How do the parallel agents work?

It launches three agents simultaneously: an Execution Tracer to follow the code path, a Pattern Analyzer to find known bug patterns, and an Entry Point Mapper to check for unexpected triggers.

When should I use the Deep Debug skill?

Use it when you are 'going in circles' with a bug, dealing with complex browser-API interactions, or when symptoms don't match your code's logic despite multiple fix attempts.

How do I trigger a deep investigation?

You can use the explicit /deep-debug command followed by a bug description, or simply tell Claude 'I'm stuck on this bug' to invoke the workflow naturally.

Can it help with React-specific bugs?

Yes, it is specifically optimized to catch common React issues such as incorrect useCallback dependencies, stale closures, and useEffect hooks running multiple times.

Does Deep Debug require special tools?

While it works best with Chrome MCP tools for browser-based debugging, it is also highly effective at analyzing backend logs, stack traces, and database query results.

Deep Debug

Name: Deep Debug
Author: jezweb

byjezweb

•

245

•

セキュリティとテスト

Executes a multi-agent investigation to resolve complex software bugs through parallel evidence gathering and cross-referenced analysis.

Deep Debug is an advanced multi-agent workflow designed to tackle stubborn or recursive software bugs that resist standard debugging efforts. By launching three specialized parallel agents—an execution tracer, a pattern analyzer, and an entry point mapper—it systematically gathers evidence from Chrome MCP tools or backend logs to identify root causes like race conditions, stale closures, or duplicate API calls. This skill is particularly effective for React state management issues and complex browser-API interactions where symptoms don't match expectations, providing a high-confidence consensus for verification and ensuring that fixes are based on concrete data rather than guesswork.

主な機能

01Chrome MCP tool integration for real-time network and console log analysis

02Cross-functional debugging covering both frontend UI and backend logic

03Parallel multi-agent investigation using specialized debugger and reviewer roles

04Automated pattern detection for React hooks, stale closures, and race conditions

05Evidence-based consensus building to ensure high-confidence bug fixes

06245 GitHub stars

ユースケース

01Resolving complex state management bugs and asynchronous race conditions

02Debugging elusive React hook dependency issues and re-render loops

03Investigating duplicate or failing API calls via Chrome network tab analysis

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jezweb/claude-skills deep-debug

For use in Claude.ai and ChatGPT

Download Skill