for external documentation and the codebase-investigator to understand the wider context of the failing code.

Can it interact with local debuggers like GDB or LLDB?

While it cannot execute debuggers directly, it generates specific commands for the user to run and provides detailed instructions on what state to inspect.

How does this skill differ from standard Claude debugging?

It enforces a strict 'tools-before-fixes' protocol, requiring evidence from internet research and codebase analysis before any code changes are proposed.

What happens if a hypothesis is proven wrong?

The workflow mandates a return to the investigation phase with the new information, preventing the implementation of incorrect or 'guess-based' fixes.

Does this skill help with performance problems?

Yes, its systematic approach is ideal for performance issues, helping to isolate bottlenecks through hypothesis testing rather than trial-and-error optimization.

Systematic Debugging

Name: Systematic Debugging
Author: withzombies

bywithzombies

•

Seguridad y Pruebas

Implements a disciplined, four-phase debugging workflow using specialized agents to identify and resolve root causes of technical issues.

The Systematic Debugging skill transforms Claude into a rigorous diagnostic engine that prioritizes evidence over guesswork. By enforcing a strict progression through investigation, hypothesis formation, and validation testing, it ensures that developers fix the underlying root cause rather than just treating symptoms. The skill leverages specialized agents like the internet-researcher to consult global knowledge bases and the codebase-investigator to map architectural dependencies, making it an essential tool for resolving cryptic errors, architectural flaws, and persistent regressions.

Características Principales

01Root cause verification to prevent 'whack-a-mole' symptom fixing

0258 GitHub stars

03Evidence-based four-phase workflow: Investigate, Hypothesize, Test, and Fix

04Structured state inspection through instrumentation and debugger guidance

05Automated deep-dive analysis of error logs and stack traces

06Multi-agent investigation via internet-researcher and codebase-investigator

Casos de Uso

01Debugging intermittent test failures in complex distributed systems

02Resolving cryptic build failures or architecture-specific linking errors

03Identifying systemic authorization flaws across multiple API endpoints

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add withzombies/hyperpowers debugging-with-tools

For use in Claude.ai and ChatGPT

Download Skill