This project is an AI-powered system that automatically generates Twitter posts in the style of any given Twitter user — just give it a username and a topic!
Inspired by Miguel Otero Pedrido's CrewAI-based LinkedIn automation, this implementation uses Agno — a lightweight, modular framework for building multimodal AI agents — to build an intelligent tweeting agent.
Given a Twitter username and a topic, the system will:
- Scrape tweets from the user to understand their tone, style, and language.
- Research trending info on your selected topic.
- Generate a new tweet using the user's style and the latest insights!
-
Twitter Scraper Agent
- Scrapes the latest tweets from a user's public Twitter profile
- Learns their tone, writing style, and common phrases
-
Web Researcher Agent
- Uses online search tools to collect recent, relevant information on your given topic
-
Doppelgänger Agent
- Uses insights from the above agents to generate tweets that look and feel like they're from the selected user
All agents are powered by Groq-hosted LLMs (e.g., LLaMA 3, Gemma 2).
. ├── agents.py # Defines the three AI agents (scraper, researcher, creator) ├── main.py # Entry point, runs the multi-agent sequence ├── twitter.py # Twitter scraping logic using Selenium ├── tools/ # (Optional) Extra utility tools ├── .env # Stores your Twitter login info