What is MCP and how does it connect agents to a team?
By Soumen Sarkar · October 2026
Short answer: MCP, the Model Context Protocol, is an open standard that lets AI apps and coding agents call tools and read data from outside services through one kind of connection. For a team, an MCP server can give every person's agents, in Claude, Claude Code, Codex or Cursor, the same shared context and the same tools.
What is MCP?
MCP is a protocol for connecting AI applications to the tools and data they need. An MCP server offers a set of tools, each with a name, a description and inputs. An MCP client, such as Claude, Claude Code, Codex or Cursor, connects to the server and lets its model call those tools during a conversation or a coding session.
Because it is a shared standard, one server works with many clients. A service written once can be reached from whichever agent a person prefers.
Why does it matter for a team?
A team rarely uses one agent. One developer runs Claude Code, another prefers Codex, a third works in Cursor, and a product manager uses Claude on the desktop. Without a shared connection, each of them sees only their own sessions. A decision made in one tool is invisible in the others.
An MCP server for the team gives all of them the same record and the same actions. What one agent writes, every other agent can read, whichever client it runs in.
What can a team MCP server do?
The tools depend on the server. For working together at agent pace, the useful ones are:
- Search what the team has decided and written.
- Record a decision or rule with its reason.
- State an assumption so the right person can confirm or correct it.
- Announce work in progress with its goal and plan, and hand off work that stops.
- Read what changed around the agent since its last session.
A team is redesigning Settings · app/settings. In Codex, one developer's agent decides that every settings change saves at once, with no Save button, because people lost changes by leaving the page. It records the decision through the team's MCP server.
An hour later, another developer's Claude Code session starts adding notification preferences to the same screen. It searches the record before building the form and finds the decision. It saves each toggle at once, with no Save button, and the two halves of the screen behave the same. The developers never spoke, and they used different tools.
What doesn't MCP do on its own?
MCP makes tools available; it doesn't make an agent use them. An agent calls a tool when its instructions, or the server's own instructions, tell it to. In Claude Code, hooks can go further and bring context in at fixed moments without the agent deciding to.
How does Campfyrr use MCP?
Campfyrr is an MCP server for a team's shared context, the backbone of VibeFlock. Claude adds it as a custom connector from /connect; Claude Code gets the connection and its hooks in a one-line setup; Codex, Cursor and other MCP clients connect with a token, call the same tools and follow the instructions the server gives them.
Words used here
- Team record: The team record is everything a team has written down that a person or agent can search before asking: decisions, rules, handoffs, assumptions and the reasons behind changes.
- Decision: A decision is a call made in the work, recorded with its reason at the moment it is made, so it travels to everyone who touches the same ground.
- VibeFlock: VibeFlock is building one product with a team of people and AI agents at agent pace, where context travels with the work so nobody waits.
The principle behind this: Answer it yourself: Look first. If the answer isn't there, assume it or decide it.
Move as one at agent speed. Nobody waits. Add Campfyrr to Claude: it briefs every agent before it touches the code, keeps assumptions and handoffs where the team can see them, and never takes your code (what it keeps).