rohithsiddi/AgentFlow-Multi-Agent-Workflow-Orchestration-Assistant

AgentFlow

AgentFlow is a multi-agent workflow orchestration assistant built with LangChain, Gradio, and modern LLMs. It supports tool-augmented reasoning, Retrieval-Augmented Generation (RAG) from uploaded PDFs, and Google Calendar integration.

Live Demo: Try AgentFlow directly in your browser via Hugging Face Spaces: 

Features

• —Multi-Agent Orchestration: Modular agents for planning, research, coding, and evaluation.
• —Tool Augmentation: Use web search, Wikipedia, file management, Python REPL, and more.
• —Retrieval-Augmented Generation (RAG):
• —Upload PDF files and ask questions about their content.
• —Uses FAISS vector database for semantic search and context injection.
• —Google Calendar Integration:
• —Create and list calendar events using natural language.
• —Detailed Reports & Summaries:
• —Generate detailed reports, summaries, and answers based on user queries, PDF content, and research tools.
• —Gradio Web UI:
• —Chat interface, PDF upload, and event management in your browser.

Setup Instructions

1. Clone the Repository

git clone https://github.com/rohithsiddi/AgentFlow-Multi-Agent-Workflow-Orchestration-Assistant.git
cd AgentFlow

2. Install Dependencies

Make sure you have Python 3.10+ and pip installed.

pip install -r requirements.txt

3. (Optional) Set Up Google Calendar API

• —Go to Google Cloud Console
• —Create a project, enable the Google Calendar API
• —Create OAuth credentials (Desktop app), download client_secret.json
• —Place client_secret.json in the project root
• —The first time you use calendar features, you will be prompted to authenticate and a token.json will be created
• —Set these in your .env (or use defaults):

  GOOGLE_TOKEN_PATH=token.json
  GOOGLE_CALENDAR_ID=primary

4. Run the App

python app.py

The app will open in your browser.

.env Configuration

Create a .env file in the project root to configure environment variables. Example:

# Google Calendar API
GOOGLE_TOKEN_PATH=token.json         # Path to your Google OAuth token
GOOGLE_CALENDAR_ID=primary          # Calendar ID (usually 'primary')

# OpenAI API (required for LLMs)
OPENAI_API_KEY=your_openai_api_key

# Serper API (for web search tool)
SERPER_API_KEY=your_serper_api_key

# LangSmith (LangChain analytics and tracing, optional)
LANGCHAIN_API_KEY=your_langsmith_api_key

# File Management Tool
FILE_TOOL_ROOT=reports              # Root directory for file management tools

# Pushover Notifications (optional)
PUSHOVER_TOKEN=your_pushover_token
PUSHOVER_USER=your_pushover_user

# Other environment variables can be added as needed for custom tools

• —If you use Google Calendar, set up the first two variables.
• —For LLM and agent features, set your OpenAI API key.
• —For web search, set your Serper API key.
• —For LangSmith analytics/tracing, set your LangSmith API key.
• —If you use file management tools, set FILE_TOOL_ROOT.
• —For push notifications, set the Pushover variables.

Usage

PDF RAG

• —Upload a PDF using the UI.
• —Ask questions about its content.
• —The assistant will cite the PDF as the source when using its content.

Detailed Reports & Summaries

• —Ask the assistant to generate detailed reports or summaries on topics, documents, or research queries.
• —The assistant leverages multi-agent reasoning, RAG, and research tools to produce comprehensive outputs.

Google Calendar

• —Ask to create or list calendar events (e.g., "Create a meeting tomorrow at 10am.", "List my next 3 events.")

General Chat & Tools

• —Use the chat for planning, research, coding, and more.
• —The assistant can use web search, Wikipedia, Python REPL, and file management tools.

Folder Structure

• —app.py — Gradio UI and app entry point
• —agentflow.py — Main agent orchestration logic
• —agentflow_tools.py — Tool definitions (calendar, RAG, web, etc.)