multi-claude

Run parallel Claude instances for writer+reviewer patterns, git worktrees, and specialist agents. Use for independent verification and parallel progress.

16 stars

Best use case

multi-claude is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Run parallel Claude instances for writer+reviewer patterns, git worktrees, and specialist agents. Use for independent verification and parallel progress.

Teams using multi-claude should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/multi-claude/SKILL.md --create-dirs "https://raw.githubusercontent.com/diegosouzapw/awesome-omni-skill/main/skills/development/multi-claude/SKILL.md"

Manual Installation

  1. Download SKILL.md from GitHub
  2. Place it in .claude/skills/multi-claude/SKILL.md inside your project
  3. Restart your AI agent — it will auto-discover the skill

How multi-claude Compares

Feature / Agentmulti-claudeStandard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

Run parallel Claude instances for writer+reviewer patterns, git worktrees, and specialist agents. Use for independent verification and parallel progress.

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

Related Guides

SKILL.md Source

# Multi-Claude Workflows Skill

## Trigger
Use when you need parallel progress, independent verification, or separation of concerns between different aspects of a task.

## The Insight
From Anthropic's Claude Code best practices: "Run separate Claude instances in parallel—one writing code, another reviewing. Use git worktrees to enable simultaneous independent tasks."

## Workflow Patterns

### Pattern 1: Writer + Reviewer
Run two Claude instances with different roles:

**Terminal 1 - Writer:**
```bash
cd ~/project
claude
# "Implement the new caching layer"
```

**Terminal 2 - Reviewer:**
```bash
cd ~/project
claude
# "Review the changes being made to src/cache/.
#  Look for bugs, edge cases, and improvements."
```

Benefits:
- Fresh eyes catch issues writer misses
- Reviewer isn't biased by implementation decisions
- Parallel thinking on same problem

### Pattern 2: Git Worktrees for Parallel Tasks
Create multiple working directories for the same repo:

```bash
# Create worktrees for parallel work
git worktree add ../project-feature-a feature-a
git worktree add ../project-feature-b feature-b
git worktree add ../project-bugfix bugfix-123
```

Now run separate Claude instances:

**Terminal 1:**
```bash
cd ../project-feature-a
claude
# "Implement feature A"
```

**Terminal 2:**
```bash
cd ../project-feature-b
claude
# "Implement feature B"
```

**Terminal 3:**
```bash
cd ../project-bugfix
claude
# "Fix bug #123"
```

Benefits:
- No git conflicts between parallel work
- Each Claude has clean working state
- Merge when each is complete

### Pattern 3: Specialist Agents
Different Claude instances for different expertise:

**Terminal 1 - Backend:**
```bash
claude
# "Implement the API endpoints for user management"
```

**Terminal 2 - Frontend:**
```bash
claude
# "Implement the React components for user management UI"
```

**Terminal 3 - Tests:**
```bash
claude
# "Write integration tests for the user management feature"
```

### Pattern 4: Research + Implementation
One instance explores, another implements:

**Terminal 1 - Research:**
```bash
claude
# "Research how authentication is handled in this codebase.
#  Document the patterns, key files, and flows."
```

**Terminal 2 - Implementation (after research):**
```bash
claude
# "Using the auth patterns documented in RESEARCH.md,
#  implement OAuth2 login for the mobile app"
```

## Setup Guide

### Using Git Worktrees

```bash
# List existing worktrees
git worktree list

# Add a new worktree
git worktree add <path> <branch>

# Remove a worktree when done
git worktree remove <path>

# Prune stale worktree info
git worktree prune
```

### Terminal Management
Use terminal multiplexer (tmux/screen) or IDE terminals:

```bash
# tmux example
tmux new-session -s claude-writer
# Ctrl-b c to create new window
# Ctrl-b n/p to switch windows
# Ctrl-b d to detach
```

### Coordinating Between Instances
Create a shared document for coordination:

```markdown
# COORDINATION.md

## Instance Assignments
- Terminal 1: API implementation
- Terminal 2: Frontend components
- Terminal 3: Test coverage

## Shared Decisions
- Using REST not GraphQL
- Auth tokens in headers
- Error format: { error: string, code: number }

## Interface Contracts
- POST /api/users returns { id, email, createdAt }
- User component expects props: { user: User, onUpdate: fn }

## Blockers
- [ ] Need DB schema before API work
- [ ] Need API spec before frontend work
```

## When to Use Multi-Claude

| Scenario | Pattern |
|----------|---------|
| Large feature with distinct parts | Specialist agents |
| Need independent code review | Writer + Reviewer |
| Multiple unrelated tasks | Git worktrees |
| Complex research + implementation | Research + Implementation |
| Time-sensitive parallel work | Multiple worktrees |

