SSalesCrew

What is a remote MCP server?

A local server lives on one laptop, so the rest of your team can't use it. A remote one is a URL everyone connects to.

An admin approves every new account by hand. Nothing is created until then. We reply by email; no newsletter, no sequence.

app.salescrew.io/inbox
The unified reply inbox with classified threads

Remote MCP server — A remote MCP server is a Model Context Protocol server that runs on the web at a URL, so AI clients connect to it over HTTP instead of launching it on your own computer. Its purpose is to let any teammate, from any compliant client, use the same tools on the same live data.

Also known as: Hosted MCP server

Why remote servers became the default for business apps

The MCP specification (version 2025-06-18, checked September 2026) defines two standard transports. With stdio, "the client launches the MCP server as a subprocess" on your machine. With Streamable HTTP, the server "operates as an independent process that can handle multiple client connections." That one is remote.

For a sales tool, the difference is practical. A local server needs setup on every laptop, plus a copy of the credentials on each one. A remote server is set up once. You connect Claude, ChatGPT or Cursor by pasting its URL, and the server decides what each person may do.

Streamable HTTP replaced an older HTTP plus SSE transport from the 2024-11-05 version of the spec. Some clients still list SSE as an option, so you'll see both names in settings screens.

How a client talks to a remote MCP server

  1. 1

    You add the server's URL

    One endpoint, such as a /mcp path.

  2. 2

    The client authenticates

    With a token or an OAuth sign-in.

  3. 3

    It posts an initialize request

    Client and server agree on a protocol version.

  4. 4

    It lists the tools

    Each with a name, a description and inputs.

  5. 5

    Tool calls go over HTTP POST

    Results come back as JSON or a stream.

A public URL puts authentication on the server

Teams often treat a remote server like a local one and skip access control. The spec's own wording: servers "SHOULD implement proper authentication for all connections." Once the endpoint is public, anyone with a valid token can reach every tool it lists. Scope each token to one person and log every call.

How a CRM can host one per customer

SalesCrew serves its MCP server over HTTP inside each client's own instance. Per-user tokens follow the same channel, profile and area rules as the app.

Questions

What is the difference between a local and a remote MCP server?
A local MCP server runs on your computer as a subprocess of the AI client and talks over stdio. A remote MCP server runs on the web and talks over Streamable HTTP. Local suits files and tools on one machine; remote suits shared business data that a whole team uses.
Which AI clients can connect to a remote MCP server?
Claude, ChatGPT and Cursor all connect to remote MCP servers. OpenAI's docs, checked September 2026, list SSE and streaming HTTP as the protocols ChatGPT developer mode supports. Each client has its own settings screen for adding the URL.
Is a remote MCP server slower than a local one?
Each tool call adds one network round trip. For CRM work that's rarely noticeable, because the model takes longer to write its answer than the call takes to return. Heavy file work on your own disk is where local servers still make sense.
Does the company running a remote MCP server see my data?
Whoever runs the server can reach whatever its tools read or write. Check where it is hosted and who holds the database. A server that runs inside your own instance keeps the data where the rest of your records already live.

Related