MCP Severs

MCP Servers (Model Context Protocol Servers) allow you to extend your AI agents with modular capabilities and external services.

By integrating MCP Servers, your agents can access real-time data, external APIs, and specialized functions beyond standard LLM responses.

---

What Are MCP Servers?

An MCP Server is a service endpoint that follows the Model Context Protocol and provides additional capabilities to your agents.

Examples of Available MCP Servers:

• Stripe MCP – Payment processing

• Brave Search MCP – Web and local search

• CoinCap MCP / BICScan MCP – Crypto prices and risk scoring

• Google Maps MCP – Location and routing data

• Perplexity Ask MCP – Real-time web question answering

• Armor Crypto MCP – Wallet and blockchain interaction

---

Exploring and Enabling MCP Servers

1. Navigate to “MCP Servers” from the sidebar.

2. Two tabs are available:

   • MCP Servers – All available servers in the marketplace

   • My MCP Servers – Servers you have enabled or submitted

3. Browse through the servers and review their name, provider, and capabilities.

4. Click “Enable Service” to activate an MCP server for your account.

Only enable MCP Servers that are relevant to your agent’s purpose to reduce latency and complexity.

---

Adding Custom MCP Servers

Developers can submit their own MCP Servers to add unique capabilities for private use or community sharing.

Steps to Submit a Custom MCP Server:

1. Navigate to MCP Servers → My MCP Servers → Submit MCP Server.

2. Fill in the required fields:

Field	Description	Example
Name	Unique name for your MCP server	GitHub Map MCP
URL	GitHub URL or repository link	https://github.com/username/github-mcp
Tags	Select functional tags for categorization	Search, Crypto, Payment
Description	Short explanation of what the server does	Fetches GitHub repo insights and stats
Server Config	JSON configuration including commands, args, and environment vars	(See example below)

Server Config Example:

{
  "mcpServers": {
    "github-mcp": {
      "command": "npx",
      "args": ["@modelcontextprotocol/server-github"],
      "env": {
        "GITHUB_PERSONAL_ACCESS_TOKEN": "YOUR_ACCESS_TOKEN"
      }
    }
  }
}

Best Practices:

• Include only the environment variables required (e.g., API tokens).

• Verify your server is MCP-compatible and responds correctly.

• Keep configuration simple to accelerate approval.

---

Using MCP Servers in Agents

Once an MCP server is enabled or submitted:

1. Open Agent Creation

2. Select MCP services for prompt-based agents.

3. Choose the MCP server you want to integrate.

4. Your agent can now call the MCP service to perform external operations.

Example Use Case:

• A Crypto Market Agent could:

  1. Query CoinCap MCP for live prices

  2. Use BICScan MCP to assess wallet risk

  3. Generate actionable investment recommendations for the user

---

6.5 Best Practices

• Enable only necessary MCPs to optimize performance.

• Test in Preview after attaching MCP services to verify outputs.

• Combine with workflows for multi-step, context-aware automation.

• Check third-party limits if MCP relies on external APIs to avoid request throttling.