AI Agents with Model Context Protocol - Python Code Examples

This repository contains Python code examples for the Coursera course "AI Agents with Model Context Protocol."

Prerequisites

Python 3.10+
pip
An API key from OpenAI, Anthropic, or Google (set in .env file)

Setup

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

cd <module>/python-code
pip install -r requirements.txt
cp .env.example .env  # Then add your API key(s)

Course Structure & Code Mapping

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

Item	Type	Code Files
Why Do We Need Model Context Protocol?	Video	—
Model Context Protocol & AI Problem Solving with Tools	Video	—
MCP Allows AI to Communicate with the Computer	Video	—
AI @ Work - What My Work with AI Looks Like	Video	—
My AI Tools	Reading	—
Learning More & Staying Connected	Reading	—

Module 1 is conceptual introduction — no code exercises.

Module 2: AI Agent Loops & Model Context Protocol

Item	Type	Code Files	Run Command
Model Context Protocol: Syntax, Semantics, Timing	Video	—	—
Model Context Protocol & AI Agents	Video	—	—
What is an MCP Server?	Video	—	—
Tool Specifications	Video	—	—
Building Your First MCP Server	Ungraded Plugin	`server.py`	`python server.py`
Building Your First MCP AI Agent	Ungraded Plugin	`agent.py`, `llm.py`	`python agent.py "question"`
Agents Talking to Tools vs. Tools with AI	Video	—	—

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

Key Files:

File	Description
`server.py`	Basic MCP tool server with `list_files` and `read_file` tools
`agent.py`	Agent loop implementation (PERCEIVE→DECIDE→ACT→OBSERVE)
`llm.py`	LLM integration utilities (OpenAI/Anthropic/Gemini)
`test_server.py`	Tests server tools directly without an agent
`server_with_resources.py`	Server with MCP resources for teaching agents
`agent_with_learning.py`	Agent that reads resources before acting

Commands:

cd 01-intro-to-mcp-agents/python-code
pip install -r requirements.txt
python server.py              # Start the basic MCP server
python agent.py "question"    # Run the basic agent
python test_server.py         # Test server tools directly

Module 3: Building AI Agents with Model Context Protocol

Item	Type	Code Files	Run Command
Resources	Video	—	—
Teaching Agents to Use Tools	Ungraded Plugin	`workspace_server.py`	`python workspace_server.py`
Teaching Agents to Seek Help	Ungraded Plugin	`workspace_server.py`	`python workspace_server.py`
Helping Agents Find Guidance	Ungraded Plugin	`workspace_server.py`	`python workspace_server.py`
Helping AI Agents on the Fly	Ungraded Plugin	`workspace_server.py`	`python workspace_server.py`
Helping AI Agents Discover Workspace-related Guidance	Ungraded Plugin	`workspace_agent.py`	`python workspace_agent.py "question"`

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

Key Files:

File	Description
`workspace_server.py`	Enhanced server with workspace-aware tools and context discovery
`workspace_agent.py`	Agent that works within a defined workspace with context awareness
`workspace/`	Sample workspace with `.context.md` files for testing

Commands:

cd 01-intro-to-mcp-agents/python-code
python workspace_server.py           # Start workspace-aware server
python workspace_agent.py "question" # Run workspace agent with context discovery

Module 4: Robust Error Handling Techniques for AI Agents

Item	Type	Code Files	Run Command
Responses are More than Data	Ungraded Plugin	`expense_server.py`	`python expense_server.py`
Designing Errors to Help AI Agents	Ungraded Plugin	`expense_server.py`	`python expense_server.py`
Errors in Complex Workflows	Ungraded Plugin	`expense_server.py`	`python expense_server.py`
Minimizing AI Agent Cognitive Burden from Error Recovery	Ungraded Plugin	`expense_server.py`	`python expense_server.py`
Helping AI Agents Find Alternative Paths to Fix Errors	Ungraded Plugin	`expense_server.py`	`python expense_server.py`

Code Location: 02-failing-forward/python-code/

Key Files:

File	Description
`expense_server.py`	MCP server demonstrating all Failing Forward patterns
`expense_agent.py`	Agent that learns from errors to complete expense tasks
`test_failing_forward.py`	Comprehensive tests for all failing forward patterns

