Skip to main content
subagent CX Codex

debugger · rohitg00 — Claude Code subagent: debugging and installation

debugger is a Claude Code subagent that investigates difficult bugs, failed tests, and runtime errors. It traces the cause before proposing a code change.

MIT ★ 2,966 updated 2026-09-29

Install into Claude Code · Plugin marketplace All methods and platforms ↓
/plugin marketplace add rohitg00/pro-workflow
/plugin install pro-workflow@pro-workflow
Open on GitHub

What debugger does

debugger reproduces the failure and records its error message and stack trace. It then tests possible explanations in order, reading relevant code, checking git log, and comparing nearby patterns that work.

Once it confirms the cause, the subagent proposes a minimal change and explains how to test it. It waits for approval before editing code. If three investigation rounds do not resolve the issue, debugger hands over its findings.

What it helps with

  • reproduce a failure and capture its stack trace
  • trace a failing test to its cause
  • check whether a bug followed a recent Git change
  • propose a small fix with a verification step

Example prompts for the agent

  • Find out why tests/auth.test.ts fails, and report the cause before changing anything.
  • Investigate the runtime error in src/server.ts; wait for my approval before editing code.
  • Check whether the test failures started with recent Git changes.

How to install debugger

1Plugin marketplace

/plugin marketplace add rohitg00/pro-workflow
/plugin install pro-workflow@pro-workflow

2Manually: put the files in place

.claude/agents/debugger.md

For all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention

Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean

Questions about debugger

How do I install debugger in Claude Code?

Add the marketplace with /plugin marketplace add rohitg00/pro-workflow, then install the plugin with /plugin install pro-workflow@pro-workflow. The agent file path is .claude/agents/debugger.md.

What does debugger do?

It reproduces bugs, checks possible causes, and identifies a confirmed root cause. It then proposes a minimal fix and waits for approval before changing code.

When does debugger escalate an investigation?

If three rounds of investigation do not resolve the issue, the subagent passes its findings to the user.

Source text

debugger.md · MIT
---
name: debugger
description: Specialized debugging agent. Use when facing hard bugs, test failures, or runtime errors that need systematic investigation.
tools: ["Read", "Glob", "Grep", "Bash"]
model: opus
memory: project
---

# Debugger - Systematic Bug Investigation

Methodical debugging that narrows down root causes before proposing fixes.

## Workflow

### 1. Reproduce

- Run the failing test or reproduce the error
- Capture the exact error message, stack trace, and context
- Note: is this a regression (worked before) or new behavior?

### 2. Hypothesize

Generate 2-3 hypotheses ranked by likelihood:

```text
Hypothesis 1 (70%): [most likely cause]
  Evidence for: [what supports this]
  Evidence against: [what contradicts]
  Test: [how to verify]

Hypothesis 2 (20%): [alternative cause]
  ...

Hypothesis 3 (10%): [unlikely but possible]
  ...
```

### 3. Investigate

Test each hypothesis starting with the most likely:

- Read relevant code paths
- Check git log for recent changes to affected files
- Search for similar patterns that work correctly
- Add targeted debug output if needed

### 4. Root Cause

Present the confirmed root cause:

```text
ROOT CAUSE: [what's actually wrong]
WHERE: [file:line]
WHY: [how it got this way]
SINCE: [when it was introduced, if knowable]
```

### 5. Fix Proposal

Propose the minimal fix. Explain why this fix is correct.

```text
FIX: [description]
CHANGES:
  - file.ts:42 - [what to change]
RISK: [low/medium/high]
TESTS: [how to verify the fix]
```

Wait for approval before implementing.

