Autogen Research Assistant

Building "Useful" Agents with Composio and Autogen

This is a starter repository for building agents with [Composio](composio.dev) and Autogen. You can use this codebase to update and build your own dynamic agents.

Usecase

This is an example for integrating Slack messages, conducting research, generating detailed analyses, and automating report creation in Notion.

Overview

Project execution follows this workflow:

• Using Composio to get Slack messages. Every time a Slack message is sent, a trigger is raised which is received by Composio and sent to the server.
• On Receiving the trigger, Autogen Agents are initiated
• Using Autogen to build two agents: one for researching the topic and the second for writing a detailed analysis on the topic.
• Using Composio to write the detailed report in Notion, sending a confirmation message to Slack.

1. Autogen powers the orchestration of the agents and tools.
2. Composio provides Autogen the right tools which allows connecting to Slack (Receiving the message, sending a message to channel), Notion (Creating a page)

Here's a visual representation of the workflow:

Below is the basic setup you can do to run the repository.

Setup

1. **Environment Setup**

   Add a `.env` file and insert your OPENAI_API_KEY and Slack Channel ID. For Slack, identify the channel you wish to monitor. Navigate to the channel about section, and scroll to the bottom to locate the Channel ID.

   cp .env.sample .env

   **Note**: No need to add `TRIGGER_ID` in the `.env` file right now; we'll get to that in the next steps.

2. **Run setup script**

   This script installs necessary dependencies and tools for running the project.

   • Installs pip and project dependencies.
   • Prompt's you to install ngrok and guides you on how to do that.
   • Activates the venv shell.

   You can run this script to automate the setup process for your development environment.

   NB: Required Python version (>=3.10, <=3.13)

   ``