Skip to main content
NeVibeCode · collection · 30 · list updated

Debugging skills

Debugging skills are instructions Claude Code or Codex follow to find the root cause of a bug: gather evidence, read logs and traces, name the cause and only then change code. These are the top 30 by installs. They go in .claude/skills for Claude Code and .agents/skills for Codex.

Ordered by install count from the source, then by repository stars on GitHub. A language model writes each description from the source file; only descriptions that passed the automatic check are listed.

  1. 1

    diagnosing-bugs

    diagnosing-bugs is a Claude Code skill for investigating hard bugs and performance regressions. It activates when someone asks to diagnose a problem or reports code that fails, throws errors, or runs slowly.

    skill · Matt Pocock · MIT

    834,802installs
  2. 2

    safe-debug

    safe-debug is a Claude Code skill that diagnoses failures in deep learning research and drafts minimal patches without code changes unless approved. It activates when training fails.

    skill · lllllllama · MIT

    363,000installs
  3. 3

    systematic-debugging

    systematic-debugging is a Claude Code and Codex skill that blocks fixes until root cause analysis is complete. It activates on test failures, bugs, or unexpected behavior.

    skill · Jesse Vincent · MIT

    306,118installs
  4. 4

    investigate-first

    The investigate-first skill keeps the agent in diagnosis mode: gather evidence, rank hypotheses, name the cause. Only then does it touch code.

    skill · JuliusBrussee · Apache-2.0

    129,600installs
  5. 5

    surgical-patch

    The surgical-patch skill constrains AI coding agents to the narrowest layer responsible for a bug. It prevents unwanted refactoring and demands focused regression checks.

    skill · JuliusBrussee · Apache-2.0

    128,900installs
  6. 6

    debugging-and-error-recovery

    The debugging-and-error-recovery Claude Code skill traces failures to their root cause and checks the fix. It activates for failing tests, broken builds, unexpected errors, or behavior that no longer matches expectations.

    skill · addyosmani · MIT

    79,600installs
  7. 7

    insforge-debug

    insforge-debug — a debugging skill for InsForge projects: traces SDK errors, RLS denials, gateway timeouts and failed edge function deploys back to a cause.

    skill · InsForge · Apache-2.0

    63,100installs
  8. 8

    convex-monitor

    The convex-monitor skill blocks on the next event in a Convex app — a dev error, a prod error or a feature request — and hands it to the agent.

    skill · convex · Apache-2.0

    43,800installs
  9. 9

    convex-sentinel

    convex-sentinel configures Sentinel error capture inside your own Convex deployment. Useful for tracking server failures and client React crashes without sending logs to third-party services.

    skill · convex · Apache-2.0

    43,700installs
  10. 10

    golang-troubleshooting

    golang-troubleshooting is a Claude Code skill for systematic root-cause debugging of Go deadlocks, panics, and race conditions. It reproduces bugs before fixes.

    skill · samber · MIT

    42,300installs
  11. 11

    vue-debug-guides

    vue-debug-guides — a skill for diagnosing Vue 3 errors: reactivity, watchers, components, templates, and hydration. When the agent sees a Vue bug, it looks up the cause in a catalog of common issues.

    skill · github.com/vuejs-ai · MIT

    28,400installs
  12. 12

    grounding-before-coding

    The grounding-before-coding skill makes the agent inspect code and data before changing a project or explaining a bug. It fits unfamiliar repositories and non-trivial work.

    skill · riekelt

    25,600installs
  13. 13

    opencli-autofix

    opencli-autofix repairs broken OpenCLI adapters when websites alter their DOM or APIs. The agent inspects trace logs, explores live pages, and applies targeted patches.

    skill · jackwener · Apache-2.0

    18,200installs
  14. 14

    hunt

    hunt pinpoints root causes before touching code when an app crashes or regresses. Enforces reproducible evidence and verification before applying any fix.

    skill · tw93 · MIT

    15,300installs
  15. 15

    chrome-devtools · github official

    Controls a live Chrome browser through Chrome DevTools MCP for page actions, debugging, and performance traces. Use it when an agent needs to inspect a web interface.

    skill · github · MIT

    15,258installs
  16. 16

    base44-troubleshooter

    base44-troubleshooter is a Claude Code skill that retrieves Base44 backend logs and workflow runs to diagnose crashed functions, silent scheduled jobs, and failed automations.

    skill · base44 · MIT

    13,900installs
  17. 17

    debugging-strategies

    The debugging-strategies skill gives an agent a step-by-step way to investigate bugs, crashes, and performance problems. Use it when a project behaves unexpectedly.

