The type-safe, event-driven workflow orchestration library that scales with your business.
Build robust, distributed workflows in Go with compile-time safety, automatic retries, and horizontal scaling out of the box.
// Define your business logic as type-safe state machines b := workflow.NewBuilder[Order, OrderStatus]("order-processing") b.AddStep(OrderCreated, ProcessPayment, PaymentProcessed) b.AddStep(PaymentProcessed, FulfillOrder, OrderCompleted) wf := b.Build(kafkaStreamer, sqlStore, roleScheduler)Unlike other orchestrators, Workflow leverages Go generics for compile-time guarantees. Catch errors before deployment, not in production.
// Your IDE knows exactly what data flows where func processPayment(ctx context.Context, r *workflow.Run[Order, OrderStatus]) (OrderStatus, error) { // r.Object is typed as *Order, OrderStatus is your enum // Compiler catches mismatches before they cause runtime errors }Built for modern distributed systems. Steps communicate through durable events, enabling:
- Loose coupling between workflow components
- Automatic retries with exponential backoff
- Horizontal scaling across multiple instances
- Fault tolerance that survives network partitions
Your choice of database, message queue, and coordination service. Start simple, scale when needed:
// Development: Everything in-memory wf := b.Build(memstreamer.New(), memrecordstore.New(), memrolescheduler.New()) // Production: Battle-tested infrastructure wf := b.Build(kafkastreamer.New(), sqlstore.New(), rinkrolescheduler.New())Production-ready monitoring without the setup overhead:
- Prometheus metrics for throughput, latency, and error rates
- Web UI for real-time workflow visualization
- Structured logging with correlation IDs
- Distributed tracing support
- Order Processing: Payment, inventory, fulfillment pipelines
- User Onboarding: Multi-step verification and activation flows
- Financial Operations: Transaction processing with compliance checks
- Data Processing: ETL pipelines with validation and cleanup
- Approval Workflows: Multi-stakeholder review processes
| Feature | Workflow | Temporal | Zeebe/Camunda |
|---|---|---|---|
| Type Safety | ✅ Compile-time (Go generics) | ❌ Runtime validation | ❌ Runtime (BPMN) |
| Architecture | ✅ Event-driven state machines | ||
| Infrastructure | ✅ Your choice (adapters) | ❌ Requires Temporal cluster | ❌ Requires external engine |
| Deployment | ✅ Library in your app | ❌ Separate server/workers | ❌ Separate engine |
| Learning Curve | ✅ Native Go patterns | ❌ BPMN modeling | |
| Language | ✅ Go-native |
go get github.com/luno/workflowpackage main import ( "context" "fmt" "github.com/luno/workflow" "github.com/luno/workflow/adapters/memstreamer" "github.com/luno/workflow/adapters/memrecordstore" "github.com/luno/workflow/adapters/memrolescheduler" ) type TaskStatus int const ( TaskStatusUnknown TaskStatus = 0 TaskStatusCreated TaskStatus = 1 TaskStatusProcessed TaskStatus = 2 TaskStatusCompleted TaskStatus = 3 ) type Task struct { ID string Name string } func main() { b := workflow.NewBuilder[Task, TaskStatus]("task-processor") b.AddStep(TaskStatusCreated, func(ctx context.Context, r *workflow.Run[Task, TaskStatus]) (TaskStatus, error) { fmt.Printf("Processing: %s\n", r.Object.Name) return TaskStatusProcessed, nil }, TaskStatusProcessed) b.AddStep(TaskStatusProcessed, func(ctx context.Context, r *workflow.Run[Task, TaskStatus]) (TaskStatus, error) { fmt.Printf("Completed: %s\n", r.Object.Name) return TaskStatusCompleted, nil }, TaskStatusCompleted) wf := b.Build(memstreamer.New(), memrecordstore.New(), memrolescheduler.New()) ctx := context.Background() wf.Run(ctx) defer wf.Stop() // Trigger a workflow runID, _ := wf.Trigger(ctx, "task-1", workflow.WithInitialValue(&Task{ ID: "task-1", Name: "Process Invoice", })) // Wait for completion wf.Await(ctx, "task-1", runID, TaskStatusCompleted) fmt.Println("✅ Workflow completed!") }Workflow provides enterprise-grade features:
- ✅ Exactly-once processing guarantees via transactional outbox pattern
- ✅ Built-in error handling with pause and retry mechanisms
- ✅ Comprehensive observability via Prometheus metrics and Web UI
- ✅ Horizontal scaling through role-based scheduling
- ✅ Infrastructure flexibility via pluggable adapters
- ✅ Production deployment patterns for various scales
| Topic | Description |
|---|---|
| Getting Started | Install and build your first workflow |
| Core Concepts | Understand Runs, Events, and State Machines |
| Architecture | Deep dive into system design and components |
| Steps | Build workflow logic with step functions |
| Adapters | Infrastructure integration guide |
| Database Setup | Complete MariaDB/MySQL & PostgreSQL setup guide |
| Callbacks | Handle external events and webhooks |
| Timeouts | Add time-based operations |
| Connectors | Integrate with external event streams |
| Hooks | React to workflow lifecycle changes |
| Configuration | Tune performance and behaviour |
| Monitoring | Observability and debugging |
| Example | Description |
|---|---|
| Getting Started | Your first workflow in 5 minutes |
| SQL Example | Complete MariaDB/MySQL integration with Docker |
| Order Processing | Complex e-commerce workflow with payments & fulfilment |
| Callbacks | Handle external webhooks and events |
| Timeouts | Schedule and timeout operations |
| Connectors | Integrate with external event streams |
- 📚 Documentation - Comprehensive guides and examples
- 🐛 Issues - Bug reports and feature requests
- 💬 Discussions - Community Q&A
go get github.com/luno/workflow # Production adapters (install as needed) go get github.com/luno/workflow/adapters/kafkastreamer go get github.com/luno/workflow/adapters/sqlstore go get github.com/luno/workflow/adapters/rinkrolescheduler go get github.com/luno/workflow/adapters/webuiReady to build reliable workflows? Get started in 5 minutes →