debug

Run an interactive debugging loop to reproduce and resolve code failures.

Updated Aug 23, 2026

Search Tags:#code analysis #debugging #bug fixing #interactive #runtime evidence

Ready to run

npx skills add https://github.com/williamw/agent-skills --skill debug-williamw

💬 Prompt for ChatGPT / Claude Chat▼

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

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

This Skill provides an interactive debugging loop to help users identify and resolve hard bugs, crashes, and unexpected behaviors in their code.

Core Features & Use Cases

Fixed-State Machine Debugging: Follows a structured process to generate hypotheses, instrument code, and reproduce issues.
Runtime Evidence: Focuses on evidence over argument, using runtime evidence to confirm or refute hypotheses.
Reproducible Failures: Assists in creating reproducible scenarios for flaky bugs, ensuring consistent debugging.
Cleanup and Documentation: Ensures that all temporary files and logs are cleaned up, and the confirmed cause is documented for future reference.

Quick Start

Run the debug skill with the command 'debug' to start the interactive debugging loop.

Frequently Asked Questions about debug

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

FAQPage Schema

How do I debug hard crashes and unexpected code behaviors interactively?▼

Interactive debugging helps resolve hard crashes and unexpected behaviors by guiding you through a structured loop of generating hypotheses, instrumenting code, and reproducing failures to gather runtime evidence and confirm root causes.

What is the best way to reproduce flaky bugs consistently during code analysis?▼

The best way to reproduce flaky bugs consistently is to create reproducible scenarios that confirm or refute your hypotheses using runtime evidence, ensuring structured and reliable code analysis during your debugging process.

How do I start an interactive debugging loop for bug fixing?▼

You start bug fixing by running the debug command to initiate the interactive loop, which requires your active participation to reproduce failures, interpret logs, and validate hypotheses based on runtime evidence.

Does interactive debugging require me to interpret logs and reproduce failures myself?▼

Yes, interactive debugging requires you to actively participate by reproducing failures and interpreting logs yourself, ensuring that hypotheses are confirmed or refuted using actual runtime evidence rather than arguments.

What are the limitations of using a fixed-state machine for code debugging?▼

A limitation of fixed-state machine debugging is that it requires strict user participation for reproducing failures and interpreting logs, relying entirely on runtime evidence to confirm or refute hypotheses rather than automated fixes.

When do I need runtime evidence to fix a bug instead of just reading the code?▼

You need runtime evidence to fix a bug when dealing with hard crashes or unexpected behaviors that cannot be resolved by reading code alone, requiring you to instrument code and reproduce issues to confirm hypotheses.