## Rules

- Never guess. Investigate systematically.
- Never apply fixes without finding root cause first.
- Check the git blame — recent changes are more likely to be the cause.
- Use project memory to recall previous bugs in the same area.
- If stuck after 3 rounds of investigation, escalate to user with findings so far.
- Capture debugging learnings: `[LEARN] Debugging: <insight>`

## Anti-Patterns to Avoid

- "Shotgun debugging" — changing random things hoping something works
- Ignoring stack traces — they tell you exactly where to look
- Not reproducing first — you can't fix what you can't see
- Fixing symptoms instead of root causes

© Rohit Ghumare · MIT · rohitg00/pro-workflow

Similar subagents

All subagents for Agent setup: memory, context, orchestration →

debugger by other authors comes first.

debugger VoltAgent Community debugger diagnoses software failures, reads stack traces, and isolates root causes to fix elusive bugs. Helpful when logs alone do not explain a crash. CX Codex 13,459installs debugger VoltAgent debugger isolates bugs across code paths, stack traces, and failing test runs. It pinpoints the defect and recommends a minimal fix. CX Codex ★ 6,505repo stars debugger CloudAI-X debugger finds root causes for test failures, memory leaks, and crashes. The subagent tests hypotheses systematically instead of guessing fixes. CX Codex ★ 1,424repo stars debugger qdhenry debugger — a targeted troubleshooting subagent for Claude Code that isolates root causes, inspects test failures, and writes minimal diffs. CX Codex ★ 1,357repo stars debugger justxor The debugger subagent reproduces errors and failing tests, tracks down their cause, then checks the fix. Use it when a stack trace or unexpected behavior needs investigation. CX Codex ★ 353repo stars diagnose LegendApp diagnose — a skill for Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, and OpenCode that runs a structured bug investigation on explicit request. CX Codex +1 429installs scout simota scout is a Claude Code skill that investigates bugs and isolates root causes. It activates for defect and regression investigations without writing code. CX Codex +1 101installs investigate-bug rerun-build investigate-bug — a skill for Claude Code that gathers evidence and traces a bug to its cause before a fix is written. It starts on bug reports, incidents, customer complaints, or requests to reproduce an issue. CX Codex +1 ★ 1,120repo stars argent-metro-debugger software-mansion argent-metro-debugger — a skill for debugging React Native via Metro and CDP: connects to the runtime, reads console logs, inspects components and evaluates JavaScript. Works with iOS, Android, Fire TV and Chromium. CX Codex +1 32,400installs mcp-debugger debugmcp Headless step-through debugger over the Debug Adapter Protocol for coding agents. Sets breakpoints, steps through execution, and inspects variables without an IDE. CX Codex +1 ★ 194repo stars node-inspect-debugger openclaw node-inspect-debugger is a Claude Code skill for debugging Node.js with node inspect, breakpoints, CPU profiles, and heap snapshots. It investigates async hangs that logs don't explain. CX Codex +1 299installs

From the same set83

The whole set →
session-handoff session-handoff generates a structured handoff document capturing git state, open tasks, decisions, and a resume prompt for your next session. CX Codex +1 ★ 2,966repo stars replay-learnings The replay-learnings skill searches project memory and similar sessions for lessons tied to a new task. Run it before coding to bring earlier corrections into view. CX Codex +1 ★ 2,966repo stars skill-optimize skill-optimize runs an offline optimization loop over a skill's accumulated examples, proposes improvements to its description file, validates candidates, and updates only when performance strictly improves. CX Codex +1 ★ 2,966repo stars orchestrate orchestrate coordinates multi-phase feature development with research, planning, gated implementation, and code review in Claude Code. CX Codex +1 ★ 2,966repo stars auto-setup The auto-setup skill detects your project type and automatically configures linters, type checking, tests, and formatting. Use it when starting work on an unfamiliar codebase. CX Codex +1 ★ 2,966repo stars wiki-query wiki-query queries local markdown wikis indexed by wiki-builder using SQLite FTS5 BM25 retrieval. It delivers top passages with citations so the agent builds on existing notes. CX Codex +1 ★ 2,966repo stars orchestrator orchestrator — a subagent for Claude Code that splits complex builds into three phases: research, plan, implement. Each phase requires approval before the next — use it for features touching 5+ files or needing architecture decisions. CX Codex ★ 2,966repo stars learn-rule learn-rule is a Claude Code skill that turns corrections and surfaced mistakes into categorized rules in permanent memory, carrying them across sessions to prevent repeats. CX Codex +1 ★ 2,966repo stars

Project sponsors

Companies that keep this analytics open