Agent Editor Workshop

​Every change to your agent's instructions or tools means a code edit, a PR, and a deploy. That's fine when you're building — but once an agent is running, the people who need to iterate on it fastest are often the ones furthest from the codebase. The new Agent Editor in Mastra Studio moves agent configuration out of code and into a visual interface with draft/publish, versioning, and access control built in.

​What you'll learn

• ​How to set up the Editor with @mastra/editor and configure storage for version snapshots

• ​Adding and configuring tools from your project, MCP clients, and integration providers like Composio and Arcade

• ​Writing instructions inline or composing them from shared prompt blocks with {{variable}} interpolation

• ​Using the draft/publish workflow to save, compare, and roll back versions without redeploying

• ​Setting display conditions on tools and prompt blocks so they activate based on request context

• ​Controlling access with Studio Auth and RBAC so the right people can edit and only admins can publish

• ​Pinning versions per environment, running A/B tests, and accessing the full editor API programmatically

​The Editor doesn't replace code — developers still define agents in TypeScript as usual. What it does is give product teams and subject matter experts a way to tune behavior, test changes against datasets, and publish updates on their own. Your agent's configuration gets a full version history, and every endpoint accepts a version parameter so you can roll back or split traffic with a single line.

​If you've been looking for a way to let your whole team iterate on agent behavior without bottlenecking on deploys, this is the feature.

​Hosted by

• ​Alex Booker, Mastra

  • ​https://x.com/bookercodes

  • ​https://mastra.ai

• ​Daniel Lew, Mastra

  • ​https://x.com/ShlomesLew

  • ​https://mastra.ai

​Recording and code examples will be available to everyone who registers.