README.md•2.21 kB
# GullakAI MCP Server
A lightweight MCP server for personal finance assistance, expense tracking, financial news simplification, and local purchasing power comparisons across Indian cities.
---
## Features
- **Core Finance Q&A**
Conversational AI that answers personal finance questions clearly and simply in multiple languages.
- **Financial News Simplifier**
Summarizes and explains complex financial news and jargon in easy-to-understand language.
- **Expense & Budget Monitor**
Manage budgets, record expenses, track debts and bills using simple chat commands.
- **Purchasing Power Checker**
Compare salary affordability and cost of living across cities using local purchasing power parity data.
---
## Setup Instructions
1. **Clone the repository and install dependencies:**
```bash
uv .venv
uv sync
source .venv/activate
```
2. **Configure environment variables:**
Create a ```.env``` file in the root folder with:
```ini
AUTH_TOKEN=your_secure_token_here
MY_NUMBER=your_whatsapp_number_here
```
3. **Add purchasing power data:**
Ensure the file purchasingpower.json is present with the structure:
```json
{
"cities": [
{"city": "Hyderabad, India", "local_purchasing_power_index": 154.1},
{"city": "Bangalore, India", "local_purchasing_power_index": 149.7}
// ... more cities
]
}
```
4. **Run the MCP server:**
```bash
python server.py
```
---
## Usage
1. Interact with the MCP tools through your WhatsApp bot or other MCP-compatible clients.
Expense commands supported:
set budget <month> <category1> <amount1> ...
spent <amount> on <category>
owe <person/description> <amount>
bill <description> <amount> due YYYY-MM-DD
summary [month]
2. Ask natural language finance questions or purchasing power comparisons.
---
## Notes
Expense and budget data is cached in memory with a 7-day expiration.
Designed for asynchronous operation suitable for WhatsApp and browser-based chatbot integration.
---
## License
This project is licensed under the MIT License.
***Let me know if you want it expanded or more examples included!***