Open Source Project On Github: https://github.com/dataspider99/Google-AI-Worker

An intelligent application that accesses Google data (emails, chat, workspace) via OAuth and integrates with Oshaani.com AI agents for automated smart work—acting like a virtual employee.

Features
Feature Description
Multi-user Session auth (browser) + API keys (scripts/cron). Each user's data is isolated.
Automation Workflows run continuously for all logged-in users (every 30 min by default).
Drive storage User credentials stored in their Google Drive (folder "Google Employee").
Task storage Action items from workflows stored in Google Tasks (list "Google Employee").
MCP server HTTP MCP endpoint at /mcp for Cursor, Claude Desktop, and other MCP clients.
Logging Configurable logging (DEBUG, INFO, WARNING, ERROR) to stdout.
Google OAuth Access Gmail, Chat, Drive, Docs, Sheets, Tasks
Oshaani Send context to AI agents for intelligent processing
Workflows
Smart Inbox – Summarize emails, highlight urgent items, draft replies
Chat Assistant – Analyze Chat spaces and respond to FAQs
Chat Auto-Reply – Auto-reply to Chat messages via Oshaani agent
Document Intelligence – Summarize Drive activity and key documents
Custom – Fully configurable workflows with your specified data
User Guide
Step 1: Prerequisites
Python 3.9+
Google Cloud project with OAuth credentials
Oshaani.com account and Agent API Key
Step 2: Installation

Clone or navigate to project

cd "Google Employee"

Create virtual environment

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

Install dependencies

pip install -r requirements.txt
Step 3: Google Cloud Setup
Go to Google Cloud Console
Create a project (or select existing)
Enable APIs: APIs & Services → Library
Gmail API
Google Chat API
Google Drive API
Google Docs API
Google Sheets API
Google Tasks API
Configure the Chat app (required – fixes 404 "Chat app not found"):
Go to Chat API Configuration
Under Application info set: App name (e.g. "Google Employee"), Avatar URL (optional, HTTPS image), Description (e.g. "AI automation for Gmail and Chat")
Turn Interactive features OFF (optional for API-only usage)
Click Save
Configure OAuth consent screen: APIs & Services → OAuth consent screen
User type: External (or Internal for Workspace)
Add scopes: gmail, chat, drive, docs, sheets, userinfo
Create credentials: APIs & Services → Credentials → Create credentials → OAuth client ID
Application type: Web application
Authorized redirect URIs: http://localhost:8000/auth/google/callback
Copy Client ID and Client Secret
Step 4: Oshaani Setup
Sign up at oshaani.com
Create an agent and train it for your use case
Publish the agent
Copy the Agent API Key (Agent → API Access)
Step 5: Configure
cp .env.example .env

Edit .env with your values

.env:

Google OAuth

GOOGLE_CLIENT_ID=your-client-id.apps.googleusercontent.com
GOOGLE_CLIENT_SECRET=your-client-secret
GOOGLE_REDIRECT_URI=http://localhost:8000/auth/google/callback

Oshaani

OSHAANI_API_BASE_URL=https://oshaani.com
OSHAANI_AGENT_API_KEY=your-agent-api-key

App

SECRET_KEY=your-secret-key
APP_BASE_URL=http://localhost:8000

Optional

AUTOMATION_ENABLED=true
AUTOMATION_INTERVAL_MINUTES=30
LOG_LEVEL=INFO
Step 6: Run
Local (development):

uvicorn main:app --reload --port 8000
Docker:

Build and run with docker-compose

docker compose up -d

Or build and run manually

docker build -t google-employee .
docker run -p 8000:8000 --env-file .env -v $(pwd)/data:/app/data google-employee
Step 7: First-Time Login
Open http://localhost:8000/auth/google
Sign in with your Google account
Approve the requested permissions
You’ll be redirected to /me when done
Step 8: Use the App
Browser (session):

Visit http://localhost:8000/docs for the Swagger UI
Endpoints use your session automatically
Try workflows such as Smart Inbox, Document Intelligence, etc.
API key (scripts/cron):

While logged in, call POST /api-key
Copy the returned API key (shown only once)
Use it in requests: Authorization: Bearer
Logout:

http://localhost:8000/auth/logout – clears session and redirects to login
API Reference
Auth
Method Endpoint Auth Description
GET /auth/google No Start Google OAuth
GET /auth/logout No Logout, redirect to login
POST /auth/logout No Logout (JSON)
GET /me Yes Current user info
POST /api-key Yes Generate API key
GET /me/drive-data Yes View Drive-stored data
Workflows
Method Endpoint Auth Description
POST /workflows/smart-inbox Yes Summarize inbox + draft replies; creates Google Tasks from action items
POST /workflows/first-email-draft Yes Read first email, AI drafts reply, create Gmail draft
POST /workflows/chat-assistant Yes Analyze Chat messages
POST /workflows/chat-auto-reply Yes Auto-reply to Chat via Oshaani
GET /workflows/chat-spaces Yes List Chat spaces
POST /workflows/document-intelligence Yes Analyze Drive/Docs
POST /workflows/custom Yes Custom workflow
POST /workflows/run-all Yes Run all workflows manually
Tasks
Method Endpoint Auth Description
GET /tasks/lists Yes List Google Task lists
GET /tasks/lists/{id}/tasks Yes List tasks in a list
POST /tasks Yes Create task (title, notes; uses "Google Employee" list by default)
MCP
Method Endpoint Auth Description
GET /mcp No MCP endpoint info
POST /mcp Yes (API key) MCP JSON-RPC (tools/list, tools/call)
Usage Examples
Browser (session)

