diagnose

Diagnose hard bugs and performance regressions through a phase-driven reproduction-to-regression-test workflow.

2|1|Updated Mar 26, 2026
One-click install
npx skills add https://github.com/huangzida/skills --skill diagnose-huangzida
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: diagnose
Source: https://github.com/huangzida/skills/tree/main/diagnose
Command: npx skills add https://github.com/huangzida/skills --skill diagnose-huangzida

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Disciplined, phase-driven approach to diagnosing hard bugs and performance regressions, enabling rapid reproduction, hypothesis testing, instrumentation, and regression verification.

Core Features & Use Cases

  • Phase-driven workflow: Build a deterministic loop to reproduce and verify bugs.
  • Hierarchical phases: Reproduce → Hypothesise → Instrument → Fix → regression-test.
  • Guidance for non-deterministic bugs: increase reproduction rate with controlled sampling and instrumentation.

Quick Start

Run the diagnose workflow when you encounter a hard bug or performance regression, reproduce it, and let the steps guide you through diagnosis.

Frequently Asked Questions about diagnose

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

FAQPage Schema
How do I debug hard bugs that are difficult to reproduce?

To debug hard bugs that are difficult to reproduce, apply a structured debugging loop that uses controlled sampling and instrumentation guidance to increase reproduction rates and confirm root causes.

What is the best workflow for diagnosing performance regressions?

The best workflow for diagnosing performance regressions is a structured hierarchical loop spanning reproduce, hypothesise, instrument, fix, and regression-test to verify root causes efficiently.

How do I verify a bug fix and prevent future regressions?

To verify a bug fix and prevent future regressions, implement structured regression-testing as the final phase of a disciplined debugging loop to ensure the bug is closed efficiently.

Does structured hypothesis-driven testing work for non-deterministic bugs?

Hypothesis-driven testing works for non-deterministic bugs by combining controlled sampling with targeted instrumentation to increase reproduction rates and confirm root causes across codebases.

When do I need to add instrumentation to confirm a root cause?

You need to add instrumentation to confirm a root cause when bugs are hard to reproduce naturally and require targeted metrics to verify hypotheses within a disciplined debugging loop.