    skill · Seth Hobson · MIT

    13,414installs
  18. 18

    memory-safety-patterns

    memory-safety-patterns — a Claude Code skill covering memory-safe patterns in Rust, C++ and C. It kicks in for systems code, resource handling and hunting down memory bugs.

    skill · Seth Hobson · MIT

    11,746installs
  19. 19

    principle-fix-root-causes

    principle-fix-root-causes is a Claude Code skill that traces a failure to the first wrong decision in the code. It starts with reproduction, then checks the state and recurring fault pattern.

    skill · backnotprop · MIT

    10,800installs
  20. 20

    sentry-fix-issues

    sentry-fix-issues finds Sentry errors via MCP, inspects stack traces and breadcrumbs, and helps write fixes. Useful for debugging production issues and triaging backlogs.

    skill · getsentry · Apache-2.0

    10,800installs
  21. 21

    cargo-diagnostics

    cargo-diagnostics — a Claude Code skill that explains why a Cargo run failed or spent too many credits. It traces runs step by step, groups batch errors by root cause, and breaks down credit use by node.

    skill · getcargo · MIT

    10,400installs
  22. 22

    love2d-core

    love2d-core — a skill for Claude Code that builds and debugs LÖVE 11.x games in Lua, handling loops and pixel-art scaling. Invoked when editing main.lua or fixing frame rates.

    skill · gamedev-skills · Apache-2.0

    10,200installs
  23. 23

    principle-attack-the-premise

    principle-attack-the-premise is a Claude Code skill that tests a shared premise after two or more fixes fail at the same gate. It checks the assumption before another fix is written.

    skill · backnotprop · MIT

    10,100installs
  24. 24

    chrome-devtools · ChromeDevTools

    This skill connects Chrome DevTools through MCP for page debugging, browser automation, performance checks, and network inspection.

    skill · ChromeDevTools · Apache-2.0

    9,400installs
  25. 25

    devtools

    devtools is a Claude Code skill that adds a debugging panel to json-render apps, showing the component tree, state, actions, streaming patches, and an element picker.

    skill · vercel-labs · Apache-2.0

    8,400installs
  26. 26

    debugging-lambda-timeouts

    debugging-lambda-timeouts — a skill for Claude Code, Codex, Cursor, and other agents that diagnoses AWS Lambda timeout failures by walking through function config, CloudWatch logs, VPC settings, cold starts, and downstream dependencies.

    skill · aws · Apache-2.0

    7,200installs
  27. 27

    platform-apex-logs-debug

    The platform-apex-logs-debug skill reads Salesforce debug logs: finds the root cause of a failure, spots governor limit hits and explains stack traces.

    skill · forcedotcom · Apache-2.0

    7,200installs
  28. 28

    ios-simulator

    Launch, inspect, and debug native, React Native, Expo, and Flutter apps on iOS Simulator from your agent. The agent can tap, type, scroll, validate code changes, and collect error logs.

    skill · callstack · MIT

    7,000installs
  29. 29

    lynx-devtool

    lynx-devtool is a Claude Code skill that inspects running Lynx applications using the agent-lynx CLI. It takes element snapshots, runs UI actions, and sends CDP commands across mobile and desktop.

    skill · lynx-community · Apache-2.0

    7,000installs
  30. 30

    mentoring-juniors official

    mentoring-juniors turns Copilot into a mentor that guides juniors with questions instead of handing over answers. Useful when you are learning or reading unfamiliar code.

    skill · github · MIT

    6,900installs

How to choose a debugging skill

The catalog has 404 debugging skills. The most installed ones solve one problem: an agent tends to patch the symptom. They block code changes until a cause is named and evidence is gathered, and they keep the fix minimal, with no refactoring on the side. The second group is tied to a stack or service: Go, Vue, Convex, AWS Lambda, Sentry, Chrome DevTools. Such a skill knows where that tool keeps its logs and how to read them.

One general method skill is enough; add stack-specific ones per project. Nearby: testing skills, code review and refactoring, DevOps.

FAQ

Is there a skill that fixes the cause of a bug, not the symptom?

Yes, that is the main class of skills in this list: they block code changes until the root cause is found. By installs, the list is led by diagnosing-bugs, safe-debug and systematic-debugging.

Which skill should I use to read logs?

One for your stack: the list covers AWS Lambda and CloudWatch logs, Salesforce debug logs and Sentry errors. General debugging skills also start from logs and traces.

How do I install a debugging skill in Codex?

Put the skill folder in .agents/skills in the project or run npx skills add owner/repo -a codex. For Claude Code the folder is .claude/skills.

Other collections

Project sponsors

Companies that keep this analytics open