MCP Servers for AI Agents: a Practical Guide

AI agents are bad at math. They hallucinate numbers, invent formulas, and confidently return wrong answers for calculations that a $5 calculator handles correctly.

MCP (Model Context Protocol) fixes this by giving agents access to external tools. Instead of guessing the answer to "what is the monthly payment on a $400K mortgage at 6.5%?", the agent calls a real mortgage calculator and returns the exact number.

Here is how it works.

What is MCP?

MCP is a protocol that lets AI models call external tools. Think of it as function calling, but standardized across models and clients.

An MCP server exposes a set of tools (functions) that the agent can discover and invoke. The agent sends structured parameters, the server runs the computation, and the result comes back as data.

No prompt engineering. No "please calculate this for me." Just a function call with inputs and outputs.

Why agents need MCP

LLMs are probabilistic. They generate text token by token, predicting the most likely next token. This works for language but fails for:

• Math: 23.7% of 847.50 is 200.8575, not "approximately 200."
• Dates: 45 business days from March 15, 2026 is a specific date. The model should not guess.
• Unit conversion: 68.5 kg to lbs has one correct answer (151.01 lbs). Not "around 150 lbs."
• Financial formulas: Compound interest with monthly contributions over 30 years has a precise result.

MCP lets the agent call a deterministic tool for these operations and return the exact answer.

Setting up MCP with Claude Desktop

Add this to your Claude Desktop configuration:

{
  "mcpServers": {
    "toolknife": {
      "type": "streamable-http",
      "url": "https://toolknife.com/api/mcp"
    }
  }
}

Restart Claude Desktop. You now have access to 273+ calculation, conversion, and developer tools.

Try asking Claude: "What is the monthly payment on a $350,000 mortgage at 6.5% for 30 years with 20% down?" It will call the mortgage calculator MCP tool and return the exact amount.

Setting up MCP with Cursor

In Cursor settings, add an MCP server with the same URL:

{
  "mcpServers": {
    "toolknife": {
      "type": "streamable-http",
      "url": "https://toolknife.com/api/mcp"
    }
  }
}

Now when you ask Cursor to "calculate the ROI of this investment" in your codebase, it can call the actual ROI calculator instead of writing an approximation.

What tools are available?

ToolKnife's MCP server exposes tools across these categories:

Each tool has typed input parameters and returns structured JSON. The agent handles the parameter mapping automatically.

Agent workflow example

A real estate assistant agent using MCP:

1. User: "I found a house listed at $450K. Can I afford it on my $95K salary?"
2. Agent calls mortgage-calculator (calculates monthly payment at current rates)
3. Agent calls affordability-calculator (checks DTI ratios)
4. Agent calls closing-costs-calculator (estimates closing costs)
5. Agent responds with a complete analysis, all numbers precise

Without MCP, the agent would estimate all these numbers and likely get several wrong.

Building your own agent with MCP

If you are building an AI agent with the Anthropic SDK, Claude Agents SDK, or any MCP-compatible framework:

# The agent automatically discovers tools from the MCP server.
# No tool definitions needed in your code.
# Just point your agent to the MCP server URL.

mcp_config = {
    "type": "streamable-http",
    "url": "https://toolknife.com/api/mcp"
}

The server handles tool discovery, parameter validation, and execution. Your agent code stays clean.

Free vs paid

Free tier requires no signup and no API key. Add an API key header for higher limits:

{
  "mcpServers": {
    "toolknife": {
      "type": "streamable-http",
      "url": "https://toolknife.com/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Set up MCP in 2 minutes or browse the full tool list.