What is an MCP server? A plain explanation
An MCP server gives an AI agent tools it can call. What a tool call looks like, and how one connection can reach many providers.
Walid Boulanouar · · 5 min read
On this page
MCP stands for Model Context Protocol. It is an open standard that lets an AI agent talk to outside software in one agreed way. An MCP server is the piece of software on the other end. It tells the agent which tools it offers and runs them when the agent asks.
The problem MCP solves
A language model on its own can only write text. It cannot look up an email address, check a ranking or read a web page. To do those things it needs tools, and every tool used to need its own glue code. One app had one way to describe a tool, another app had a different one.
MCP gives every tool the same shape. An agent that speaks MCP can connect to any MCP server without custom code for that server. You add the server once in your agent's settings, and the agent sees what it can do.
Three parts
There are three roles. The agent is the program you talk to. The MCP server is the connection point. The provider is the service that holds the data or does the work.
| Part | What it does | Example |
|---|---|---|
| Agent (the MCP client) | Reads your request, decides which tool to call, reads the result | Claude Code, Cursor, Codex |
| MCP server | Lists its tools, accepts calls, returns results | A server your agent is connected to |
| Provider | Holds the data or runs the job | A search API, an email finder, a scraper |
Sometimes the server and the provider are the same company. Sometimes the server sits in front of several providers and passes calls along. The agent does not need to care which.
What a tool call looks like
When you ask an agent to find recent news about a company, it picks a tool from the server's list and sends a request. Under the hood this is a small JSON message. The tool name below is made up to show the shape.
{
"method": "tools/call",
"params": {
"name": "search_web",
"arguments": { "query": "Acme Corp funding news", "limit": 5 }
}
}The server runs the search and answers with text or structured data. The agent reads the answer and either replies to you or makes another call.
Each tool also carries a description and a list of inputs. The agent reads those to decide when the tool fits.
Where looot fits
looot is one MCP connection to many data providers. Instead of adding a search server, an email server and a scraping server one by one, you connect your agent to looot once. The agent can then search the catalog for a job in plain words, read what an endpoint takes and returns, and call it.
looot is not the official MCP server of any provider listed in the catalog. It is a separate service that calls those providers' APIs for you and bills each call. If you want to see what an agent can do with a single provider, the Exa page shows example jobs and prompts. The full list is on the MCP index.
How to try one
- Pick an agent that supports MCP, such as Claude Code or Cursor.
- Add the server in the agent's MCP settings, or give the agent a setup file and let it do the steps.
- Sign in if the server asks you to. Most hosted servers use a normal browser sign-in.
- Ask for one small job and watch which tool the agent calls.
- Check the result before you ask for a large batch.
Prompt for your agent
Prompt to see what a server offers
What to watch for
Tools that read data are low risk. Tools that write, send or spend need care. Check what a server can do before you connect it. If a server can send email or change records, tell your agent to ask first.
For a comparison with calling a plain API, read MCP vs API for agents. To set up looot, see set up looot with one line.
Questions
Keep reading
MCP vs API for agents: when each one fits
MCP and a plain API both let an agent reach outside data. What the agent handles in each, when to pick which, and a prompt to test both on your job.
Walid Boulanouar · · 5 min read
Set up looot with one line: skill.md and llms.txt
Give your agent one URL and it can set up looot itself. What skill.md and llms.txt are, what each one tells an agent, and a prompt you can paste today.
The looot team · · 3 min read
Build your own CRM with an agent and Supabase
Let an AI agent build a small CRM in your own Supabase project. The five tables, the loop, seven rules, the exact prompt and one step not available yet.
Walid Boulanouar · · 5 min read
Try it with the agent you already use
Start your workspace, top up, and paste one prompt into your agent.