Skip to main content

MCP Tool

The MCP Tool allows Amplifi agents to connect with external systems, APIs, or enterprise data sources via a standardized interface. Amplifi comes with a built-in MCP — Tavily Web Search — already configured and ready to use.

You can also add external MCP tools for your own data pipelines, enterprise search APIs, or logic.

Purpose​

The MCP Tool allows you to bring your proprietary data or enterprise logic into agent responses. It acts as a bridge between Amplifi and your custom backend — whether that's a document search API, a ticketing system, or a specialized LLM orchestrator.

This is especially useful when your data is hosted externally or governed by strict access policies, but you still want to expose it securely through a conversational interface.

How It Works​

When an agent is equipped with the MCP tool, it can execute standard input/output commands using tools like npx or uvicorn. The external system (your MCP backend) processes the query through stdin/stdout, performs its logic (e.g., document search, vector search, summarization), and returns a response in a structured format.

Amplifi then renders this response in the conversation — as plain text, formatted cards, charts, or any custom UI defined on your side.

You can configure this via:

  • Command Configuration: Define the executable command and arguments for your MCP
  • Environment Variables: Set any required environment variables for your MCP
  • Package Dependencies: Specify MCP packages that can be executed with standard tooling

Configuration Example​

{
  "your_mcp_name": {
    "command": "npx",
    "args": ["-y", "@modelcontextprotocol/your-mcp-package", "arg1"],
    "env": {}
  }
}

Notes​

  • Replace "your_mcp_name" with the identifier you wish to assign (e.g., "custom_mcp").
  • Replace "@modelcontextprotocol/your-mcp-package" with your actual MCP package.
  • Modify "arg1" to the appropriate command required by your MCP.
  • Use the "env" section to define any environment variables your MCP requires.

Typical Use Cases​

  • Internal document search: Query your proprietary vector store and serve results through chat.
  • LLM routing: Use Amplifi as a frontend, and route complex prompts to a backend LLM orchestrator.
  • Custom AI tools: Connect to systems like LangChain agents, knowledge graphs, or structured APIs.
  • Enterprise dashboards: Let agents pull and display live business metrics from internal services.

Best Practices​

  • Keep your response payload clean and well-structured.
  • Include context-aware logic in your backend (e.g., summarization, filtering).
  • Use versioned endpoints to support iterative improvements without breaking agents.

The MCP Tool makes Amplifi highly extensible — bringing your own intelligence, data, and APIs into the loop while keeping the UX conversational and seamless.