debug

Coordinate a four-phase debugging workflow for unclear failures.

1|Updated Feb 25, 2026

Search Tags:#workflow #testing #debugging #investigation #agents #root-cause

Ready to run

npx skills add https://github.com/danalexilewis/taskgraph --skill debug-danalexilewis

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: debug
Source: https://github.com/danalexilewis/taskgraph/tree/main/.cursor/skills/debug
Command: npx skills add https://github.com/danalexilewis/taskgraph --skill debug-danalexilewis

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Helps orchestrate a disciplined debugging workflow when root cause is unclear, tests fail unexpectedly, or a user triggers /debug.

Core Features & Use Cases

Four-phase debugger that guides Root Cause Investigation, Pattern Analysis, Hypothesis and Testing, and Implementation.
Roles-based workflow with optional sub-agents (investigator, implementer) to split phases.
Escalation and guardrails after 3 failed attempts to avoid endless debugging.

Quick Start

Run the debug skill to start a structured root-cause investigation for unclear failures.

Frequently Asked Questions about debug

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

FAQPage Schema

How do I debug a failing test when the root cause is unclear?▼

To debug an enigmatic bug with an unclear root cause, you can trigger a structured debugging workflow that coordinates root cause investigation, pattern analysis, hypothesis testing, and implementation. This systematic process enforces discipline to identify the actual failure.

What is a systematic debugging workflow for software engineering agents?▼

A systematic debugging workflow is a four-phase process—Root Cause Investigation, Pattern Analysis, Hypothesis and Testing, and Implementation—that guides agents through unclear failures. It uses optional sub-agents to split investigation and implementation roles.

How do I stop agents from endlessly debugging the same failed fix?▼

To stop endless debugging loops, the workflow includes built-in guardrails that trigger an escalation after three failed fix attempts. This prevents agents from repeatedly applying ineffective patches to the same root cause.

Can I split investigation and implementation roles during root-cause analysis?▼

Yes, you can split roles during root-cause analysis by using optional sub-agents. The workflow supports assigning an investigator agent for pattern analysis and an implementer agent for the actual fix.

When should I use a structured root-cause investigation instead of ad-hoc debugging?▼

You should use a structured root-cause investigation when tests fail unexpectedly, you face enigmatic bugs, or you trigger a /debug request. It enforces a disciplined workflow needed when the underlying failure is not immediately obvious.