MCP Servers
Connect your AI agent to external apps and services through MCP.
What is MCP?
MCP stands for Model Context Protocol. It is an open standard that lets AI agents communicate with external tools through a defined interface. Instead of building custom API integrations for every service, you connect an MCP server and the agent automatically learns how to use it.
Think of MCP servers as plugins for your agent. The agent already has built-in skills for things like web search and browser control. MCP servers add connections to third-party services: GitHub for code management, Slack for team communication, Notion for knowledge and documentation. The protocol works the same way regardless of which service you are connecting.
How it works on ClawStart
OpenClaw has native MCP support baked in. When you add an MCP server to your agent's config, it connects and discovers what operations are available. After that the agent can use those operations whenever it needs to. You tell your agent "create an issue on my GitHub repo" and it knows which MCP tool to call, what parameters to pass, and how to show you the result.
Setting up an MCP server usually comes down to providing an API token, like a GitHub personal access token or a Slack bot token. The token lives in your agent's isolated config file and nobody else has access to it.
Available servers
GitHub
Manage repositories, issues, and pull requests.
Slack
Send messages, read channels, and manage workspaces.
Notion
Create pages, query databases, and organize knowledge.
Custom MCP servers
The MCP ecosystem keeps growing. If there is an MCP server for a tool you use, chances are it will work with your ClawStart agent. The protocol is standardized, so any compliant server should connect without issues. Need help getting a custom MCP server running? Write to us at support@clawstart.io.