1. Sign in at http://localhost:8000/auth/google

2. Use Swagger UI at http://localhost:8000/docs

Or with curl (cookies from browser):

curl -X POST "http://localhost:8000/workflows/smart-inbox" -b cookies.txt -c cookies.txt
API key (scripts / cron)

Generate key first (while logged in):

curl -X POST "http://localhost:8000/api-key" -b cookies.txt -c cookies.txt

Use the key:

curl -X POST "http://localhost:8000/workflows/smart-inbox" \
-H "Authorization: Bearer ge_your_api_key"

Chat auto-reply (need space name from /workflows/chat-spaces):

curl -X POST "http://localhost:8000/workflows/chat-auto-reply?space_name=spaces/xxx" \
-H "Authorization: Bearer ge_your_api_key"
MCP (Cursor / Claude)

List tools

curl -X POST http://localhost:8000/mcp \
-H "Authorization: Bearer ge_your_api_key" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'

Call smart_inbox

curl -X POST http://localhost:8000/mcp \
-H "Authorization: Bearer ge_your_api_key" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"smart_inbox","arguments":{}}}'
Automation
After login, workflows run automatically for all users with credentials:

Smart Inbox – Summarize emails, suggest replies
Document Intelligence – Summarize Drive activity
Chat Auto-Reply – Reply to latest messages in each Chat space
Configuration (.env):

AUTOMATION_ENABLED=true (default)
AUTOMATION_INTERVAL_MINUTES=30
Manual trigger: POST /workflows/run-all

Drive Storage
User data is stored in the user’s Google Drive:

Folder: Google Employee (in Drive root)
File: user_data.json (credentials and settings)
Local bootstrap: Minimal refresh token kept locally for server restarts
Users who signed in before Drive storage may need to re-auth to grant drive.file scope.

MCP Server
HTTP MCP server at /mcp for Cursor, Claude Desktop, and other MCP clients.

Auth: Authorization: Bearer

Tools:

Tool Description
smart_inbox Summarize Gmail and suggest replies (creates Tasks from action items)
first_email_draft Read first email, AI drafts reply, create Gmail draft
chat_spaces List Google Chat spaces
document_intelligence Summarize Drive activity
chat_auto_reply Auto-reply to Chat (requires space_name)
run_all_workflows Run all workflows
list_task_lists List Google Task lists
list_tasks List tasks in a list (requires task_list_id)
create_task Create task in Google Tasks (requires title)
Logging
Logs go to stdout. Set LOG_LEVEL in .env: DEBUG, INFO, WARNING, ERROR. Default: INFO.

2026-02-08 23:19:37 | INFO | google_employee | User logged in: user@example.com
Project Structure
├── main.py # FastAPI app
├── config.py # Configuration
├── storage.py # Credentials + API keys
├── logging_config.py # Logging setup
├── auth/
│ ├── deps.py # Auth dependencies (get_current_user)
│ └── google_oauth.py # Google OAuth
├── services/
│ ├── google_data.py # Fetch Gmail, Chat, Drive
│ ├── oshaani_client.py # Oshaani API client
│ ├── orchestrator.py # Workflow orchestration
│ ├── automation.py # Automation runner
│ └── drive_storage.py # Drive read/write
├── mcp_server/
│ └── server.py # MCP HTTP server
├── ARCHITECTURE.md
└── requirements.txt
Troubleshooting
404 "Google Chat app not found"
Configure the Chat app in Google Cloud: Chat API Configuration → set App name, Avatar URL, Description → Save.

403 PERMISSION_DENIED
Google Cloud – Enable Gmail, Chat, Drive, Docs, Sheets APIs
OAuth scopes – Re-auth via /auth/logout then /auth/google
Workspace – Chat API may need admin approval for chat.messages scopes
Unverified app – Use "Advanced" → "Go to app (unsafe)" if needed
Missing client_id
Set GOOGLE_CLIENT_ID and GOOGLE_CLIENT_SECRET in .env from Google Cloud Console.

MCP 401
Ensure Authorization: Bearer is sent. Generate an API key via POST /api-key.

Security
Store tokens and API keys securely
Use environment variables for secrets
Request minimal OAuth scopes
Rotate API keys regularly
Use HTTPS in production