FinPilot

FinPilot is a modular AI-driven assistant built using the LlamaIndex library. It leverages a Structured Planner Agent to interact with tools and execute tasks. The project also includes an MCP (Model Context Protocol) server for managing tool specifications.

---

🔧 Setup

git clone <repo-url>
pip install --upgrade pip
pip install uv
echo 'OPENAI_API_KEY="xyz"' > .env
uv sync  # (if using Python 3.10 virtualenv)
source .venv/bin/activate  # activate the virtual environment

---

🧠 Components

MCP Server

The MCP server provides tool specifications and handles tool-related queries. Built with the mcp[cli] package.

Agent

The agent.py script launches a Structured Planner Agent using:

• 🛠️ Tools: Loaded from the MCP server
• 🧠 LLM: OpenAI's GPT-4o
• 📜 Prompts: Custom planning instructions

Run MCP Server

uv run --active server.py
# MCP inspector:
mcp dev server.py

Run the Agent

python agent.py

---

📦 Project Structure

finbot/
├── agent.py
├── prompts.py
├── requirements.txt
├── .env
├── README.md
└── mcp_server/

---

🔍 Planning Flow Diagram

---

🎥 Working Demo

---

📝 Logging

We use the loguru library. Tool metadata is logged at init for better traceability.

---

📜 License

MIT License