IonBin
All pages
Search
Concept

Model Context Protocol (MCP)

MCP is an open protocol for connecting AI applications to outside tools and data. A server can support multiple assistants, but transport, authentication, permissions and feature compatibility still need checking.

Verified Oct 18, 2026 Review by Apr 16, 2027 4 sources

What it is

The MCP project describes it as a USB-C port for AI applications: one standard plug between an assistant and the systems it needs. An AI application is the host. Its MCP client connects to an MCP server, which exposes tools or data. A native connector may use MCP internally without offering a server that other assistants can connect to.

How it works

The server describes the tools and data it offers (for example “search contacts” or “create invoice”). The host decides which descriptions to make available to the model. When a tool call is permitted, the client sends it to the server and returns the result to the host.

The latest specification is version 2026-07-28. Authorization is optional in the spec. Servers reached over HTTP that support it should follow its OAuth 2.1-based flow, while stdio servers should take credentials from the environment instead. Transport, authentication, protocol version and optional features must also match the client.

Our Agent-ready page records documented MCP servers and named native connectors, with their sources and limits. A named connector is not proof that every MCP client can use it.

Benefits

  • Reuse a standard interface. One server can support several clients, subject to their transport, authentication and feature support.
  • Current data and permitted actions. Servers can expose live records or actions; some integrations are read-only.
  • Less glue. Fewer hand-built connectors and automations between your AI tool and your stack.

Limits and risks

  • Context cost. Tool descriptions and results can consume context. How much depends on the host’s tool discovery and loading behavior; MCP does not prescribe prompt construction.
  • Permissions. A connected agent can do whatever the connection allows. Scope access before you connect (see letting an agent act safely).
  • Unofficial servers. Community-built servers vary in quality and security. We only mark MCP as “yes” with an official source.

Checked specs

Snapshot from official sources on Oct 18, 2026.

Latest specification version2026-07-28 [2]
AuthorizationOptional; HTTP servers that support it should follow the spec's OAuth 2.1-based flow [3]

Best-fit stacks

Stack 06

Ops & admin automation

Hand repeat admin (receipts, logs, notices, handoffs) to workflows your agent can trigger.

Stack 03

Sales, CRM & outreach

Find the right accounts, enrich them, and keep every deal in one CRM, with an agent doing the busywork.

Stack 08

Dev, web & IT

Design, code, host and ship sites and internal tools, with an AI coding agent doing most of the typing.

Agent compatibility

Tool in this entryHow agents connect
Claude Agent platform MCPAPI/CLIFile import
ChatGPT Agent platform API/CLIFile import
HubSpot (free CRM) MCPAPI/CLIFile import
Notion MCPAPI/CLIFile import
Zapier Automation hub MCPAPI/CLIFile import

Levels come from each tool's recorded, sourced compatibility. How the levels work.

Related entries

Sources

  1. What is the Model Context Protocol (MCP)? · MCP projectofficial accessed Oct 18, 2026
  2. Specification (version 2026-07-28) · MCP projectofficial accessed Oct 18, 2026
  3. Authorization · MCP projectofficial accessed Oct 18, 2026
  4. Architecture overview · MCP projectofficial accessed Oct 18, 2026

Practitioner views are restated in our words and labeled as views, not measured results. Product facts come from official sources dated above.

Suggest an edit

IonBin's wiki isn't user-edited. Corrections go to our Directory Curator, who checks them against primary sources before anything changes. Spot something wrong or out of date on “Model Context Protocol (MCP)”? Email us with a link to a primary source.

Suggest an edit by email or write to contact@ionbin.com