MCP (Model Context Protocol)

Imagine asking an AI assistant: "Who on my team has bandwidth next week for a React project?" — and instead of getting a generic answer, it actually checks your live resource schedule, filters by skill tags, and returns real names with real availability numbers. That's what MCP makes possible.

MCP — Model Context Protocol — is an open standard developed by Anthropic that defines how AI models connect to external tools, databases, and applications. Think of it as USB-C for AI: a universal plug that lets any compatible model talk to any compatible tool, without custom integrations for every combination.

Why It Matters for Agencies

Agencies run on data scattered across tools — schedules in one app, budgets in another, time tracking in a third. Traditionally, getting answers that span multiple systems means logging into each one separately, exporting data, and piecing things together.

MCP changes that. When your tools support MCP, an AI assistant can:

• Pull current allocations and utilization rates directly
• Check project budgets and burn rates
• Review team capacity and FTE counts
• Create or adjust allocations based on natural language requests

Instead of navigating a UI to answer a question, you ask it in plain language and the AI handles the tool interaction.

How MCP Works (Simply)

The protocol defines three roles:

1. MCP Host — The AI application (like Claude Desktop, Cursor, or any compatible client) 2. MCP Server — The connector that exposes a tool's data and actions (e.g., a WrkSched MCP server) 3. MCP Client — The protocol bridge that maintains the connection between host and server

When you ask the AI a question that requires external data, the host recognizes it needs information from the MCP server, sends a structured request, gets back structured data, and incorporates it into the response. All of this happens in the background — you just see the answer.

WrkSched's MCP Support

WrkSched includes a native MCP server that exposes scheduling data to any compatible AI client. This means you can connect Claude, or any MCP-compatible assistant, directly to your resource schedule and interact with it conversationally.

Practical examples:

• "Show me everyone who's overallocated next week"
• "What's our design team's bandwidth for March?"
• "Create a placeholder allocation for the new client project"
• "What's the profitability forecast for Project Alpha?"

This is the bridge between AI-powered scheduling as a concept and a tool you can actually use today. The protocol is open-source, actively developed, and gaining adoption across the industry — making it a future-proof investment for agencies building their tech stack.

Security and Control

MCP includes built-in permission layers. The server defines exactly which data and actions are exposed. Read-only access for reporting? Fine. Write access for creating allocations? Optional and configurable. The resource manager controls what the AI can and can't do, just like any other integration.