dev-debug

Diagnose hard bugs through reproduction, bisection, and debugger configuration generation.

2|Updated May 16, 2026
One-click install
npx skills add https://github.com/avbel/ai-skills --skill dev-debug-avbel
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: dev-debug
Source: https://github.com/avbel/ai-skills/tree/main/skills/dev-debug
Command: npx skills add https://github.com/avbel/ai-skills --skill dev-debug-avbel

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Bugs that survive the first obvious fix often get patched with guesses that hide the real cause. This Skill enforces a root-cause-first debugging discipline: build a failing feedback loop, test ranked hypotheses, and only then fix — plus it generates real debugger setups for your IDE when print-debugging stalls. ## Core Features & Use Cases - Systematic reproduction workflow: Builds a deterministic failing command (test, curl, or git bisect run wrapper) before any fix attempt, and ranks 3-5 falsifiable hypotheses to test in order. - Debugger config generation: Produces VS Code launch.json entries, JetBrains run configurations, and CLI fallbacks (node --inspect-brk, rust-lldb, gdb, rr) tailored to the bug location, with breakpoint locations and watch expressions. - Fix and lock-in: Commits the regression test first, fixes at the owning boundary, runs the full suite, and records the root cause. - Use Case: A flaky race condition in a Node service resists logging. The Skill raises the reproduction rate with a 100x loop, then generates an rr record/replay workflow or an attach config for the running server. ## Quick Start Ask the agent to help debug a failing test or endpoint and generate a VS Code launch.json configuration for it.

Frequently Asked Questions about dev-debug

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

FAQPage Schema
How do I debug a bug that keeps coming back after fixing?▼

Build a deterministic failing command first — a test, curl call, or git bisect run wrapper — then list 3-5 falsifiable hypotheses and test only the top one with runtime evidence. Never write a fix before confirming the root cause.

How to set up VS Code debugger for Node or Rust tests?▼

For Node, create a launch.json entry of type node pointing at the test binary, or run node --inspect-brk with chrome://inspect. For Rust, build with cargo test --no-run and debug the binary via CodeLLDB or rust-lldb with breakpoints at file:line.

What is the best way to debug race conditions and heisenbugs?▼

Use rr record and rr replay on Linux to reverse-continue from the crash back to the cause, since debugger pauses can mask races. Alternatively, add timestamped logging with thread IDs and reconstruct the interleaving.

When should I use git bisect instead of reading diffs?▼

Use git bisect run with your failing command whenever a regression spans more than a handful of commits. It mechanically finds the exact offending commit, which is faster and more reliable than reading diffs manually.

Why does print debugging stop working on complex bugs?▼

Print debugging fails when state mutates unexpectedly, timing matters, or the failure point is far from the cause. After two failed hypotheses, switch to an interactive debugger with conditional breakpoints at the suspect locations.