MCP Servers
AI tools —
on the open standard
The Model Context Protocol (MCP) is how AI agents discover and use tools. UCM gives you local, remote, and embedded MCP servers — and a Connection Gateway to reach dozens of enterprise systems through three tools.
The Reality
AI agents are only as capable as their tools
An AI assistant without access to tools can only reason on what is in its context window. With tools, it can search knowledge bases, read documents, create records, and take action in real systems.
The Model Context Protocol is the open standard that makes this work — a common language for AI agents and tools to communicate. UCM implements MCP natively. Whether you are using Claude Desktop, Claude Code, or building your own agent, UCM's MCP servers give AI the tools it needs to do meaningful work.
How It Works
Three server types. One consistent tool interface
Local MCP Server
Published to npm as ucm-mcp-server · Stdio transport · For Claude Desktop and Claude Code
Remote MCP Server
Web-accessible endpoint · Streamable HTTP · Enterprise cloud access
Agent MCP Server
Embedded in UCM Agent API · Tools available inside every UCM agent conversation
Key Capabilities
One consistent interface
UCM's MCP tools cover the full lifecycle: discover content, read artefacts, publish knowledge, manage repositories, access SharePoint, connect to remote systems.
The Unified Connection Gateway adds three tools that proxy any external connection — so an AI agent can reach Atlassian Jira, Confluence, or any custom service through a single, governed access point.
- MCP tools across discovery, artefacts, repositories, SharePoint, and connections
- Local MCP server: npm package for Claude Desktop and Claude Code
- Remote MCP server: HTTP endpoint for cloud and enterprise deployments
- Unified Connection Gateway: access dozens of external tools through 3 gateway tools
- Tool schemas loaded on demand — no context window bloat
- Full authentication support: API key, OAuth, and platform-managed credentials
MCP tools available
Unified Connection Gateway
One gateway. Any external system
Three tools give an AI agent access to every external connection your organisation has registered — SharePoint, Atlassian, Remote MCP Servers, and any custom service — without needing separate integrations or bespoke authentication code.
Any AI Agent
Claude · GPT · Custom
Unified Connection Gateway
ucm_connection_list
Discover registered external systems
ucm_connection_access
Fetch tool schemas for any connection
ucm_connection_call_tool
Execute tools on any external system
Microsoft 365
SharePoint · Teams · OneDrive
Atlassian
Jira · Confluence
Remote MCP Servers
Any MCP-compliant service
Custom Systems
APIs · databases · internal tools
How it works in practice: An agent calls ucm_connection_list to discover available systems, then ucm_connection_access to get the tool schemas for a specific system. From that point, ucm_connection_call_tool executes any tool on any system — authentication, credential refresh, and OAuth flows handled transparently.
MCP Tool Categories
Every AI capability your organisation needs
UCM's MCP tools are organised into five categories — covering everything from knowledge discovery to enterprise system access.
Get Started
Give your AI agents the tools they need
Install the local MCP server, configure the remote endpoint, or connect via the embedded agent API — all tools are ready from day one.