debug

Diagnose bugs through a five-phase reproduce, narrow, diagnose, fix, verify workflow with regression bisect.

6|Updated Mar 27, 2026
One-click install
npx skills add https://github.com/greglas75/zuvo --skill debug-greglas75
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: debug
Source: https://github.com/greglas75/zuvo/tree/main/skills/debug
Command: npx skills add https://github.com/greglas75/zuvo --skill debug-greglas75

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Debugging often devolves into guesswork, random code changes, and fixes that address symptoms instead of root causes. This Skill enforces a disciplined investigation process that turns a bug report or stack trace into a confirmed root cause, a verified fix, and a permanent regression test. ## Core Features & Use Cases - Five-Phase Framework: Systematically reproduce the failure, narrow the search space, diagnose the root cause with ranked hypotheses, then fix and verify with targeted and full test suites. - Automated Regression Bisect: Use the --regression flag to run git bisect with a minimal reproducer test and pinpoint the exact commit that broke previously working behavior. - Quality Gates and Reports: Produce a structured debug report with root cause analysis, CQ/Q self-evaluation scores, adversarial review, and defense-in-depth guards. - Use Case: A user reports "checkout totals are wrong since last deploy." Invoke the skill with the error details; it classifies the bug, traces the call chain, confirms the root cause at a specific file and line, applies a minimal fix, and writes a regression test. ## Quick Start Ask the AI to debug the issue by pasting your error message or stack trace, or say "debug this regression with --regression" to bisect the breaking commit.

Frequently Asked Questions about debug

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

FAQPage Schema
How do I debug a bug systematically instead of guessing?

Follow a five-phase process: reproduce the failure with clear expected versus actual behavior, narrow the search space using logs and recent changes, diagnose by tracing the execution path with ranked hypotheses, then apply a minimal fix and verify with tests. This skill automates that workflow.

How to find which commit broke my code with git bisect?

Use the --regression flag to run automated git bisect. The skill writes a minimal reproducer test that fails on HEAD, verifies a known-good commit, then runs git bisect to identify the exact breaking commit before fixing.

Can this debug workflow handle flaky or intermittent test failures?

Yes. Intermittent failures are flagged as potential race conditions, shared mutable state, or test-order dependencies. The skill runs the failing test multiple times to confirm flakiness and applies a dedicated async/flaky diagnostic sequence.

When should I not use git bisect for regression debugging?

Skip bisect when the bug is not a regression, when no deterministic test can reproduce it, when the change is only one or two commits old (just read the diff), or when uncommitted changes cannot be stashed cleanly.

What does the debug report include after a fix?

The report includes reproduction details, a hypothesis table with evidence, the root cause at a specific file and line, the before/after fix, verification results, CQ and Q self-evaluation scores, and a permanent regression test.