diagnose

Debug intermittent production failures and performance regressions using a structured 6-phase workflow with falsifiable hypotheses and regression tests.

4|2|Updated Jun 20, 2026
One-click install
npx skills add https://github.com/saitarrun/devforge-ai --skill diagnose-saitarrun
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: diagnose
Source: https://github.com/saitarrun/devforge-ai/tree/main/skills/diagnose
Command: npx skills add https://github.com/saitarrun/devforge-ai --skill diagnose-saitarrun

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill eliminates the wasted time and guesswork of ad-hoc debugging for hard-to-reproduce functional bugs and performance regressions by enforcing a structured, repeatable diagnosis workflow that ensures you build a reliable repro loop before chasing root causes.

Core Features & Use Cases

  • 6-Phase Structured Workflow: Guides you through building a feedback loop, reproducing the bug, generating falsifiable hypotheses, targeted instrumentation, fixing with regression tests, and post-fix cleanup to avoid missing critical steps.
  • Edge Case Handling: Includes specific guidance for non-deterministic bugs, performance regressions, and scenarios where no obvious reproduction path exists.
  • Use Case: If your team is seeing intermittent 500 errors on the checkout endpoint after a recent deploy, use this Skill to build a fast, deterministic repro loop, rank potential root causes, instrument the code to pinpoint the issue, and add a regression test to prevent the bug from returning.

Quick Start

Use the diagnose skill to investigate the intermittent 500 errors reported in the production logs for the user checkout endpoint.

Frequently Asked Questions about diagnose

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

FAQPage Schema
How do I debug hard-to-reproduce functional bugs and intermittent production failures?

Debug intermittent production failures by building a deterministic pass/fail feedback loop, generating falsifiable ranked hypotheses, and using targeted code instrumentation to pinpoint the root cause instead of ad-hoc code inspection.

What is the best way to diagnose a performance regression after a recent deployment?

Diagnose a performance regression by building a deterministic reproduction loop, ranking potential root causes, and instrumenting the code to isolate the slowdown, then locking in the fix with regression tests at correct code seams.

How do I troubleshoot non-deterministic edge case bugs when standard debugging approaches fail?

Troubleshoot non-deterministic edge case bugs by applying a disciplined 6-phase workflow that includes specific guidance for scenarios where no obvious reproduction path exists, ensuring a reliable repro loop is built before chasing root causes.

How do I stop guessing during root cause analysis and eliminate wasted debugging effort?

Stop guessing during root cause analysis by enforcing a repeatable diagnosis workflow that requires building a deterministic feedback loop and generating falsifiable hypotheses, eliminating wasted effort from ad-hoc code inspection.

Where should I add regression tests to lock in bug fixes and prevent future recurrence?

Add regression tests at correct code seams during the targeted fix phase of the debugging workflow to lock in fixes, verify the deterministic pass/fail loop, and prevent intermittent defects or performance regressions from returning.