How is this different from standard AI debugging?

While standard AI might suggest fixes based on patterns, this skill enforces a scientific methodology: it forms hypotheses, searches for specific evidence, and verifies findings before proposing any code changes.

Can it help with intermittent or 'flaky' bugs?

Yes. The skill includes specific patterns for investigating timing issues, race conditions, and state-dependent bugs by correlating logs and analyzing environment variables systematically.

What is the 'Observable Actions' principle?

This principle ensures the AI only reports what it has actually verified through its tools (like reading files or running commands), preventing hallucinations and ensuring every claim is backed by evidence.

Does this skill automatically modify my code?

No. It follows a 'User in Control' philosophy, where it proposes a targeted fix and explains the reasoning, only applying changes once you have explicitly approved the solution.

Scientific Bug Diagnosis

Name: Scientific Bug Diagnosis
Author: rsmdt

byrsmdt

•

158

•

개발자 도구

Applies a systematic, evidence-based debugging methodology to identify and resolve complex software issues through conversational investigation.

The Bug Diagnosis skill transforms Claude into a specialized debugging agent that follows the scientific method to isolate and fix software defects. Instead of jumping to conclusions, it uses a structured approach involving hypothesis formation, targeted experimentation, and progressive disclosure. By adhering to the 'Observable Actions' principle, it only reports verified findings, ensuring a high-trust, collaborative investigation where the user remains in control of the direction and final implementation of any fixes.

주요 기능

01Progressive disclosure of technical details to maintain focus

02Systematic root cause analysis and execution path tracing

03Evidence-based reporting using the Observable Actions principle

04User-controlled investigation flow with proactive check-ins

05Scientific hypothesis formation and testing

06158 GitHub stars

사용 사례

01Performing deep root cause analysis on complex logic errors or race conditions

02Investigating intermittent production bugs with limited initial context

03Systematically tracing execution paths across multiple files to isolate regressions

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add rsmdt/the-startup bug-diagnosis

For use in Claude.ai and ChatGPT

Download Skill