How does this skill handle AI tool failures?

Tools wrapped in the 'use step' directive are automatically retried, ensuring that transient API errors or infrastructure hiccups don't break the entire agent execution.

Does it support streaming responses?

Yes, it includes specialized API routes and transport layers that support both standard streaming and resumable streams for interrupted connections.

What is a durable AI workflow?

A durable workflow is a process that persists its state at every step, allowing it to survive server crashes, redeploys, or network interruptions and resume exactly where it left off.

Which AI models are supported?

The infrastructure is model-agnostic and supports major providers including OpenAI, Anthropic, and AWS Bedrock through the Vercel AI SDK and Gateway.

Can I monitor my agent's progress in real-time?

Yes, the skill integrates with an observability dashboard accessible via 'npx workflow web' which displays run statuses, step traces, and data flow.

Durable AI Agent Workflows

Name: Durable AI Agent Workflows
Author: mattwoodco

bymattwoodco

•

生产力与工作流

Builds resilient AI agent infrastructure with durable workflows that persist through failures, crashes, and redeploys.

The Workflow skill automates the setup of Workflow DevKit within Next.js projects, providing a robust foundation for building production-grade AI agents. It implements durable execution patterns where every LLM call and tool invocation is persisted and automatically retried upon failure. This skill is essential for developers creating complex, multi-step agentic workflows that require high reliability, observability via a local dashboard, and resumable streaming capabilities for seamless user experiences.

主要功能

01Automatic retries for AI tool invocations and LLM calls

02Durable state persistence for workflows and individual steps

03Extensible architecture using 'use workflow' and 'use step' directives

04Resumable streaming support for persistent chat sessions

053 GitHub stars

06Local observability dashboard for tracing and debugging runs

使用场景

01Building mission-critical AI agents that must survive infrastructure failures

02Creating reliable chat interfaces with built-in reconnection logic

03Implementing long-running background tasks like multi-stage content generation

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add mattwoodco/skills workflow

For use in Claude.ai and ChatGPT

Download Skill