Skip to content
← Back to Blog

July 4, 2026

The MCP Connector Is Live: Connect Any Tool to Your AI Agents

The MCP connector is now live in AgentTeams. Until today, your agents could use the tools in our catalog, Slack, Google Workspace, Microsoft 365, HubSpot, and the rest. Now they can use almost anything. Paste the address of any MCP server, choose which of its tools your agent is allowed to call, and it starts working, no waiting on us to build the integration first.

This is the first integration you add yourself. If a tool your team relies on speaks MCP, whether it is a product you buy, an internal service your engineers built, or a niche app that would never make a mainstream catalog, your agents can reach it. The catalog is no longer the ceiling.

What MCP is, in one paragraph

MCP (the Model Context Protocol) is an open standard for connecting tools to AI. A growing list of products publish an MCP server, a single address that exposes their actions in a format any AI agent can understand. Instead of us hand-building a wrapper for every product on earth, we speak the standard, so any MCP server becomes a set of tools your agent can use the moment you connect it.

What is live

  • Connect by URL. Paste a server address, add a token or API key if the server needs one, and connect. The credential is encrypted before it is saved, and the whole thing stays scoped to the individual agent.
  • Automatic tool discovery. We ask the server what it offers and show you the full list. You pick which tools the agent can call with a switch each, or select all, so a server with a hundred actions does not mean the agent gets a hundred it does not need.
  • Works everywhere your agent works. Once connected, the tools are available across every channel the agent runs on, dashboard chat, Slack, email, and the rest, and on whichever engine powers the agent. Nothing extra to wire up per channel.
  • Manage and refresh any time. Turn individual tools on or off later, and re-discover when a server adds or renames tools, without disconnecting and starting over.

How to connect one

  1. Open your agent in AgentTeams and go to the Tools tab.
  2. Under Custom, choose MCP Server. Give it a name, paste the server URL, and add the token or API key if it needs one.
  3. We connect, list the server's tools, and hand you the switches. Enable the ones you want the agent to use, save, and you are done.

A word on trust

A native integration like Slack or HubSpot is one we built and vetted. An MCP server is one you point us at, so its tool names and descriptions are written by that server, not by us. We treat that content as untrusted by default and show you exactly what a server reports before anything is enabled, so you only ever switch on tools from a server you recognize and trust. Credentials are encrypted at rest, access is scoped to the single agent, and sensitive actions still pass through the same approval gates as every other tool. The same care we put into agent guardrails applies here too.

What you can do with it

  • Reach an internal service. If your engineers expose an internal tool over MCP, a billing lookup, a deploy trigger, a data query, an agent can use it the same way it uses Slack, no custom integration project.
  • Connect a product before we do. A vendor you already pay for ships an MCP server, an email-marketing platform, a monitoring tool, an analytics product. Connect it today instead of waiting for it to reach our catalog.
  • Cover the long tail. The niche, vertical, or regional app your team lives in but that no mainstream platform will ever integrate, if it speaks MCP, your agent can work with it.

Where this fits

The MCP connector joins Slack, Google Workspace, Microsoft 365, HubSpot, GitHub, and the rest of the catalog, and then goes past it. The pattern that makes any of this work is the one behind every multi-agent setup: one agent, several tools, one outcome. The MCP connector just means the set of tools an agent can reach is now open-ended.

Ready to connect one? Open an agent's Tools tab and choose MCP Server under Custom. New to AgentTeams? Start with the first-agent onboarding guide, or see how we compare to other AI agent platforms.