Mastra is an open-source TypeScript framework designed for developers who want to build AI agents and workflows in the same language as the rest of their stack. Instead of stitching together a Python agent library plus a separate orchestration layer, teams define typed agents, workflows, and tools directly in TypeScript. Core building blocks include Agents (instructions, models, tools, runtime behavior), Workflows (typed steps, retries, branching), a Harness for multi-mode agents, and Memory that adds durable context through observational memory and semantic recall.
Beyond the framework, Mastra ships an opinionated observability layer: traces, metrics, logs, evaluations, datasets, and scorers, so you can see exactly what an agent did on every run and measure quality before shipping changes. Agents can register on a single Mastra server that runs locally and deploys to Mastra Cloud, and integrations cover major model providers plus MCP servers for tool use.
The project is Apache 2.0 open source and can be self-hosted, while Mastra Cloud adds managed deployment, collaboration, and enterprise controls such as RBAC, SSO, and VPC deployment. Pricing follows a freemium model with a free open-source path and contact-for-pricing cloud plans.
