diagnose

Name: diagnose
Availability: InStock
Author: Haohao

Diagnose bugs and performance regressions by reproducing failures and proving hypotheses with evidence.

779|80|Updated Jun 20, 2025

Search Tags:#debugging #failure analysis #regression testing #engineering workflow #reproducibility #fault isolation #performance regression

Ready to run

npx skills add https://github.com/Haohao-end/openagent --skill diagnose-haohao-end

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: diagnose
Source: https://github.com/Haohao-end/openagent/tree/main/api/internal/core/skills/catalog/diagnose
Command: npx skills add https://github.com/Haohao-end/openagent --skill diagnose-haohao-end

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you systematically diagnose bugs and performance regressions by turning vague “something is broken” reports into evidence-backed, reproducible hypotheses.

Core Features & Use Cases

Feedback-loop first troubleshooting: prioritize writing failing tests, curl scripts, fixtures, or harnesses to lock the exact failure.
Reproduction and signal selection: ensure you observe the same failure, then validate assumptions using the strongest minimal logs or measurements.
Root-cause to regression lock: fix the underlying issue first, then add a regression test to prevent recurrence.

Quick Start

Ask the diagnose skill to help you reproduce the reported failure, identify the most likely root causes, and propose the minimal evidence to confirm each hypothesis.

Frequently Asked Questions about diagnose

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

FAQPage Schema

How do I debug intermittent failures when I don't have enough logging?▼

You can diagnose intermittent failures without excessive logging by creating deterministic reproduction steps and using minimal-strong-signal verification to prove hypotheses with reproducible evidence.

What's the best way to narrow down a root cause for a performance regression?▼

To isolate a performance regression, establish deterministic reproduction steps, measure degraded behavior with minimal strong signals, and lock the regression through tests or fixtures to prove your root-cause hypothesis.

How does fault isolation work in an engineering workflow when behavior breaks unexpectedly?▼

Fault isolation turns vague broken behavior reports into evidence-backed hypotheses by requiring deterministic reproduction steps to observe the exact failure and validating assumptions before fixing the underlying issue.

How do I lock a regression test after fixing an intermittent failure?▼

You lock a regression test after fixing an intermittent failure by adding tests, fixtures, or harnesses that reproduce the exact failure scenario, ensuring the same issue is handled safely and preventing recurrence.

Can I reproduce a complex failure if I only have a vague report that something is broken?▼

Yes, you can reproduce a complex failure from a vague report by prioritizing a feedback-loop first approach, writing curl scripts or fixtures to lock the exact failure, and proving hypotheses with reproducible evidence.