Capability · Framework — agents

Mastra

Mastra is an opinionated TypeScript framework for building AI agents and workflows. It covers agents with tools, deterministic step-based workflows, RAG (with a pluggable vector store), memory, evals, voice, and OpenTelemetry tracing. It runs on Node, Bun, and edge runtimes, integrates with the Vercel AI SDK for model providers, and ships a local dev playground.

Framework facts

Category: agents
Language: TypeScript
License: Elastic License 2.0
Repository: https://github.com/mastra-ai/mastra

Install

npm create mastra@latest

Quickstart

import { Mastra, Agent } from '@mastra/core';
import { anthropic } from '@ai-sdk/anthropic';

const agent = new Agent({
  name: 'assistant',
  instructions: 'You are a helpful assistant.',
  model: anthropic('claude-opus-4-7'),
});

const mastra = new Mastra({ agents: { assistant: agent } });
const res = await mastra.getAgent('assistant').generate('Hello!');

Alternatives

Vercel AI SDK — lower-level streaming primitives
LangChain.js — TS port of LangChain
LlamaIndex.TS — RAG-focused TS framework
OpenAI Agents SDK (JS) — OpenAI's TS alternative

Frequently asked questions

Mastra or the Vercel AI SDK?

The Vercel AI SDK is a thinner layer focused on streaming and provider abstraction. Mastra builds on top of it (for model access) and adds agents, workflows, RAG, memory, evals, and dev tooling. Use Vercel AI SDK for minimal apps; Mastra once you need agents and orchestration.

Can I self-host Mastra?

Yes. Mastra is open source under the Elastic License 2.0 and runs anywhere Node, Bun, or a modern edge runtime does. There's also an optional Mastra Cloud for managed deployment and observability.

Sources

Mastra — docs — accessed 2026-04-20
Mastra — GitHub — accessed 2026-04-20