diagnose

Diagnoses bugs and performance regressions using falsifiable experimentation and deterministic pass/fail feedback loops.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/kileroppo/AI-Coding-Guidance-Skills --skill diagnose-kileroppo
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: diagnose
Source: https://github.com/kileroppo/AI-Coding-Guidance-Skills/tree/main/skills/diagnose
Command: npx skills add https://github.com/kileroppo/AI-Coding-Guidance-Skills --skill diagnose-kileroppo

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps you systematically diagnose hard bugs and performance regressions by turning vague symptoms into a reliable feedback loop that reveals the root cause.

Core Features & Use Cases

  • Build a deterministic feedback loop: creates pass/fail signals via failing tests, HTTP/CLI scripts, browser automation, trace replay, or bisection harnesses.
  • Reproduce before theorizing: verifies the failure matches the user-described symptom and is reproducible (or reproducible at a debuggable rate for flaky issues).
  • Generate falsifiable hypotheses: produces 3–5 ranked, testable predictions before making changes.
  • Instrument with controlled experiments: probes one variable at a time using targeted debugger/logging, with a cleanup discipline.
  • Fix with regression protection: writes a regression test before the fix when a correct test seam exists, then re-runs the original loop and removes instrumentation.

Quick Start

Ask your agent to diagnose this by running a reproducible loop: reproduce the reported failure, generate falsifiable hypotheses, instrument targeted probes, apply the fix, and verify the original scenario with a regression test.

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?

Debugging hard bugs requires converting vague symptoms into a reliable feedback loop with deterministic pass/fail signals. You can reproduce failures using tests, HTTP scripts, browser automation, or trace replay to verify the issue before theorizing.

What is the best way to find a performance regression in my codebase?

Finding a performance regression involves building a deterministic feedback loop to measure pass/fail signals. You can use bisection harnesses or stress runs to reproduce the regression, generate falsifiable hypotheses, and instrument targeted probes.

How do I systematically diagnose a flaky test failure?

Diagnosing a flaky test failure requires stepwise discipline across reproduce, hypothesize, instrument, fix, and regression-test phases. You create a debuggable reproduction rate, generate 3-5 ranked falsifiable predictions, and probe one variable at a time.

Can I use bisection to find the root cause of a software regression?

Bisection helps find software regressions by creating a deterministic pass/fail signal via a bisection harness. This builds a reliable feedback loop to stepwise reproduce, hypothesize, instrument, and fix the identified root cause.

What should I do when my debugging loop is not working for non-deterministic failures?

When a debugging loop fails for non-deterministic failures, you should explicitly stop if no reliable feedback loop exists. Debugging requires a deterministic pass/fail signal via tests, scripts, or trace replay to proceed safely.