GoogleDocsMCP

# 🧠 Google Sheets Analyzer MCP (Model Context Protocol) This project is an implementation of a **Model Context Protocol (MCP)**-based system that allows you to: - **Read** and analyze Google Sheets from a specific **Google Drive folder** - Use tools like `get_insights`, `get_future_recommendations`, etc. - Ask questions and get responses using **Claude AI** (Anthropic) --- ## 📂 Project Structure ``` ├── flows/ ├── lib/ │ └── server/ │ └── server.py # Loads sheets from Google Drive and exposes MCP tools │ └── client/ │ └── client.py # Connects to MCP server and queries using Claude ├ └── configs/ │ └── configuration.py # Centralized config file ├── environment.yml # Conda environment definition ├── README.md ``` --- ## ⚙️ Features - ✅ Load Google Sheets from a **specific Drive folder** - ✅ Process and convert them into pandas DataFrames - ✅ Tools: - `get_insights`: Analyze the data and summarize trends - `get_future_recommendations`: Suggest actions based on insights - ✅ Chat-based interface powered by **Claude AI** - ✅ Fully functional **MCP Server/Client architecture** --- ## 🏗️ Setup Guide ### 1. 🔐 Google Cloud API Setup #### ✅ Enable APIs - Go to [Google Cloud Console](https://console.cloud.google.com/) - Enable: - Google Drive API - Google Sheets API #### ✅ Create a Service Account - Go to `APIs & Services > Credentials > Create Credentials > Service Account` - Assign roles like `Viewer` or `Drive Reader` - Click on your service account → "Keys" → Add new key → `JSON` - Save the file as `credentials/gcp_credentials.json` #### ✅ Share Folder with Service Account - Copy the **client_email** from the `gcp_credentials.json` - Go to Google Drive → open the **target folder** → Share with the client email ### 2. 🤖 Get Claude (Anthropic) API Key 1. Go to [https://console.anthropic.com](https://console.anthropic.com/) 2. Get your API key under “API Keys” 3. Set it as an environment variable: ```bash export ANTHROPIC_API_KEY="your-api-key" ``` --- ## 📦 Create & Activate Conda Environment ```bash conda env create -f environment.yml conda activate gsheet-mcp ``` --- You can override these using `.env` or `export` in your terminal. --- ## 🚀 How to Run ### 1. Start the MCP Server ```bash python lib/server/server.py ``` It will: - Authenticate using your Google credentials - Load spreadsheets from the target folder - Start a FastMCP server exposing tools like `get_insights` ### 2. Run the MCP Client In a new terminal (same environment): ```bash python lib/client/client.py ``` You can then start chatting with Claude and ask things like: ``` What are the top insights from the inventory sheet? What future recommendations can you give for budgeting? Summarize fund trends across all files. ``` --- ## 🧪 Example Prompt > "Give me future recommendations based on trends from all spreadsheets. Be concise and format your answer in bullet points." --- ## ✅ Checklist - [x] Google Cloud project & credentials - [x] Folder added to **My Drive** - [x] Folder shared with service account - [x] Anthropic key set in env - [x] Claude model set to `claude-3-haiku-20240307` --- ## 🛠️ TODO / Ideas - Add authentication via `gcloud auth` (for personal scripts) - Support other Claude models via config - Add tool for plotting charts from sheet data - Use embeddings to auto-suggest questions --- ## 🧑‍💻 Credits Created by [Harri200191!] to explore Claude-powered spreadsheet analytics with MCP.