Configuring the Agent

Each MCP-enabled application has an agent that represents its AI-facing interface.

View all settings

Setting	Description
Name	Text input, required. Unique identifier for the agent (e.g., Customer Orders).
Instructions	Multi-line text, required. Describes agent purpose and available data.

Agent Configuration

Name

The agent name identifies your application to AI clients. It should be:

• Descriptive - clearly indicates the application's purpose
• Unique - distinct across your organization
• Readable - uses proper casing like Customer Orders or Inventory Manager

Examples:

Customer Orders
Finance Agent
Inventory Manager

Instructions

Instructions describe what the application does and how AI should interact with it. Good instructions include:

• Purpose of the application
• Types of data available
• Common use cases
• Important constraints

Example:

Use this application to query financial data and generate
summaries. Available data includes transactions, revenue,
expenses, and budget information. Always specify date ranges
when querying historical data.

Best Practices

Do	Don't
Use clear, descriptive agent names	Use generic names like "app" or "agent"
Provide context about data and capabilities	Leave instructions empty
Document constraints and requirements	Assume AI will know implicit rules
Keep instructions concise but complete	Write lengthy paragraphs of documentation

Next Steps

• Adding Tools - Define what AI can access
• Generating Code - Export your MCP-ready application