MCP Server

At a glance

The MCP (Model Context Protocol) server lets you control Ontologie from your AI clients — Claude Desktop, VS Code, Cursor, or any compatible client. You can query your ontology, run workflows, search documents, and manipulate data directly from your usual AI interface.

Example: from Claude Desktop, ask "What are the active suppliers in my ontology?" and the agent queries your data directly via MCP, without opening the application.

Video tutorial — Connect MCP to Claude Code

What you will do

• Create an API key with the MCP preset
• Configure your AI client to connect to Ontologie
• Use MCP tools to interact with your data

Key vocabulary

Term	Meaning
MCP	Model Context Protocol — an open standard for connecting AI models to data sources and tools.
Tool	An action the AI client can execute (search, create entity, run workflow, etc.).
Resource	Readable content exposed by the server (ontology schema, dashboard list, etc.).
Prompt	A pre-configured query template for common use cases.

Before you start

1. Have an API key with the MCP preset (see API Keys)
2. Have an MCP client installed

Configuration by client

Claude Desktop

Open the Claude Desktop configuration file (claude_desktop_config.json) and add:

{
  "mcpServers": {
    "ontologie": {
      "url": "https://api.ontologie-growthsystemes.com/api/mcp",
      "headers": {
        "X-API-Key": "df_YOUR_MCP_KEY",
        "x-workspace-id": "YOUR_WORKSPACE_ID"
      }
    }
  }
}

VS Code (Copilot MCP extension)

Add to your VS Code settings.json:

{
  "mcp": {
    "servers": {
      "ontologie": {
        "url": "https://api.ontologie-growthsystemes.com/api/mcp",
        "headers": {
          "X-API-Key": "df_YOUR_MCP_KEY",
          "x-workspace-id": "YOUR_WORKSPACE_ID"
        }
      }
    }
  }
}

Cursor

Add in Cursor's MCP configuration (Settings > MCP Servers):

{
  "ontologie": {
    "url": "https://api.ontologie-growthsystemes.com/api/mcp",
    "headers": {
      "X-API-Key": "df_YOUR_MCP_KEY",
      "x-workspace-id": "YOUR_WORKSPACE_ID"
    }
  }
}

Tip
Store your API key in an environment variable rather than in plaintext in the configuration file.

Available tools

The default public surface exposes 39 tools across 3 categories. This total is computed from the backend preview-stable contract. Additional tools may appear on preview-full when the workspace plan, modules, and feature flags allow them.

Category	Example tools	Usage
Ontology	Describe, query, traverse, inspect plans	Read and understand the governed model
DataForge Capabilities	Answer questions, dossiers, freshness reports	Access composed read-only business capabilities
Administration	Health and usage	Inspect service readiness and consumption

Your client's tools/list response is the source of truth for the current workspace because scopes, modules, and flags can further restrict the result. See the generated stable tool reference.

Scopes and permissions

MCP tools require specific scopes on your API key:

Scope	Access
mcp.read	Read-only (list, search, view)
mcp.write	Write (create, update, delete)
mcp.workflow.execute	Trigger workflow executions
mcp.agent.execute	Invoke the AI agent

The MCP preset on the API Keys page automatically configures the recommended scopes.

Expected result

Your AI client is connected to Ontologie. You can query your data, run workflows, and search documents directly from Claude Desktop, VS Code, or Cursor.

See also

• API Keys — Create a key with the MCP preset
• Permissions — MCP scopes reference
• Support and contact

What's new

• See MCP updates