ccswarm: AI Multi-Agent Orchestration System

Version 0.4.0 - Rust Edition 2024

ccswarm is a workflow automation system that coordinates specialized AI agents using Claude Code CLI. It provides task delegation, template-based application generation, and Git worktree isolation for parallel development.

What ccswarm Does

• Task Delegation: Analyzes tasks and routes to specialized agents (Frontend, Backend, DevOps, QA)
• Auto-Create: Generates complete applications from natural language descriptions
• Git Worktree Isolation: Each agent works in isolated Git worktrees
• Template System: Predefined templates for common development tasks
• TUI Monitoring: Real-time terminal UI for task tracking
• Session Management: Agent session lifecycle management
• Resource Monitoring: Track agent resource usage

Quick Start

Prerequisites

• Rust (Edition 2024 compatible toolchain)
• Git 2.20+
• Claude Code CLI (claude command available)

Installation

git clone https://github.com/nwiizo/ccswarm.git
cd ccswarm
cargo build --release

Basic Usage

# Initialize a project
./target/release/ccswarm init --name "MyProject"

# Create a TODO app
./target/release/ccswarm auto-create "Create a simple TODO app" --output /tmp/todo-app

# List templates
./target/release/ccswarm template list

# Check system health
./target/release/ccswarm doctor

# Start TUI monitoring
./target/release/ccswarm tui

CLI Commands

Project Structure

ccswarm/
├── Cargo.toml              # Workspace configuration (edition = "2024")
├── crates/
│   ├── ccswarm/            # Main CLI and orchestration
│   │   ├── src/
│   │   │   ├── cli/        # CLI commands
│   │   │   ├── orchestrator/ # Task delegation
│   │   │   ├── agent/      # Agent management
│   │   │   ├── session/    # Session management
│   │   │   ├── template/   # Template system
│   │   │   ├── providers/  # Claude Code integration
│   │   │   ├── subagent/   # Parallel execution
│   │   │   ├── tui/        # Terminal UI
│   │   │   ├── workflow/   # Workflow engine
│   │   │   └── resource/   # Resource monitoring
│   │   └── tests/          # Integration & mockall tests
│   └── ai-session/         # Session management library
│       ├── src/
│       │   ├── core/       # Core session management
│       │   ├── context/    # Context compression
│       │   ├── coordination/ # Multi-agent coordination
│       │   └── security/   # Capability-based security
│       └── tests/
└── docs/                   # Documentation

Development

# Run all tests (242 lib tests + 37 mockall tests)
cargo test --workspace

# Run only library tests
cargo test --lib -p ccswarm

# Run mockall tests
cargo test --test mockall_tests -p ccswarm
cargo test --test mockall_tests -p ai-session

# Check code quality
cargo fmt && cargo clippy -- -D warnings

# Build release
cargo build --release --workspace

Testing with Mockall

The project uses mockall for mock-based testing. See tests/mockall_tests.rs for examples:

use mockall::mock;
use mockall::predicate::*;

mock! {
    pub MyService {
        fn execute(&self, input: &str) -> Result<String>;
    }
}

#[test]
fn test_with_mock() {
    let mut mock = MockMyService::new();
    mock.expect_execute()
        .times(1)
        .withf(|input| input.contains("test"))
        .returning(|_| Ok("result".into()));

    assert!(mock.execute("test input").is_ok());
}

Documentation

• Architecture - System design
• Application Spec - Feature details
• Development Standards - Coding guidelines

Limitations

• Claude Code Required: Auto-create requires Claude Code CLI (claude command)
• macOS/Linux Only: Windows is not supported due to Unix-specific dependencies

License

MIT License - see LICENSE for details.