Model Context Protocol (MCP) is an open standard that lets AI tools — Claude, Cursor, GitHub Copilot, and a growing list of other agents — connect to external data sources and tools through a uniform interface.
Also called: MCP · Anthropic MCP
The Model Context Protocol is an open specification published by Anthropic in late 2024. It defines how AI tools (clients) discover and call external systems (servers) — typically databases, APIs, file systems, or operational data layers. A server exposes resources and tools according to the protocol; a client connects to it and the model gets a uniform way to read and act on that external context.
Before MCP, every vendor invented their own glue. OpenAI had ChatGPT Plugins, then Actions. Anthropic had its own tool-use API. IDE assistants each had bespoke bridges to local files and project state. Each integration was point-to-point and didn’t transfer between tools. MCP collapses that fragmentation into a single contract.
MCP is now supported by Claude (Anthropic’s own product), Cursor, GitHub Copilot, Continue, Cline, and a steadily growing list of agentic IDEs and chat tools. On the server side, vendors are publishing MCP servers for their products so AI tools can talk to them: databases, version-control systems, project-management tools, internal company data layers.
The protocol is transport-agnostic (stdio, websocket, HTTP) and language-agnostic (servers exist in TypeScript, Python, Rust, Go). Aquifer’s MCP server is one of dozens in the production-grade catalog.
For end users, MCP means “the AI tool I already use can read my data without me copy-pasting it into the chat.” That’s the productivity unlock. For software companies, MCP means “we ship one server and our customers can use it from every AI tool, present and future.” That’s the strategic unlock — no longer betting on one AI vendor’s ecosystem.
Aquifer publishes an MCP server that exposes a customer’s connected operational data — Procore projects, Sage 300 CRE cost data, ArcGIS asset records, SharePoint document corpora — to any MCP-compatible AI tool. Customers configure once; their Claude, their Cursor, their custom agents all get the same access. The companion piece for OpenAI tools is the ChatGPT Action, which Aquifer ships in parallel.
See Your AI of Choice, Connected to Your Operational Data for a customer story and the architecture that makes it work.
We use a small number of first-party cookies to understand which pages visitors find useful. No advertising, no third parties. Read our privacy notice.