Does avad-debug help with testing?

Yes. Every fix requires a mandatory regression test that is proven to fail without the fix and pass with it, ensuring the bug does not return in future builds.

What happens if a bug cannot be identified quickly?

The skill follows a 3-strike rule. If three consecutive hypotheses fail to explain the bug, it stops to suggest an architectural review or escalation to a human developer rather than continuing to guess.

What is the 'Iron Law' of avad-debug?

The Iron Law states: 'No fixes without root cause investigation first.' It prevents the application of superficial fixes that only address symptoms while leaving the underlying problem intact.

How does scope locking prevent side effects?

Scope locking identifies the narrowest directory containing the affected files and restricts Claude's editing capabilities to that area, preventing accidental changes to unrelated parts of the codebase.

Can this skill search for external solutions?

Yes, it uses sanitized WebSearch to find known issues in frameworks and libraries, stripping away sensitive project data before searching to maintain security.

Systematic Debugging Pro

Name: Systematic Debugging Pro
Author: agwacom

byagwacom

0•

安全与测试

Employs a rigorous four-phase investigation methodology to identify root causes and implement verified fixes without regressions.

avad-debug is a specialized Claude Code skill designed to eliminate 'whack-a-mole' debugging by enforcing a strict 'Iron Law': no fixes are permitted without a verified root cause. It guides the AI through a structured workflow—investigation, pattern analysis, hypothesis testing, and implementation—ensuring that every bug is deeply understood before any code is changed. With built-in scope locking to prevent side effects and mandatory regression testing, it transforms haphazard troubleshooting into a disciplined, predictable engineering process.

主要功能

01Automated generation of structured Debug Reports including evidence and regression test results.

02Pattern analysis for common failures like race conditions, null propagation, and state corruption.

03Three-strike rule that triggers architectural review if multiple hypotheses fail.

04Scope locking integration to prevent unintended 'while-I'm-at-it' edits to unrelated code.

050 GitHub stars

06Four-phase debugging framework: Investigate, Analyze, Hypothesize, and Implement.

使用场景

01Performing deep root cause analysis on legacy code regressions where the origin is unclear.

02Enforcing high-quality bug fixes that include mandatory regression tests and minimal code churn.

03Troubleshooting intermittent race conditions or complex 'heisenbugs' that are difficult to reproduce.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add agwacom/avadbot avad-debug

For use in Claude.ai and ChatGPT