Open-Source Visual Workflow Designer for AI Agent Systems

Musashi is designed specifically for AI workflow design without execution, focusing on:

• 🎯 Design-First Approach - Pure workflow design tool without runtime overhead
• 🚀 Lightweight & Fast - Minimal dependencies, optimized Docker image (~150MB)
• 🔧 Framework Agnostic - Export workflows as JSON for any execution engine
• 🌐 Open Source - MIT licensed, community-driven development

• 🖱️ Drag & Drop Interface - Intuitive node-based editor powered by React Flow
• 📐 Auto Layout - Automatic graph alignment using Dagre algorithm
• 🔌 Rich Node Types - Agent, Function, MCP Server, User Input, Output nodes
• 🔗 Smart Connections - Type-safe input/output matching with validation

• 👥 Multi-User Support - Real-time collaboration with conflict resolution
• 🔐 Role-Based Access - Admin, Editor, Viewer permissions per workspace
• 📊 Workflow History - Track changes with detailed update logs
• 🔄 Version Control - Optimistic locking prevents conflicting edits

• 📦 JSON Export/Import - Git-friendly workflow format
• 🔍 TypeScript Support - Full type safety in frontend
• 📝 OpenAPI Docs - Auto-generated API documentation
• 🐳 Docker First - Production-ready containerized deployment

• 🔒 JWT Authentication - Secure token-based access control
• ✅ Container Signing - Cosign-verified Docker images
• 🛡️ Security Scanning - Automated Trivy vulnerability checks
• 📋 SBOM Generation - Software bill of materials included

┌─────────────────────────────────────────────────────┐
│  Musashi Workflow Editor                      ⚙️ 🔍 │
├─────────────────────────────────────────────────────┤
│  ┌──────────┐    ┌──────────┐    ┌──────────┐     │
│  │   User   │───▶│  Agent   │───▶│  Output  │     │
│  │  Input   │    │   GPT-4  │    │  Result  │     │
│  └──────────┘    └──────────┘    └──────────┘     │
│        │              │                │           │
│        └──────────────┼────────────────┘           │
│                       ▼                            │
│                 ┌──────────┐                       │
│                 │ Function │                       │
│                 │  Process │                       │
│                 └──────────┘                       │
└─────────────────────────────────────────────────────┘

# Pull the latest image from GitHub Container Registry
docker pull ghcr.io/iml1111/musashi:latest

# Run with a single command (requires external MongoDB)
docker run -d \
  --name musashi \
  --restart unless-stopped \
  -p 80:8080 \
  -e MONGODB_URL="mongodb://host.docker.internal:27017" \
  -e DATABASE_NAME="musashi" \
  -e SECRET_KEY="$(openssl rand -hex 32)" \
  --add-host host.docker.internal:host-gateway \
  ghcr.io/iml1111/musashi:latest

# Access in browser
open http://localhost

Note:

• The image path follows lowercase convention: ghcr.io/iml1111/musashi
• For first-time publishing, set the package to Public in GitHub Packages for anonymous pull
• The image includes OCI labels like org.opencontainers.image.source linking back to this repository

# Download and run with one command
curl -O https://raw.githubusercontent.com/iml1111/musashi/main/docker-compose.yml
echo "SECRET_KEY=$(openssl rand -hex 32)" > .env
docker-compose up -d

# Access at http://localhost

For detailed installation options, see INSTALL.md

When Musashi starts for the first time, it automatically creates an admin account:

Security Notice:

• The admin account is created automatically on first startup
• For production deployments, immediately change the default password
• Create individual user accounts for team members
• Never expose the default credentials in production environments

For complete environment variable reference and examples, see INSTALL.md

Note: The container exposes only port 8080, which serves both the frontend and proxies API requests to the backend internally.

# Health check
curl http://localhost/api/v1/health
# Response: {"status": "healthy", "api": "v1"}

# API health check (via nginx proxy)
curl http://localhost:8080/api/v1/health
# Response: {"status": "healthy", "api": "v1"}

# Docker health check status
docker inspect musashi --format='{{.State.Health.Status}}'
# Response: healthy

# Pull new image and restart
docker-compose pull
docker-compose up -d

# Test MongoDB connection
docker exec musashi python -c "
from pymongo import MongoClient
client = MongoClient('mongodb://host.docker.internal:27017')
print(client.server_info()['version'])
"

# Solutions
# 1. Check if MongoDB is running
docker ps | grep mongo

# 2. Check network settings (macOS/Windows)
--add-host host.docker.internal:host-gateway

Automatically created on first run:

• Username: admin
• Password: changeme123! (immediate change recommended)

# Map to different host port
docker run -d --name musashi -p 8888:8080 ghcr.io/iml1111/musashi:latest

We recommend using a reverse proxy:

server {
    listen 443 ssl;
    server_name musashi.example.com;
    
    ssl_certificate /path/to/cert.pem;
    ssl_certificate_key /path/to/key.pem;
    
    location / {
        proxy_pass http://localhost:80;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
    }
}

See INSTALL.md for backup and restore procedures.

Supports both AMD64 and ARM64 architectures:

• linux/amd64: Intel/AMD processors
• linux/arm64: Apple Silicon (M1/M2), ARM servers

# Real-time logs
docker logs -f musashi

# Last 100 lines
docker logs --tail 100 musashi

# Since specific time
docker logs --since 2h musashi

• 📖 Installation Guide - Detailed installation and setup instructions
• 🔐 Security Policy - Security vulnerability reporting and policies
• 🤝 Contributing Guide - How to contribute to the project
• 📝 Changelog - Version history and changes
• 📋 API Documentation - OpenAPI/Swagger documentation

Musashi is an open-source project. Contributions are welcome!

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

For more details, see CONTRIBUTING.md.

This project is distributed under the MIT License. For more details, see the LICENSE file.

• GitHub Repository
• Container Image (GHCR)
• Issue Tracker