Oracle MCP Server

DataGrout's Oracle MCP Server connects AI agents to Oracle Fusion Cloud through OAuth2, using DataGrout's application credentials or your own. Write operations are disabled by default and require explicit configuration per integration. 1000+ tools cover Financials, Procurement, Inventory, Suppliers, Tax, and Workforce, individually selectable rather than exposed as one fixed set.

Setup using CLI:

Go to the DataGrout GitHub and set up using command line interface - https://github.com/DataGrout/conduit-sdk

Setup using DataGrout Portal:

1. Sign up for a DataGrout account
Go to https://app.datagrout.ai/ and create an account.

2. Create a server

• Click the server dropdown in the top-left of the dashboard.

• Select Create New Server.

• Enter a name (required) and an optional description.

• Click Create Server.

• Your server is assigned a UUID and generates an MCP endpoint: https://gateway.datagrout.ai/servers/{your-uuid}/mcp

• You can retrieve this URL any time later under Settings > Server Info.

3. Create an access token

• Open the server dropdown and click the Settings icon.

• Go to the Authentication tab.

• Click Create Access Token.

• Give it a name, or leave it blank for an auto-generated one.

• Copy the token immediately — it is shown only once and cannot be retrieved again later.

4. Add the Oracle integration

• Go to the Home page — this is the integrations marketplace.

• Find Oracle in the list and click Add.

• You'll be redirected to the Oracle integration detail page to complete setup.

5. Configure authentication

• Choose authentication type: OAuth2.

• Choose whether to use DataGrout's application credentials or your own.

6. Authorize the connection

• Click Connect.

• You're redirected to Oracle's own hosted login/authorization screen to sign in and grant access.

• Once authorized, you're returned to the DataGrout dashboard.

7. Configure tools and policy

• On the integration's detail page, select which of the 1000+ available tools to expose for this integration, spanning Financials, Procurement, Inventory, Suppliers, Tax, and Workforce.

• Open the Policy tab to review or adjust the security policy. This inherits from the parent Hub Server's policy and can only be tightened, never loosened — write access must be explicitly enabled per tool if you want it, since it's disabled by default.

MCP Client Configuration:

DataGrout runs one MCP endpoint per server — the same URL and token work across every connected client, and every integration you add (Salesforce included) becomes available through it. No separate config is needed per integration.

{
  "mcpServers": {
    "DataGrout": {
      "url": "https://gateway.datagrout.ai/servers/YOUR_UUID/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      }
    }
  }
}

Use this same block for Claude Desktop, Claude Code, Cursor, and Windsurf. VS Code's MCP config may use "servers" instead of "mcpServers" — check your version.

Two other authentication options are available for this endpoint: mutual TLS (production/enterprise agents) and OAuth 2.1 (user-facing apps, via Client Credentials or Authorization Code + PKCE).

Configuration reference:

Setting	Values	Default
Auth type	OAuth2, mTLS, or Bearer token	—
Credentials	DataGrout app credentials or user-provided	—
Tool exposure	Individually selectable from 1000+ tools	none enabled until selected
Write access	Enabled per tool, per integration	disabled
Policy	Inherits from parent Hub Server; tightened, not loosened	inherited

Tools: 1000+ tools across Financials, Procurement, Inventory, Suppliers, Tax, and Workforce, individually enabled/disabled per integration. No named example tools confirmed.

Interaction settings: cost estimate before execution, execution receipt, execution summary after.

Links:

• Product: https://datagrout.ai/integrations/oracle-mcp-server

• Docs: https://library.datagrout.ai/

• Sign up: https://app.datagrout.ai/