Tools

MCP tools replace per-framework tool definitions. Write one tool schema on the server; every MCP client discovers and calls it identically.

MCP tools are standardized power outlets — any MCP client can plug in, regardless of who manufactured the server.

Scenario

A GitHub MCP server exposes tools: create_issue, search_code, get_pull_request. Cursor's agent calls search_code to find relevant files in a repo.

Solution

In Model Context Protocol, apply Tools to this scenario: A GitHub MCP server exposes tools: create_issue, search_code, get_pull_request. Identify the inputs, run the technique, validate the output, and note one thing you would monitor in production.

Do this before moving to the next module — reading alone is not enough.

Open the Code Walkthrough below and run it locally. Change one parameter related to Tools (e.g. model, temperature, top_k, or tool name), observe the difference in output, and write 2–3 sentences explaining what changed.

Highlighted lines show where Tools happens in the code.

1# Tools — minimal example2from openai import OpenAI3
4client = OpenAI()  # create API client5
6# Ask the model to explain this topic7response = client.chat.completions.create(  # core API call for Tools8    model="gpt-4o-mini",9    messages=[10        {"role": "system", "content": "You explain tools clearly."},11        {"role": "user", "content": f"What is tools?"},12    ],13    temperature=0,14)15print(response.choices[0].message.content)  # show output for debugging

• Treating Tools as a black box without evaluation
• Ignoring cost and latency in production
• Skipping error handling for tools

Quick recap — the most important points from this module.