Model Context Protocol (MCP) is a standard for connecting AI applications to external tools and data sources. An MCP server exposes capabilities such as document search, calendar access, database queries, and workflow actions; an agent client discovers and invokes them with structured inputs. section 18.2
MCP replaces custom integration code with named tools, descriptions, and input schemas while hiding system-specific API translation behind the server. It standardizes exposure and invocation, but authorization remains separate: each tool call still requires a policy decision before execution. section 18.2
{
"name": "documents.search",
"description": "Search customer documents available to the current task.",
"inputSchema": {
"type": "object",
"properties": {
"customerId": { "type": "string" },
"query": { "type": "string" }
},
"required": ["customerId", "query"]
}
}
A document-search tool illustrates how MCP describes capabilities and required inputs without exposing the underlying API implementation. section 18.2
