📰 News SQLite MCP Server

A Model Context Protocol (MCP) server for managing and querying a SQLite database of news articles

---

Built with FastMCP, this server exposes its capabilities to any MCP-compatible client — including AI orchestrators, agents, and assistants.

✨ Features

Tool	Description
get_schema	Retrieve the full schema of the SQLite database (tables, columns, types, constraints)
get_data_from_table	Query rows from the new_details table with optional WHERE, ORDER BY, LIMIT, and OFFSET support
prepare_query	Convert natural-language questions into SQL queries using an LLM (OpenAI, Groq, or DeepSeek)
post_into_table_details	Insert a new news article record into the database

🗄️ Database Schema

The server manages a single table called new_details:

Column	Type	Constraints
id	INTEGER	PRIMARY KEY, AUTOINCREMENT
timestamp	TEXT	NOT NULL
source	TEXT	NOT NULL
news	TEXT	NOT NULL
header	TEXT	NOT NULL
keywords	TEXT	NOT NULL

📋 Requirements

• Python 3.10+
• pip

🚀 Installation

1. Clone & Navigate

cd mcp_server

2. Create Virtual Environment

python -m venv venv
source venv/bin/activate   # On Windows: venv\Scripts\activate

3. Install Dependencies

pip install -r requirements.txt

4. Configure Environment Variables

cp .env.example .env

Edit .env to set your preferred LLM provider credentials:

# LLM Provider
LLM_PROVIDER=openai

# At least one of these API keys:
OPENAI_API_KEY=sk-...
DEEPSEEK_API_KEY=...
GROQ_API_KEY=...

# Toggle providers on/off
USE_OPENAI=true
USE_DEEPSEEK=false
USE_GROQ=false

# Optional: model overrides
OPENAI_MODEL=gpt-4o-mini
GROQ_MODEL=llama3-8b-8192

🎯 Usage

Running the Server

Start the server with stdio transport (default for MCP):

python server.py

Alternatively, run it directly via the FastMCP CLI:

fastmcp run server.py

🛠️ MCP Tools

Once the server is running, any MCP client can discover and invoke the following tools:

---

1. get_schema

Returns the full database schema as a human-readable string.

{
  "name": "get_schema",
  "arguments": {}
}

---

2. get_data_from_table

Query data from the new_details table.

Parameter	Type	Required	Default	Description
where_clause	string	No	null	SQL WHERE clause (e.g., source = 'Times of India')
limit	int	No	50	Max rows to return (max 500)
offset	int	No	0	Row offset for pagination
order_by	string	No	id DESC	ORDER BY clause

---

3. prepare_query

Convert a plain-English question into a SQLite query using the configured LLM.

Parameter	Type	Required	Description
user_question	string	Yes	Natural-language question for the DB

---

4. post_into_table_details

Insert a new news article.

Parameter	Type	Required	Description
source	string	Yes	News source name (e.g., "BBC", "Times of India")
news	string	Yes	Article content or summary
header	string	Yes	Article headline/title
keywords	string	Yes	Comma-separated keywords
timestamp	string	No	ISO-format timestamp (defaults to current UTC)

🔗 Integration with MCP Clients

Via .continue/config.json (Continue.dev)

{
  "mcpServers": {
    "news-sqlite-server": {
      "command": "python",
      "args": ["/absolute/path/to/mcp_server/server.py"]
    }
  }
}

Via claude_desktop_config.json (Claude Desktop)

{
  "mcpServers": {
    "news-sqlite-server": {
      "command": "python",
      "args": ["/absolute/path/to/mcp_server/server.py"]
    }
  }
}

📁 Project Structure

mcp_server/
├── data/               # SQLite database directory (auto-created)
├── venv/               # Virtual environment
├── .env                # Environment variables (not committed)
├── .env.example        # Example environment config
├── instructions.md     # Original design instructions
├── requirements.txt    # Python dependencies
├── server.py           # MCP server implementation
├── image/
│   └── Coverimage.png  # Cover image
└── README.md           # This file

📦 Dependencies

• FastMCP – Framework for building MCP servers.
• OpenAI Python SDK – LLM integration (also used for Groq and DeepSeek via compatible endpoints).
• python-dotenv – Environment variable management.
• httpx – HTTP client for LLM API calls.

📄 License

This project is part of a larger research backend. See the root project for license details.