resonance-engineering-debugger

Diagnose root causes of bugs through hypothesis-driven reproduction and isolation.

Updated Dec 1, 2025
One-click install
npx skills add https://github.com/manusco/wolfgang --skill resonance-engineering-debugger-manusco
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: resonance-engineering-debugger
Source: https://github.com/manusco/wolfgang/tree/main/.agents/skills/engineering/debugger
Command: npx skills add https://github.com/manusco/wolfgang --skill resonance-engineering-debugger-manusco

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? It prevents blind patches and shotgun fixes by enforcing a strict scientific debugging protocol: no fix is applied until the bug is reproduced deterministically and the root cause is proven with evidence. ## Core Features & Use Cases - 9-Step Debugging Protocol: Search memory, reproduce, isolate via binary search or git bisect, hypothesize, instrument, verify, fix surgically, log the RCA, and attest completion. - Agent Trajectory Diagnosis: Classifies AI agent failures into five modes (Planning, Tool Selection, Tool Argument, Memory, Reasoning) and applies minimal prompt fixes. - Defense in Depth Hardening: Traces every layer the bad data crossed and makes the illegal state unrepresentable so the whole bug class cannot recur. - Use Case: When users report an intermittent wrong cart total, the Skill refuses to patch immediately, lists competing hypotheses (race condition, caching, floating-point), and builds a deterministic reproduction script before any code change. ## Quick Start Ask the debugger to investigate a bug report or flaky test by describing the error, the environment where it occurred, and any logs you have.

Frequently Asked Questions about resonance-engineering-debugger

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

FAQPage Schema
How do I debug an intermittent or flaky bug?▼

Build a reproduction script that triggers the failure deterministically before attempting any fix. List multiple hypotheses such as race conditions, caching, or floating-point errors, then isolate the failing surface with binary search or git bisect.

How to find the root cause of a production-only bug?▼

Treat environment divergence as the primary suspect: diff environment variables, config, and database schema between staging and production. Never accept 'works in staging' as proof the logic is correct, and escalate after three failed hypotheses.

What is a shotgun fix and why avoid it?▼

A shotgun fix changes several variables at once without isolating the cause, making the fix unverifiable. Without a reproduction case you cannot know which change mattered, and swallowed exceptions may hide the real failure.

How do I debug an AI agent that gives wrong answers?▼

Extract the agent's trajectory of thoughts and tool calls, then classify the failure into one of five modes: Planning, Tool Selection, Tool Argument, Memory, or Reasoning. Apply the minimal prompt constraint for that mode and re-run the same input to verify.

When should I use git bisect for debugging?▼

Use git bisect when a regression appeared and you know a past commit worked. Mark the current commit bad and a known-good commit good, then test each midpoint checkout until git identifies the exact commit that introduced the bug.