diagnose

Isolate bug causes through reproducible feedback loops and hypothesis testing.

Updated Jul 17, 2026
One-click install
npx skills add https://github.com/kartikkabadi/skills --skill diagnose-kartikkabadi
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: diagnose
Source: https://github.com/kartikkabadi/skills/tree/main/diagnose
Command: npx skills add https://github.com/kartikkabadi/skills --skill diagnose-kartikkabadi

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Diagnose helps you systematically investigate bugs, regressions, and performance problems instead of guessing. It creates a reliable feedback loop so you can reproduce the issue, narrow the cause, and verify the fix.

Core Features & Use Cases

  • Reproduction-first debugging: Build a deterministic loop using tests, scripts, traces, or harnesses before changing code.
  • Hypothesis-driven investigation: Rank possible causes, then instrument one variable at a time to confirm or eliminate them.
  • Regression protection: Turn the minimized repro into a failing test, apply the fix, and rerun the original scenario to confirm the bug is gone.
  • Use cases: Debugging flaky failures, isolating performance regressions, tracing intermittent UI issues, and validating fixes with repeatable tests.

Quick Start

Use the diagnose skill to reproduce the bug, propose ranked hypotheses, and build a minimal regression test for the failure.

Frequently Asked Questions about diagnose

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

FAQPage Schema
How do I systematically debug flaky tests and intermittent failures?

Systematically debug flaky tests by building a deterministic reproduction loop first, then ranking falsifiable hypotheses and instrumenting one variable at a time to isolate the cause. This prevents guessing and confirms the fix with a regression test.

What is hypothesis-driven debugging and how does it isolate code regressions?

Hypothesis-driven debugging isolates code regressions by ranking possible causes and testing them one by one through targeted instrumentation. You eliminate incorrect assumptions sequentially until the root cause is confirmed and validated.

How do I trace performance regressions across CLI, API, and UI workflows?

Trace performance regressions across CLI, API, and UI workflows by establishing reproducible feedback loops with traces or scripts. You propose ranked hypotheses, instrument specific bottlenecks, and verify the slowdown is resolved.

Can I diagnose broken behavior without deterministic reproduction steps?

Diagnosing broken behavior requires deterministic reproduction steps as a prerequisite. You must build a reliable feedback loop using tests, scripts, or harnesses before changing code to ensure you can accurately isolate the cause.

What is the best way to fix hard bugs without randomly changing code?

The best way to fix hard bugs without randomly changing code is reproduction-first debugging. You build a minimal failing test that reproduces the defect, apply the fix at the correct seam, and rerun the scenario to confirm the bug is gone.

When should I not use a systematic debugging approach for failing tests?

You should avoid a systematic debugging approach when you lack reproducible steps or when the defect stems from obvious syntax errors or missing dependencies that are immediately identifiable from standard stack traces.