Commands:

cd 02-failing-forward/python-code
pip install -r requirements.txt
python expense_server.py           # Start the expense server
python expense_agent.py "request"  # Run the expense agent
python test_failing_forward.py     # Run failing forward tests

Pattern Mapping:

Coursera Item	Pattern	Implementation
Responses are More than Data	Response-as-Instruction	`next_action`, `hint` fields in responses
Designing Errors to Help AI Agents	Errors as Curriculum	`submit_expense` validation with guidance
Errors in Complex Workflows	Error Chains	`request_late_expense_approval` → `check_approval_status`
Minimizing AI Agent Cognitive Burden	Pre-filled Parameters	`suggested_params` in error responses
Helping AI Agents Find Alternative Paths	Alternative Actions	`alternatives` array in responses

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

Item	Type	Code Files	Run Command
Managing AI Agent Cognitive Load	Ungraded Plugin	`agent_heavy_server.py` `tool_heavy_server.py`	`python agent_heavy_server.py` `python tool_heavy_server.py`
Predictability, Lower Cost, Speed: Scripted Orchestration	Ungraded Plugin	`scripted_orchestration_server.py`	`python scripted_orchestration_server.py`
Prompts and MCP	Video	—	—
Self-Prompting: Adding Reasoning to Tools	Ungraded Plugin	`self_prompting_server.py`	`python self_prompting_server.py`
AI Agent Tool Design for Common Errors	Ungraded Plugin	`validate_at_source_server.py`	`python validate_at_source_server.py`
AI Agents, MCP, & Identity / Security	Video	—	—
Wrapping Up	Video	—	—
Final Assessment	Assignment	—	—

Code Location: 03-intelligence-budget/python-code/

Key Files:

File	Description
`agent_heavy_server.py`	Minimal tools approach - agent does most processing
`tool_heavy_server.py`	Rich tools approach - tools pre-process data for agent
`hybrid_server.py`	Balanced approach combining both strategies
`scripted_orchestration_server.py`	Tools that let agent write scripts for batch operations
`self_prompting_server.py`	Tools that make isolated LLM calls for semantic reasoning
`validate_at_source_server.py`	Tools with layered validation (format → business → semantic)
`database.py`	Shared mock database used by all server examples
`test_all.py`	Comprehensive test suite comparing all approaches

Commands:

cd 03-intelligence-budget/python-code
pip install -r requirements.txt
python agent_heavy_server.py   # Start agent-heavy server
python tool_heavy_server.py    # Start tool-heavy server
python hybrid_server.py        # Start hybrid server
python test_all.py             # Run comparison tests

Pattern Mapping:

Coursera Item	Pattern	Implementation
Managing AI Agent Cognitive Load	Intelligence Budget	Agent-heavy vs tool-heavy approaches
Predictability, Lower Cost, Speed	Scripted Orchestration	`execute_script` tool for batch operations
Self-Prompting: Adding Reasoning	Self-Prompting	Isolated LLM calls within tools
AI Agent Tool Design for Common Errors	Validate at Source	Layered validation stack (format → business → semantic)

Quick Reference: Coursera Item → Code File

Module	Coursera Item	Code File
2	Building Your First MCP Server	`01-intro-to-mcp-agents/python-code/server.py`
2	Building Your First MCP AI Agent	`01-intro-to-mcp-agents/python-code/agent.py`
3	Teaching Agents to Use Tools	`01-intro-to-mcp-agents/python-code/workspace_server.py`
3	Teaching Agents to Seek Help	`01-intro-to-mcp-agents/python-code/workspace_server.py`
3	Helping Agents Find Guidance	`01-intro-to-mcp-agents/python-code/workspace_server.py`
3	Helping AI Agents on the Fly	`01-intro-to-mcp-agents/python-code/workspace_server.py`
3	Helping AI Agents Discover Workspace-related Guidance	`01-intro-to-mcp-agents/python-code/workspace_agent.py`
4	Responses are More than Data	`02-failing-forward/python-code/expense_server.py`
4	Designing Errors to Help AI Agents	`02-failing-forward/python-code/expense_server.py`
4	Errors in Complex Workflows	`02-failing-forward/python-code/expense_server.py`
4	Minimizing AI Agent Cognitive Burden from Error Recovery	`02-failing-forward/python-code/expense_server.py`
4	Helping AI Agents Find Alternative Paths to Fix Errors	`02-failing-forward/python-code/expense_server.py`
5	Managing AI Agent Cognitive Load	`03-intelligence-budget/python-code/agent_heavy_server.py` `03-intelligence-budget/python-code/tool_heavy_server.py`
5	Predictability, Lower Cost, Speed: Scripted Orchestration	`03-intelligence-budget/python-code/scripted_orchestration_server.py`
5	Self-Prompting: Adding Reasoning to Tools	`03-intelligence-budget/python-code/self_prompting_server.py`
5	AI Agent Tool Design for Common Errors	`03-intelligence-budget/python-code/validate_at_source_server.py`

