claude-squad

🥇Gold

Claude Squad enables operations teams to manage multiple AI terminal agents like Claude Code, Aider, Codex, OpenCode, and Amp. It streamlines workflows by integrating these agents into a single interface, improving efficiency in code generation, debugging, and automation tasks.

6,9263980Updated Yesterday

Intermediate30min to implementautomation

Saves ~20 min per use

Quick InstallView Source

git clone https://github.com/smtg-ai/claude-squad.git

Works with:

Claude

Overview

About This Skill

https://smtg-ai.github.io/claude-squad/

How to Use

[{"step":"Install and configure Claude Squad","action":"Run `pip install claude-squad` and initialize the terminal agents using the provided setup script. Configure API keys for GitHub, Stripe, and other services your workflow requires.","tip":"Use the `claude-squad --init` command to generate a default configuration file with agent templates. Customize the `agents.json` file to define roles, permissions, and fallback behaviors."},{"step":"Define your workflow scope","action":"Create a new workflow file (e.g., `workflows/bug_fix_workflow.json`) specifying the task name, agents involved, and success criteria. Include error handling rules and dependency chains between agents.","tip":"Start with a simple 2-agent workflow (e.g., Claude Code + Aider) before scaling to 3+ agents. Use the `claude-squad validate` command to check your workflow syntax."},{"step":"Execute the workflow","action":"Run `claude-squad run --workflow bug_fix_workflow.json --repo acme-inc/ecommerce-api`. Monitor progress in real-time via the CLI dashboard or web interface.","tip":"Use the `--debug` flag to log detailed agent interactions. For long-running workflows, set up notifications via Slack or email using the webhook integration."},{"step":"Review and iterate","action":"After execution, review the generated artifacts (PRs, test files, logs) in the `output/` directory. Use the `claude-squad analyze` command to generate a performance report and identify bottlenecks.","tip":"Compare execution times across workflow runs to optimize agent handoffs. Adjust agent priorities (e.g., `priority: high` for critical tasks) to improve efficiency."},{"step":"Scale to team workflows","action":"Create shared workflow templates in a team repository. Use the `claude-squad deploy` command to push workflows to a central server for team access.","tip":"Tag workflows with version numbers (e.g., `v1.2`) and maintain a changelog in your team's documentation. Use the `--parallel` flag to run multiple workflows simultaneously."}]

Use Cases

Simultaneously manage multiple coding tasks using different AI agents without conflicts.

Automatically accept prompts in the background to speed up development workflows.

Review and apply code changes from various agents in a single terminal window.

Create isolated git workspaces for each AI session to prevent code conflicts.

Setup & Installation

Quick Install

No install command available. Check the GitHub repository for manual installation instructions.

Alternative Install (Git Clone)

git clone https://github.com/smtg-ai/claude-squad

Requirements

Claude Code or compatible AI agent
Works with: Claude

Quick Start Guide

Install the Skill

Copy the install command above and run it in your terminal.

Open Your AI Agent

Launch Claude Code, Cursor, or your preferred AI coding agent.

Try It Out

Use the prompt template or examples below to test the skill.

Customize

Adapt the skill to your specific use case and workflow.

Usage Examples

Prompt Template

Set up a [TASK_NAME] workflow in Claude Squad using [TERMINAL_AGENT_1], [TERMINAL_AGENT_2], and [TERMINAL_AGENT_3]. Define the handoff points between agents, error handling for [COMMON_ERROR], and success criteria for [OUTPUT_DELIVERABLE]. Provide a step-by-step execution plan with estimated time per step.

Example Output

### Claude Squad Workflow Execution Report

**Workflow:** Automated Bug Fix & Test Suite Generation
**Agents Assigned:**
- Agent 1: Claude Code (Primary orchestrator)
- Agent 2: Aider (Code modification specialist)
- Agent 3: OpenCode (Test suite generator)

**Execution Timeline:**
1. **00:00-00:05** - Agent 1 (Claude Code) clones the repository from GitHub (repo: `acme-inc/ecommerce-api`) and identifies the top 3 failing tests in `test_payment_gateway.py`.
2. **00:05-00:15** - Agent 2 (Aider) implements fixes for the failing tests: 
   - Fixed `stripe_charge()` method to handle null `customer_id` (PR #1243)
   - Added retry logic for network timeouts in `process_refund()` (PR #1244)
   - Resolved race condition in `update_inventory()` (PR #1245)
3. **00:15-00:25** - Agent 3 (OpenCode) generates a new test suite (`test_integration_suite.py`) with 15 edge cases covering:
   - Concurrent payment attempts
   - Database transaction rollbacks
   - API rate limiting
4. **00:25-00:30** - Agent 1 runs the full test suite (87 tests) and confirms 100% pass rate. All agents log their actions to `claude-squad.log` with timestamps.

**Outputs Generated:**
- 3 pull requests (linked in GitHub)
- 1 new test file (120 lines)
- Updated documentation in `docs/api_reference.md`

**Error Handling:**
- If Agent 2 fails to apply a fix, it automatically falls back to Agent 1 to generate a manual patch.
- If test suite generation fails, Agent 3 triggers a rerun with reduced test cases.

**Success Metrics:**
- All tests passing (100%)
- Code coverage increased from 78% to 89%
- Zero regressions in existing functionality

**Next Steps:**
- Merge PRs after code review
- Schedule performance testing in staging environment