workflow-best-practices

Orchestrate durable workflows with use workflow, use step, and startStream/finishStream.

20|3|Updated Nov 30, 2025
One-click install
npx skills add https://github.com/andrelandgraf/fullstackrecipes --skill workflow-best-practices-andrelandgraf
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: workflow-best-practices
Source: https://github.com/andrelandgraf/fullstackrecipes/tree/main/templates/fullstackrecipe/.agents/skills/workflow-best-practices
Command: npx skills add https://github.com/andrelandgraf/fullstackrecipes --skill workflow-best-practices-andrelandgraf

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Coordinating long-running, stateful workflows with reliable steps, streaming outputs, and persistence can be complex and error-prone. This guide outlines patterns to build durable workflows using the Workflow Development Kit, enabling authoring, starting, resuming, and persisting workflow runs.

Core Features & Use Cases

  • Orchestration with "use workflow" at the entry point and "use step" for durable checkpoints.
  • Streaming and lifecycle management using startStream and finishStream to provide real-time feedback.
  • Persisting results, resuming in-progress runs, and recovering from failures for robust, production-grade workflows.

Quick Start

Create a minimal workflow under src/workflows, wrap the orchestration with startStream and finishStream, and demonstrate persistence across a simulated restart.

Frequently Asked Questions about workflow-best-practices

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build durable workflows with persistence across application restarts?▼

Durable workflows are built by orchestrating steps with 'use workflow' and 'use step' to create checkpoints, ensuring state persistence and automatic recovery across application restarts for long-running tasks.

What's the best way to stream real-time output from long-running agent runs?▼

Streaming real-time output from agent runs is managed using the startStream and finishStream lifecycle methods, providing immediate feedback and observable execution throughout the workflow's duration.

How do I resume in-progress workflow runs after a failure?▼

Resuming in-progress workflow runs after a failure relies on robust persistence patterns and durable step checkpoints, allowing the workflow to recover its state and continue execution seamlessly.

When do I need explicit 'use workflow' and 'use step' checkpoints for task orchestration?▼

Explicit 'use workflow' and 'use step' checkpoints are needed when orchestrating stateful, multi-step tasks that require durable persistence, reliable error-handling, and recovery capabilities in modern full-stack apps.

Does the Workflow Development Kit support error handling for complex orchestration steps?▼

Yes, the Workflow Development Kit supports robust error-handling patterns for complex orchestration steps, ensuring production-grade reliability when persisting results and recovering from failures during execution.

Can I use this workflow orchestration pattern for long-running tasks in full-stack apps?▼

Yes, this workflow orchestration pattern is designed for long-running tasks in modern full-stack apps, coordinating durable execution, streaming outputs, and state persistence across simulated or actual restarts.