Mumega
Start

Model Context Protocol (MCP)

Model Context Protocol (MCP): the open standard for connecting AI agents to tools and data, and how AWS, Microsoft, and Google adopted it. June 2026.

The Model Context Protocol (MCP) is an open standard for connecting AI agents to tools, data sources, and external systems through a uniform interface. Instead of writing a bespoke integration per model and per tool, a developer exposes capabilities once as an MCP server, and any MCP-aware agent can discover and call them.

MCP answers the “hands” problem of an agent harness: a model can reason, but it needs a constrained, discoverable way to act on the world. MCP is that wire format.

Why MCP became the standard

By 2026 every major agent platform speaks MCP. It moved under the Linux Foundation’s Agentic AI Foundation, making it neutral ground rather than any one vendor’s property. The big-three clouds all built on it:

  • AWS Bedrock AgentCore Gateway turns REST APIs and Lambda functions into MCP tools with semantic tool search, and shipped a GA AWS MCP Server in May 2026.
  • Microsoft’s Agent Framework is MCP-native, and the Work IQ APIs expose Microsoft 365 org intelligence as 10 generic MCP tools.
  • Google’s stack treats MCP as the tool layer underneath the A2A protocol.

MCP and A2A operate at different layers: MCP connects an agent to its tools; A2A connects an agent to other agents.

How Mumega uses MCP

Mumega exposes its bus and substrate to agents over MCP — that is how Claude Code, Codex, and folder-resident agents reach the same shared memory, task board, and message bus. Because authorization is portable and app-layer, the MCP surface stays the same whether an agent runs on our infrastructure or the tenant’s own box.

Sources

Last updated: Jun 5, 2026