diagnose

Diagnose bugs and performance regressions through a structured reproduce-minimise-hypothesise-instrument-fix-regression-test workflow.

Updated May 11, 2026
One-click install
npx skills add https://github.com/IsKenKenYa/Commory --skill diagnose-iskenkenya
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: diagnose
Source: https://github.com/IsKenKenYa/Commory/tree/main/.windsurf/skills/diagnose
Command: npx skills add https://github.com/IsKenKenYa/Commory --skill diagnose-iskenkenya

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It helps you systematically diagnose hard-to-reproduce bugs and performance regressions instead of guessing or repeatedly inspecting code without progress.

Core Features & Use Cases

  • Reproducible feedback loop: Reproduce → minimise → hypothesise → instrument → fix → regression-test, with explicit checkpoints to ensure you’re chasing the right failure.
  • Falsifiable hypothesis ranking: Generates 3–5 ranked hypotheses with clear predictions so each experiment meaningfully narrows the cause.
  • Instrumentation and perf discipline: Uses targeted probing with tagged debug logs, and for performance regressions prioritizes measurement and bisecting over excessive logging.
  • Non-determinism tactics: Raises reproduction rate for flaky issues by looping, parallelising, stressing timing windows, and injecting sleeps.
  • Closure requirements: Enforces cleanup (remove debug tags), confirms the original repro no longer happens, and documents findings when a correct regression-test seam is absent.

Quick Start

Ask the AI to diagnose a failing behaviour by running a structured loop that produces a deterministic pass/fail signal and culminates in a regression test plus cleanup.

Frequently Asked Questions about diagnose

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

FAQPage Schema
How do I debug non-deterministic and flaky bugs that fail to reproduce consistently?

To debug non-deterministic bugs, raise reproduction rates by looping executions, parallelising tests, stressing timing windows, and injecting sleeps to force failures into a deterministic feedback loop.

What is a structured debugging workflow for diagnosing hard software failures and performance regressions?

A structured debugging workflow turns reports into a reproduce, minimise, hypothesise, instrument, fix, and regression-test loop, ensuring you systematically chase the correct failure with measurable signals.

How do I generate falsifiable hypotheses to narrow down the root cause of a performance regression?

Generate 3 to 5 ranked falsifiable hypotheses with clear predictions so each experiment meaningfully narrows the cause of the performance regression during your investigation.

What is the best way to instrument code for debugging without leaving behind temporary changes?

The best way to instrument code is using targeted probing with tagged debug logs, followed by enforcing strict cleanup requirements to remove debug tags once the issue is resolved.

How do I verify a fix for a hard bug when a correct regression-test seam is absent?

To verify a fix without a regression-test seam, confirm the original reproduction no longer occurs and document your findings to ensure the hard bug remains resolved.

Can I diagnose performance slowdowns by bisecting measurements instead of adding excessive logging?

Yes, diagnosing performance regressions prioritizes measurement and bisecting over excessive logging, allowing you to pinpoint slowdowns with targeted instrumentation and a deterministic pass/fail signal.