MCP Fundamentals

Learn the basics of AI engineering, bring the UX of ChatGPT to your Angular app, simplify forms, provide suggestions and shortcuts to your users, and build agentic web apps using our JavaScript runtime.

The Basics of AI Engineering

Learn how to use Large Language Models (like GPT-5) in your Angular app using Hashbrown.

1. The completionResource() function to generate simple text input completion (think Gmail).
2. The factors to consider when choosing a model.
3. The basics of the model APIs: tokens, temperature, json mode, and verbosity.
4. How to measure token usage and cost.

Model Content Protocol

Learn how MCP works through building a server and testing it with Postman.

1. Learn the architecture of the protocol.
2. Build a sample MCP server that exposes tools to the model.
3. Connect, list tools, and disconnect from the MCP server using Postman.
4. Expose resources to the model.

Integrating Remote MCP

Learn how to integrate a remote MCP server with tool calling in Hashbrown in a React / Angular web application.

1. Create an MCP client and connect to the remote MCP server using the streamable HTTP transport.
2. Fetch the list of tools from the remote MCP server.
3. Integrate remote MCP tool calling with client tools in Hashbrown.

Trust & Safety

For trust and safety, an intelligent agent (human) should be in the loop and be able to allow or deny tool invocations.

1. Integrate user interface for providing the user with control over the tools that the model can access.
2. Create a clear visualization of the tools that are being invoked by the model.
3. Integrating human confirmation prompts for tool calling.

Prerequisites

• You should have a basic understanding of TypeScript and either React or Angular.
• It is helpful if you have some experience using the OpenAI models and APIs.

Schedule

• Start: 8 am PT / 11 am ET
• End: 3 pm PT / 6 pm ET
• Breaks: two 15-minute breaks and one 30-minute break