parallel-debugging Skill for Claude: What It Does and How to Install It

Makes Claude debug hard bugs with competing hypotheses, evidence with file and line citations, and a final root cause ruling.

Installs: 10.2K; on skills.sh
Stars: 42.1K; on GitHub
Files: 1; besides SKILL.md
Updated: Oct 2026; last repo commit

This skill is a method for debugging with Analysis of Competing Hypotheses. Claude proposes causes across 6 failure categories (logic error, data issue, state problem, integration failure, resource issue and environment), then investigates each one for evidence.

It sets evidence rules (4 evidence types, citations as file:line, 3 confidence levels) and a 4-step arbitration protocol that ends with a fix-validation checklist. references/hypothesis-testing.md holds task templates, a report template and common hypothesis patterns.

Good for

Developers stuck on a bug with several plausible causes
Teams using Claude Code agent teams for investigation
Anyone who wants evidence, not guesses, in a root cause report

Install parallel-debugging

Claude Code

Copy this prompt.

Install the parallel-debugging skill in Claude Code. Run this command:

/plugin marketplace add wshobson/agents
/plugin install agent-teams@claude-code-workflows

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Pick one route. The plugin routes install the whole agent-teams plugin: 6 skills, 4 agents and 7 commands (counted from the plugin folder in the repo tree, which has no hooks or MCP config files). The plugin README says its team commands need Claude Code's experimental Agent Teams feature, turned on with an environment variable. The skills-only routes install just this skill, with no agents or commands.

From the project README. Source.

What's in the skill

When it loads

Debug complex issues using competing hypotheses with parallel investigation, evidence collection, and root cause arbitration. Use this skill when debugging bugs with multiple potential causes, performing root cause analysis, or organizing parallel investigation workflows.

Sections in SKILL.md

When to Use This Skill
Hypothesis Generation Framework
1. Logic Error
2. Data Issue
3. State Problem
4. Integration Failure
5. Resource Issue
6. Environment
Evidence Collection Standards
What Constitutes Evidence
Citation Format
Confidence Levels
Result Arbitration Protocol
Step 1: Categorize Results
Step 2: Compare Confirmed Hypotheses
Step 3: Determine Root Cause

Key facts

Author: wshobson (repo owner)
License: MIT (crawl data and agent-teams plugin.json)
Installs on skills.sh: 10,217 (Nov 2, 2026)
GitHub stars: 42,081 for the whole wshobson/agents repo (Nov 2, 2026)
Last repo commit: Oct 30, 2026
What it includes: 2 files including SKILL.md: references/hypothesis-testing.md (no scripts)
Needs: A bug to investigate. The wording is built around parallel investigator agents, which the agent-teams plugin provides
Part of: The agent-teams plugin (version 1.0.4) in wshobson/agents. Counted from the plugin folder: 6 skills, 4 agents and 7 commands
Works in: Claude Code and Codex CLI have plugin routes in the README. The plugin README ties its commands to Claude Code's experimental Agent Teams feature

What it includes

6 failure categories with example causes
Logic error, data issue, state problem, integration failure, resource issue and environment, each with a short list of typical causes.
Try: “Debug this intermittent 500 on POST /users using competing hypotheses.”
Read · Hypothesis generation
Direct, correlational, testimonial and absence evidence
Rates each type as strong, medium or weak and requires citations as file:line.
Read · Evidence collection
High above 80%, medium 50 to 80%, low below 50%
Each level has criteria such as direct evidence and a clear causal chain.
Read · Confidence levels
Categorize, compare, pick a root cause, validate the fix
Results are Confirmed, Plausible, Falsified or Inconclusive. If several are equally likely it may be a compound issue.
Read · Result arbitration
5 checks before declaring the bug fixed
Includes that the original reproduction no longer fails and that tests are added or updated.
Read · Fix validation checklist
Task template, report template, decision tree, 4 error patterns
Patterns cover 500 errors, race conditions, works-locally-fails-in-production and regression after deploy.
Read · Hypothesis testing reference

From the official docs

Permissions

Low risk

The skill itself is guidance. The plugin route adds agents that can run shell commands, so review what they propose to run.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

The plugin route adds agents and commands
agent-teams has 4 agents and 7 commands. The team-debugger agent lists Bash in its tools. The skills-only routes install none of them. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Turning on read-only	No read-only mode. Use version control.
Sign-in and keys	SKILL.md and the reference file mention no account or key.
Scopes	The skill bundles no scripts The plugin adds a team-debugger agent whose tools include Read, Glob, Grep and Bash

Questions

What does the parallel-debugging skill do?

It gives Claude a structured way to debug when a bug has several possible causes. Claude lists competing hypotheses, collects cited evidence for each, and picks the best supported root cause.

How do I install it?

Use the install box at the top of this page. It lists the agent-teams plugin and two skills-only commands.

Is this skill from Anthropic?

No. It is in wshobson/agents, a community collection, inside the agent-teams plugin.

Do I need the agent-teams plugin?

Not to read the method, but the skill talks about investigators and arbitration. The plugin adds a team-debugger agent and a /team-debug command, and its README says the team commands need Claude Code's experimental Agent Teams feature, turned on with an environment variable.

What else is in the plugin?

6 skills, 4 agents and 7 commands, counted from the plugin folder. The repo tree shows no hooks or MCP config files there.

Does it work in the Claude app?

The install box has no Claude app upload for it, because the skill is written for agent teams in a coding tool. Use Claude Code or Codex.

Is it safe?

The skill bundles no scripts. If you use the plugin, its team-debugger agent lists Bash among its tools, so it can run commands in your project. Review the commands it proposes.

Information

updated Nov 2, 2026

Repository: wshobson/agents
Folder: plugins/agent-teams/skills/parallel-debugging

wshobson/agents skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 5

parallel-debugging SKILL.md · github.com
agent-teams README · github.com
wshobson/agents README · github.com
agent-teams plugin.json · github.com
hypothesis-testing.md · github.com