Build crash-safe AI agents on AgentKit and step.ai — durable tool calls, human-in-the-loop approval, realtime progress, and when not to reach for an agent loop at all.
VoltAgent Best PracticesSkill
Summary
VoltAgent's official conventions skill — when to reach for an agent versus a workflow, the expected project layout, and the core snippets for agents, workflows, memory and servers.
Features
- Agent vs workflow decision table — adaptive reasoning against explicit control flow
- Canonical src/ layout with agents/, tools/ and workflows/
- Agent construction with the provider/model string format spelled out
- createWorkflowChain pipelines typed with Zod input and result schemas
- VoltAgent bootstrap registering agents and workflows behind honoServer()
- Memory, server and observability conventions in one reference
Install This Skill
Add this skill to your favorite AI agent in a few steps.
Skill Content
Usage Instructions
Learn how to use this skill with different AI agents.
Description
VoltAgent is a TypeScript framework for building AI agents and multi-step workflows, with its own memory, server and observability pieces. This is the vendor's own conventions skill: a compact reference an agent loads before writing VoltAgent code, so what it produces looks like the framework's documentation rather than an approximation of it.
The decision it makes first
Agent or workflow. An Agent is for open-ended tasks where the model picks tools and reasons adaptively; a workflow is for multi-step pipelines with explicit control flow and suspend/resume. Getting this backwards produces either an agent asked to follow a rigid script or a workflow trying to improvise, and it is the single most common structural mistake in a new VoltAgent project.
Layout and snippets
The expected shape is src/index.ts with agents/, tools/ and workflows/ beside it. From there the skill carries the canonical snippets: constructing an Agent from @voltagent/core with name, instructions and a provider/model string (openai/gpt-4o-mini, anthropic/claude-3-5-sonnet — the slash format is a detail models routinely get wrong); building a typed pipeline with createWorkflowChain and Zod schemas for input and result; and the VoltAgent bootstrap that registers agents and workflows behind honoServer().
The rest covers memory, server configuration and observability conventions.
Where it fits
It is one of four skills in VoltAgent's official pack, alongside create-voltagent for project setup, voltagent-core-reference for the VoltAgent class options and lifecycle methods, and voltagent-docs-bundle, which reads the version-matched docs embedded in @voltagent/core so answers track the version you actually installed rather than whatever the model remembers.
MIT licensed. Install with npx skills add VoltAgent/skills, or clone the repository and point your agent at the directory.
Related Skills
Inngest's durable-execution playbook as an agent skill — steps and memoisation, event and cron triggers, idempotency, cancellation, retries and non-retriable errors.
Skill: Inngest Flow Control
by Inngest
Seven ways to throttle agent and background workloads — concurrency keys, throttling, rate limiting, debounce, singleton, priority and batching — with a decision guide that picks the right one.
Skill: AWS Lambda Durable Functions
by AWS Labs
Build multi-step Lambda workflows that survive interruption and run for up to a year — the replay model, step operations, wait/callback patterns, saga-style compensation and local testing.