Mastra
An open-source TypeScript framework for building AI agents, workflows, RAG, and tool-calling, with memory, model routing, and built-in observability.
Mastra is an open-source TypeScript framework for building AI agents, workflows, RAG, and tool-calling, from the team behind Gatsby. It targets JavaScript/TypeScript engineers who want typed agents, graph-based workflows, memory, and observability in one stack instead of dropping to Python.
Mastra is an open-source TypeScript framework for building AI agents, workflows, RAG, and tool-calling, from the team behind Gatsby (Sam Bhagwat and co-founders). It packages the primitives an agentic app needs — typed agents, a graph-based workflow engine, agent memory, model routing, and observability — into one modular framework so you stay in TypeScript instead of reaching for a separate Python stack.
It is aimed at JavaScript/TypeScript engineers building production agents and AI features: you define an agent with instructions, a model, and tools in one place, then compose deterministic multi-step workflows around it. That makes it a natural fit when your app, your tooling, and your AI logic all live in the same Node/TypeScript codebase.
Highlights
- Typed agents — declare instructions, model, tools, and runtime behavior in a single typed object; the agent reasons over tools and iterates until it returns a final answer.
- Graph-based workflows — orchestrate multi-step processes with explicit control flow via
.then(),.branch(), and.parallel(), separate from open-ended agent loops. - Memory and RAG — built-in conversation history plus semantic/retrieval memory for grounding agents in your own data.
- Model routing and observability — connect to many model providers through one interface, with built-in scorers and tracing to measure and refine agent behavior.
In an AI-assisted workflow
Scaffold a project, define an agent, and call it from your TypeScript app. See where it sits among peers in Agent Frameworks in 2026.
npm create mastra@latest
# then, in code:
# import { Agent } from "@mastra/core/agent";
# const agent = new Agent({ name: "support", instructions: "...", model, tools });
# const res = await agent.generate("Summarize this ticket");TIP
Reach for Mastra when your stack is already TypeScript and you want agents, workflows, memory, and tracing in one framework — rather than splitting agent logic into a separate Python service.
Good to know
Mastra's core framework is open source under Apache-2.0 and free to self-host; code under ee/ directories is source-available under the Mastra Enterprise License, and the hosted Mastra Platform is a separate commercial offering with a free Starter tier and paid usage-based tiers. Confirm current pricing and license terms on the official site. For broader framework trade-offs, see LangGraph vs. CrewAI.
Frequently asked questions
- What is Mastra?
- Mastra is an open-source TypeScript framework for building AI agents, workflows, RAG, and tool-calling, built by the team behind Gatsby. It bundles typed agents, a graph-based workflow engine, agent memory, model routing across many providers, and built-in observability so JavaScript/TypeScript engineers can build agentic apps without dropping to Python.
- Is Mastra free?
- Mastra's core framework is open source under Apache-2.0 and free to self-host (deploy to Node.js, Vercel, Cloudflare, Netlify, or your own infra). The optional Mastra Platform is a commercial hosted offering with a free Starter tier and paid usage-metered tiers — confirm current pricing on the official site.
- How does Mastra compare to LangGraph?
- Both build controllable, multi-step agents, but Mastra is TypeScript-native and ships agents, workflows, memory, and observability as one framework, while LangGraph is a Python-first (with JS port) low-level graph library focused on explicit state machines. If your stack is Node/TypeScript, Mastra keeps you in one language; LangGraph is the more common choice in Python shops.
Related
- Which Agent Framework in 2026? LangGraph vs CrewAI vs AutoGen vs OpenAI Agents SDK vs Claude Agent SDKA decision guide to the major AI agent frameworks — control vs. abstraction, multi-agent models, state and durability, and which fits your project.
- LangGraph vs CrewAI: Agent Frameworks Compared (2026)LangGraph vs CrewAI — explicit state-machine control vs role-based crew abstractions. Which agent framework fits your reliability bar and team.
- Building Multi-Step Agent WorkflowsPatterns for building multi-step agent workflows in Claude Code: decompose tasks, fan-out to parallel subagents, verify every step, and orchestrate.
- Multi-Agent OrchestrationFour patterns for coordinating multiple agents — fan-out, pipeline, orchestrator-worker, and verify/critic — and when each earns its overhead.
- Production Tool & Function Calling: Feed Errors Back as ObservationsHow agents use tools — the call/observe/retry loop, why errors must return to the model, and the schemas, idempotency, and limits that keep it reliable.