Local & Remote MCP Servers

Allow MCP Servers

This section details the complete workflow for analyzing, configuring, and enabling an MCP server for a specific project in the JFrog Platform.

Step 1: Select and Analyze

1. In the JFrog Platform, navigate to AI/ML > Discovery > MCP Servers.
2. Click on an MCP card to open the Model Details pane.
3. Review the available tabs to ensure the tool is safe and usable:
   • MCP Server Info: Displays general information about the MCP Server.
   • Config: Displays Environment Variables, for example if the MCP requires API keys or runtime arguments.
   • Identified Tools: Displays an AI-generated list of tools predicted to be in the package. This is a discovered suggestion list, not a final manifest. You can also see the which projects the MCP server is already allowed in (Allowed in project field).

📘

You can only allow the MCP one project at a time. To allow it for additional projects, repeat the procedure for the other projects.

Step 2: Add to Registry

1. Click Add to Registry in the Model Details pane.
2. Select a Project: Choose the specific project context where this tool will be active.

Step 3: Define Tool Policy

You must now choose how to govern the tools within this server. Select one of the following options:

• Allow all tools: Automatically approves all current tools and any tools added to this server in future updates.
• Select tools manually (Recommended): Opens the Allow List and Deny List configuration. This enables you to define granular policies using static text and Regex patterns. See Configure Tool Policies.

Step 4: Configure Required Variables

Configure the Environment Variables and Runtime Arguments.

Step 5: Save Configuration

Click Save Configuration. The MCP now appears in the AI Catalog Registry page and is available for developers in that project via the MCP Gateway.