Debug team

Skill claude-hangar/claude-hangar/core/skills/debug-team

Production-grade configuration management for Claude Code. Hooks, agents, skills, multi-project orchestration.

Install

npx -y skills add claude-hangar/claude-hangar --skill debug-team

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Launch parallel debugging agents for systematic root-cause analysis. Use when: "debug team", "help me debug", "debug-team", "find the bug".

SKILL.md

3.4 KB, as published. Nobody here has run it

/debug-team — Multi-Agent Debugging Team

Launch a parallel debugging team that investigates bugs from multiple angles simultaneously. Combines deep code analysis, error pattern matching, and test-driven verification.

Team Composition

Agent	Role	Focus
explorer-deep	Code Analyst	Trace execution paths, find root cause in code
build-resolver-*	Build Expert	Resolve build/test/dependency failures (auto-detected)
tdd-guide	Test Expert	Write reproduction test, verify fix

Instructions

Step 1: Capture Bug Context

From the user's $ARGUMENTS or conversation, extract:

Symptom: What's failing? (error message, unexpected behavior, test failure)
Trigger: When does it happen? (specific input, sequence, timing)
Scope: Which files/modules are involved?

If insufficient context, ask: "What error or unexpected behavior are you seeing?"

Step 2: Detect Stack

Scan the project for build tools to select the right build-resolver:

tsconfig.json or package.json with TypeScript → build-resolver-typescript
go.mod → build-resolver-go
pyproject.toml, setup.py, requirements.txt → build-resolver-python
None detected → skip build-resolver, use explorer-deep twice with different prompts

Step 3: Launch Parallel Agents

Launch all applicable agents simultaneously:

Agent({
  subagent_type: "explorer-deep",
  description: "Root cause analysis",
  prompt: "Investigate this bug: [symptom]. Trace the execution path from [trigger point]. Find the root cause. Check: data flow, state mutations, error handling, race conditions, edge cases. Report: SYMPTOM → CONTEXT → CAUSE → FIX suggestion."
})

Agent({
  subagent_type: "build-resolver-[lang]",
  description: "Build/dependency analysis",
  prompt: "Analyze this error: [error message]. Check: dependency versions, import resolution, build configuration, type errors, missing modules. Report what's broken and how to fix it."
})

Agent({
  subagent_type: "tdd-guide",
  description: "Reproduction test",
  prompt: "Write a failing test that reproduces this bug: [symptom]. The test should: 1) Set up the minimal scenario, 2) Trigger the bug, 3) Assert the expected (correct) behavior. The test MUST fail with the current code. Report the test and where to place it."
})

All agents MUST be launched in a single message (parallel execution).

Step 4: Synthesis

After all agents complete, synthesize findings:

## Debug Team Report

### Bug Summary
[one-line description]

### Root Cause
[from explorer-deep findings]

### Evidence
- Code path: [file:line → file:line → failure point]
- Build/dependency factor: [from build-resolver, if applicable]

### Reproduction Test
[from tdd-guide — the failing test]

### Recommended Fix
[concrete fix based on root cause analysis]

### Prevention
[what to change so this class of bug can't recur]

Step 5: Offer to Fix

Present the fix plan and ask: "I've identified the root cause. Shall I:

Apply the fix and run the reproduction test to verify?
Just show the fix for manual review?"