Workflow Orchestrator Instant

Design and execute complex multi-step business workflows with state management, error recovery, and parallel processing.

When to Use This Command

Run this command when you need to:

• Orchestrate a multi-step process with dependencies, decision points, and parallel branches
• Implement reliable workflow execution with automatic retry, compensation, and rollback logic
• Model business processes as state machines with auditable transition histories

Consider alternatives when:

• Your task is a simple linear script with no branching or error recovery needs
• You need a persistent workflow engine like Temporal or Camunda (this command generates designs, not runtime infrastructure)

Quick Start

Configuration

Copy
name: workflow-orchestrator-instant
type: command
category: automation

Example Invocation

Copy
claude command:run workflow-orchestrator-instant --process order-fulfillment --mode design

Example Output

[Analysis] Process: order-fulfillment
  States identified: 8 (pending, validated, paid, picking, packed, shipped, delivered, cancelled)
  Decision points: 3 (payment check, inventory check, shipping method)
  Parallel branches: 1 (notification + inventory update)

[Design] State machine generated:
  Transitions: 12 defined
  Compensation flows: 4 rollback paths
  Error boundaries: 3 retry policies

[Output] Files created:
  - workflows/order-fulfillment.state-machine.json
  - workflows/order-fulfillment.diagram.md
  - workflows/order-fulfillment.error-handlers.ts

Core Concepts

Orchestration Overview

Aspect	Details
Patterns	Sequential, parallel split/join, exclusive choice, event gateway
State Management	Persistent state with version control and migration support
Error Handling	Retry with exponential backoff, circuit breaking, dead letter queues
Transactions	Saga pattern with compensation logic for distributed operations

Orchestration Workflow

[Process Requirements]
        |
[Model States + Transitions]
        |
[Define Decision Points]
   /         \
[Parallel]  [Sequential]
   \         /
[Error Boundaries + Compensation]
        |
[Monitoring + Audit Trail]
        |
[Deploy Workflow Definition]

Configuration

Parameter	Type	Default	Description
process	string	required	Name of the business process to orchestrate
mode	string	design	Operation mode: design, validate, or execute
max-retries	number	3	Maximum retry attempts for failed steps
timeout	string	30s	Default timeout per workflow step
audit	boolean	true	Enable full audit trail logging for state transitions

Best Practices

1. Start with Process Mapping - Before defining states and transitions, map the entire business process on paper. Identify every decision point, parallel branch, and failure scenario upfront to avoid rework.

2. Design Compensation Before Happy Path - For every state transition, define the compensating action that undoes it. Reliable workflows require rollback paths to be first-class citizens, not afterthoughts.

3. Use Idempotent Operations - Every workflow step should produce the same result when executed multiple times. This is essential for safe retries and recovery after partial failures.

4. Keep State Transitions Atomic - Each transition should move the workflow from one consistent state to another. Avoid transitions that leave the system in an intermediate or ambiguous state.

5. Monitor with Business Metrics - Track not just technical metrics (latency, error rate) but business KPIs (orders processed per minute, SLA compliance). These reveal workflow health from the user perspective.

Common Issues

1. State Inconsistency After Crash - If the orchestrator crashes mid-transition, state may be inconsistent. Use write-ahead logging or event sourcing to reconstruct the last known good state on restart.

2. Deadlock in Parallel Branches - Parallel branches that compete for the same resource can deadlock. Use resource locking with timeouts and design branches to acquire locks in a consistent order.

3. Compensation Chain Failure - When a compensation step itself fails, the workflow enters an unrecoverable state. Implement a dead letter queue to capture these cases for manual intervention.