skills.rest
GongYuanCaiJi
GongYuanCaiJiCOMMUNITY
Explore all skills by this author →
GitHub

skill-debug

Diagnose bugs through a four-phase root cause investigation and hypothesis testing workflow.

Updated Aug 15, 2026
Search Tags:#debugging#troubleshooting#root cause analysis#test failures#hypothesis testing#systematic debugging#fix loops
One-click install
npx skills add https://github.com/GongYuanCaiJi/dsh-claude-octopus --skill skill-debug-gongyuancaiji
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skill-debug
Source: https://github.com/GongYuanCaiJi/dsh-claude-octopus/tree/main/.claude/skills/skill-debug
Command: npx skills add https://github.com/GongYuanCaiJi/dsh-claude-octopus --skill skill-debug-gongyuancaiji

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When facing bugs, test failures, or unexpected behavior, developers often jump straight to random fixes that waste hours and mask root causes. This Skill enforces a systematic four-phase debugging process that finds root causes in 15-30 minutes instead of 2-3 hours of thrashing. ## Core Features & Use Cases - Four-Phase Process: Root Cause Investigation, Pattern Analysis, Hypothesis Testing, and Implementation, with a hard gate preventing fixes before investigation completes. - 3-Strike Rule & Self-Regulation Score: Stops runaway fix loops by tracking reverts, unrelated file changes, and repeated errors, forcing a pause to consult the user after failed attempts. - Multi-Agent Debugging: Parallelizes investigation across multiple AI providers via the orchestrate.sh script, examining error messages, recent changes, data flow, and environment simultaneously. - Use Case: When a production API endpoint returns intermittent 500 errors, use this Skill to reproduce the failure, trace data flow across service boundaries, form a single testable hypothesis, and implement one verified fix. ## Quick Start Ask the AI to debug why your test suite is failing and have it follow the systematic four-phase process before proposing any fix.

Frequently Asked Questions about skill-debug

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

FAQPage Schema
How do I debug a bug systematically instead of guessing fixes?▼

Follow the four-phase process: investigate the root cause by reading errors and reproducing the issue, compare against working examples, form one specific hypothesis and test it minimally, then implement a single verified fix. Never propose fixes before completing root cause investigation.

What should I do after multiple failed fix attempts?▼

Apply the 3-strike rule: after three failed fixes, stop and show your work to the user before continuing. Repeated failures usually indicate an architectural problem, so question whether the underlying pattern is sound rather than attempting a fourth fix.

Can this debugging workflow handle multi-component systems?▼

Yes. For systems spanning API, service, and database layers, add diagnostic instrumentation at each boundary and run once to gather evidence showing where the failure occurs. Trace bad values up the call stack and fix at the source, not the symptom.

When should I not use a systematic debugging process?▼

Skip it for explanation questions like why a technology is used, decision support questions, known issues with clear solutions, and documentation or architecture questions. It is designed specifically for active bugs and failures.

What is the self-regulation score in debugging fix loops?▼

The WTF score quantitatively detects runaway fix loops by adding points for reverts, touching unrelated files, fixes spanning more than three files, and attempts beyond fifteen. If the score exceeds 20 percent, stop immediately and consult the user.

Related Skills

Explore More Software Engineering Skills→
legotec73legotec73

investigate

Diagnose software failures by tracing symptoms to root causes.

Community
Advanced
robkamrobkam
12

bugfix-red-green-proof

Enforce a red-green-proof loop starting with a failing regression test.

Community
Advanced
oikon48oikon48
1

parallel-explore

Launch four Explore agents in parallel to investigate distinct codebase areas.

Community
Basic
nakano1122nakano1122
2

observability

Guide observability design with structured logging, tracing, metrics, health checks, alerting, and error tracking.

Community
Intermediate