Temporal Workflow Scaffolder

Temporal is a durable execution platform for long-running workflows. It provides fault-tolerant, resumable workflows where state is automatically persisted. Use Temporal when you need: multi-step processes that take seconds to days (order processing, user onboarding), reliable distributed transactions with rollback/compensation, cron-like scheduled workflows, human-in-the-loop approval flows, or any workflow where reliability and observability are critical.

Workflows are deterministic orchestration code — they define the sequence of steps, handle retries, signals, and timers. They must be deterministic (no I/O, random numbers, or current time) because Temporal replays them to recover state. Activities are the actual work — calling external APIs, writing to databases, sending emails. Activities can do any I/O and are retried automatically on failure.

A Task Queue is a routing mechanism that matches Workflow and Activity tasks to Workers. Workers poll specific task queues and execute tasks. Using separate task queues for different services (e.g. 'payment-processing' vs 'notification-service') allows independent scaling and deployment of worker pools.

Temporal retries Activities automatically on failure with configurable policies: maximumAttempts limits total tries. initialInterval is the first retry delay. backoffCoefficient multiplies the delay each retry (exponential backoff). maximumInterval caps the delay. Workflow-level retries (for the entire workflow run) are separate from activity-level retries. The default retry policy retries indefinitely with exponential backoff up to 100 seconds.

Signals are asynchronous messages sent to a running workflow to change its state or trigger behavior — like cancelling an order or approving a request. The workflow handles signals via setHandler (TS/Go) or signal handlers (Python). Queries are synchronous read-only requests to inspect a workflow's current state — like asking 'what step is this order on?' without affecting execution. Queries are answered immediately from the workflow's in-memory state.

No. All code generation runs entirely in your browser using JavaScript. Your workflow names, activity names, and configuration never leave your machine.