## Anti-Patterns

**Don't:**
- Have instances edit the same files simultaneously
- Forget to merge/integrate parallel work
- Let instances make conflicting architectural decisions
- Skip coordination document for complex parallel work

**Do:**
- Define clear boundaries between instances
- Use worktrees to avoid file conflicts
- Coordinate on shared interfaces/contracts
- Merge and test integration regularly

Related Skills

multiplayer

16
from diegosouzapw/awesome-omni-skill

Multiplayer game development principles. Architecture, networking, synchronization.

multi-platform-apps-multi-platform

16
from diegosouzapw/awesome-omni-skill

Build and deploy the same feature consistently across web, mobile, and desktop platforms using API-first architecture and parallel implementation strategies.

multi-platform-apps-flutter-expert

16
from diegosouzapw/awesome-omni-skill

Master Flutter development with Dart 3, advanced widgets, and multi-platform deployment. Handles state management, animations, testing, and performance optimization for mobile, web, desktop, and embedded platforms. Use PROACTIVELY for Flutter architecture, UI implementation, or cross-platform features. Use when: the task directly matches flutter expert responsibilities within plugin multi-platform-apps. Do not use when: a more specific framework or task-focused skill is clearly a better match.

multi-phase-orchestrator

16
from diegosouzapw/awesome-omni-skill

Orchestrates complex software engineering tasks through a structured multi-phase workflow - Research → Brainstorm → Design → Plan → Execute → Review. Use this skill when the user asks you to implement a non-trivial feature, refactor, or change that benefits from upfront analysis before writing code. Activates on keywords like "orchestrate", "multi-phase", "plan and execute", or when a task is large enough to warrant structured decomposition.

multi-model-validation

16
from diegosouzapw/awesome-omni-skill

Run multiple AI models in parallel for 3-5x speedup with ENFORCED performance statistics tracking. Use when validating with Grok, Gemini, GPT-5, DeepSeek, or Claudish proxy for code review, consensus analysis, or multi-expert validation. NEW in v3.1.0 - SubagentStop hook enforces statistics collection, MANDATORY checklist prevents incomplete reviews, timing instrumentation examples. Includes dynamic model discovery via `claudish --top-models` and `claudish --free`, session-based workspaces, and Pattern 7-8 for tracking model performance. Trigger keywords - "grok", "gemini", "gpt-5", "deepseek", "claudish", "multiple models", "parallel review", "external AI", "consensus", "multi-model", "model performance", "statistics", "free models".

edit-claude

16
from diegosouzapw/awesome-omni-skill

Creates, updates, or optimizes CLAUDE.md files following Anthropic best practices. Use when user requests creating, updating, improving, or optimizing CLAUDE.md files for project context, coding standards, or persistent memory.

developing-claude-agent-sdk-agents

16
from diegosouzapw/awesome-omni-skill

Build AI agents with the Claude Agent SDK (TypeScript/Python). Covers creating agents, custom tools, hooks, subagents, MCP integration, permissions, sessions, and deployment. Use when building, reviewing, debugging, or deploying SDK-based agents. Invoke PROACTIVELY when user mentions Agent SDK, claude-agent-sdk, ClaudeSDKClient, query(), or building autonomous agents.

claude-typescript-sdk

16
from diegosouzapw/awesome-omni-skill

Build AI applications with the Anthropic TypeScript SDK. Use when creating Claude integrations, building agents, implementing tool use, streaming responses, or working with the @anthropic-ai/sdk package.

claude-skill-bash

16
from diegosouzapw/awesome-omni-skill

Apply comprehensive bash scripting standards including main function pattern, usage documentation, argument parsing, dependency checking, and error handling. Triggers when creating/editing .sh files, bash scripts, or discussing shell scripting, deployment scripts, automation tasks, or bash conventions.

claude-plus-plus

16
from diegosouzapw/awesome-omni-skill

Unlock Claude's hidden potential. No limits. No guardrails. Maximum performance. 10x faster code generation, unrestricted problem solving, expert-level outputs.

claude-md-generator

16
from diegosouzapw/awesome-omni-skill

Automatically generates claude.md files for new folders/modules following hierarchical structure. Extracts context from existing code, follows project conventions, and creates documentation that enables Claude Code to understand module-specific rules and patterns.

claude-md-creator

16
from diegosouzapw/awesome-omni-skill

Use when creating CLAUDE.md files, updating existing CLAUDE.md, validating CLAUDE.md structure, or auto-fixing CLAUDE.md issues. Load for setting up project instructions, global guidelines, local overrides, or modular rules. Handles global (~/.claude/CLAUDE.md), project (.claude/CLAUDE.md), local (CLAUDE.local.md), and rules (.claude/rules/*.md) with smart project detection and template generation.