AI Agents with Model Context Protocol & TypeScript - Code Examples

This repository contains code examples for the Coursera course “AI Agents with Model Context Protocol & TypeScript.”

Prerequisites

• Node.js 18+
• npm
• An OpenAI API key (set in .env file)

Setup

Each module has its own code/ directory. To set up any module:

cd /code
npm install
cp .env.example .env  # Then add your OPENAI_API_KEY

---

Course Structure & Code Mapping

Module 1: Getting Started with Model Context Protocol (MCP) for AI Agents

Module 1 is conceptual introduction — no code exercises.

---

Module 2: AI Agent Loops & Model Context Protocol

Code Location: 01-intro-to-mcp-agents/code/

Key Files:

Commands:

cd 01-intro-to-mcp-agents/code
npm install
npm run server       # Start the basic MCP server
npm run agent        # Run the basic agent
npm run test-server  # Test server tools directly

---

Module 3: Building AI Agents with Model Context Protocol

Code Location: 01-intro-to-mcp-agents/code/

Key Files:

Commands:

cd 01-intro-to-mcp-agents/code
npm run workspace-server  # Start workspace-aware server
npm run workspace-agent   # Run workspace agent with context discovery

---

Module 4: Robust Error Handling Techniques for AI Agents

Code Location: 02-failing-forward/code/

Key Files:

Commands:

cd 02-failing-forward/code
npm install
npm run server  # Start the expense server
npm run agent   # Run the expense agent
npm run test    # Run failing forward tests

Pattern Mapping:

---

Module 5: Faster, More Predictable, More Capable AI Agents

Code Location: 03-intelligence-budget/code/

Key Files:

Commands:

cd 03-intelligence-budget/code
npm install
npm run agent-heavy  # Start agent-heavy server
npm run tool-heavy   # Start tool-heavy server
npm run hybrid       # Start hybrid server
npm run test         # Run comparison tests

Pattern Mapping:

---

Quick Reference: Coursera Item → Code File

---

Environment Setup

Create a .env file in each module’s code/ directory:

OPENAI_API_KEY=your-api-key-here

---

Project Structure

code-repo/
├── README.md
│
├── 01-intro-to-mcp-agents/
│   └── code/
│       ├── server.ts              # Module 2: Building Your First MCP Server
│       ├── agent.ts               # Module 2: Building Your First MCP AI Agent
│       ├── llm.ts                 # LLM utilities
│       ├── workspace-server.ts    # Module 3: All workspace/context items
│       ├── workspace-agent.ts     # Module 3: Workspace-related Guidance
│       ├── test-server.ts         # Server testing
│       └── workspace/             # Sample workspace with .context.md files
│
├── 02-failing-forward/
│   └── code/
│       └── src/
│           ├── expense-server.ts       # Module 4: All error handling items
│           ├── expense-agent.ts        # Agent for expense workflows
│           └── test-failing-forward.ts # Pattern tests
│
└── 03-intelligence-budget/
    └── code/
        └── src/
            ├── agent-heavy-server.ts          # Module 5: Managing Cognitive Load
            ├── tool-heavy-server.ts           # Module 5: Managing Cognitive Load
            ├── hybrid-server.ts               # Balanced approach
            ├── scripted-orchestration-server.ts # Module 5: Scripted Orchestration
            ├── self-prompting-server.ts       # Module 5: Self-Prompting
            ├── validate-at-source-server.ts   # Module 5: Tool Design for Errors
            ├── database.ts                    # Shared mock database
            └── test-all.ts                    # Comparison tests

---

License

MIT