Environment Setup

Create a .env file in each module's python-code/ directory:

OPENAI_API_KEY=your-openai-key-here
ANTHROPIC_API_KEY=your-anthropic-key-here
GOOGLE_API_KEY=your-google-key-here

You only need to provide the API key(s) for the LLM provider(s) you plan to use.

Project Structure

code-repo/
├── README.md
├── .gitignore
│
├── 01-intro-to-mcp-agents/
│   └── python-code/
│       ├── server.py              # Module 2: Building Your First MCP Server
│       ├── agent.py               # Module 2: Building Your First MCP AI Agent
│       ├── llm.py                 # LLM integration utilities
│       ├── workspace_server.py    # Module 3: All workspace/context items
│       ├── workspace_agent.py     # Module 3: Workspace-related Guidance
│       ├── server_with_resources.py  # Server with MCP resources
│       ├── agent_with_learning.py    # Agent with active learning
│       ├── test_server.py         # Server testing
│       ├── test_resources.py      # Resource testing
│       ├── requirements.txt       # Python dependencies
│       ├── README.md              # Setup instructions
│       └── workspace/             # Sample workspace with .context.md files
│
├── 02-failing-forward/
│   └── python-code/
│       ├── expense_server.py      # Module 4: All error handling items
│       ├── expense_agent.py       # Agent for expense workflows
│       ├── test_failing_forward.py # Pattern tests
│       ├── requirements.txt       # Python dependencies
│       └── README.md              # Setup instructions
│
└── 03-intelligence-budget/
    └── python-code/
        ├── agent_heavy_server.py           # Module 5: Managing Cognitive Load
        ├── tool_heavy_server.py            # Module 5: Managing Cognitive Load
        ├── hybrid_server.py                # Balanced approach
        ├── scripted_orchestration_server.py # Module 5: Scripted Orchestration
        ├── self_prompting_server.py        # Module 5: Self-Prompting
        ├── validate_at_source_server.py    # Module 5: Tool Design for Errors
        ├── database.py                     # Shared mock database
        ├── test_all.py                     # Comparison tests
        ├── requirements.txt                # Python dependencies
        └── README.md                       # Setup instructions

License

MIT

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

AI Agents with Model Context Protocol - Python Code Examples

Prerequisites

Setup

Course Structure & Code Mapping

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

Module 2: AI Agent Loops & Model Context Protocol

Module 3: Building AI Agents with Model Context Protocol

Module 4: Robust Error Handling Techniques for AI Agents

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

Quick Reference: Coursera Item → Code File

Environment Setup

Project Structure

License

About

Uh oh!

Releases

Packages

Uh oh!

Contributors

Uh oh!

Languages

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
01-intro-to-mcp-agents/python-code		01-intro-to-mcp-agents/python-code
02-failing-forward/python-code		02-failing-forward/python-code
03-intelligence-budget/python-code		03-intelligence-budget/python-code
.gitignore		.gitignore
README.md		README.md

Folders and files

Latest commit

History

Repository files navigation

AI Agents with Model Context Protocol - Python Code Examples

Prerequisites

Setup

Course Structure & Code Mapping

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

Module 2: AI Agent Loops & Model Context Protocol

Module 3: Building AI Agents with Model Context Protocol

Module 4: Robust Error Handling Techniques for AI Agents

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

Quick Reference: Coursera Item → Code File

Environment Setup

Project Structure

License

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages