MCP Tool Server (FastAPI)

A lightweight Model Context Protocol (MCP)-style server built with FastAPI that exposes backend tools (e.g., sentiment analysis, calculator) to LLM/agent workflows via structured JSON requests and responses.

Key Features

• Tool routing layer to invoke backend utilities through a single API
• Structured JSON I/O designed for LLM/agent compatibility
• Example tools included: sentiment analysis + calculator
• Clean, minimal project structure for easy extension

Project Structure

. ├── main.py ├── tools_registry.py ├── tools/ ├── models/ ├── ml_training/ └── README.md

API Endpoints

POST /invoke-tool

Invokes a registered backend tool using structured JSON input.

Request

{
  "tool": "sentiment",
  "input": "I really like how this MCP server works"
}

{
  "tool": "sentiment",
  "result": {
    "label": "positive",
    "score": 0.91
  }
}

**Response**
```json
{
  "tool": "sentiment",
  "result": {
    "label": "positive",
    "score": 0.91
  }
}

### Available Tools
- `sentiment` – sentiment analysis on input text
- `calculator` – evaluates arithmetic expressions