Mastra playground

Quick setup

1. Clone repo
2. Install dependencies: npm i
3. Create .env file with single line content OPENAI_API_KEY=[your openai api key]
4. Run frontend: npm run frontend:dev (default: http://localhost:5173/)
5. Run backend: npm run backend:dev (default: http://localhost:4111/ - frontend hard-coded to use this)
6. Open UI and have conversation
7. Check traces in Mastra Studio observability tab: http://localhost:4111/observability (or whatever port you run backend at)

Agentic architectural approaches

This repo contains prototypes for the following approaches:

• single agent
• single learning agent
• agent network
• multi-step workflow

To configure which approach is used, set the approach value in src/config.ts. Both backend and frontend dev-servers should get hot-reloaded when you change the config file. If they don't, then manually restart them. Backend uses in-memory storage, so restarting the server will reset the storage.

Example conversation messages

Koala quiz:

1. I'd like a quiz about koalas.
2. I am a high-school biology teacher. I will give a lecture about koalas, covering basic topics. I want to print and hand out a simple quiz with 8 easy questions. To help students practise. Please create quiz now.
3. Please add a bonus question about an aspect that we don't have a question for yet. And then show me the complete quiz again.