Twitter Agent

This tutorial guides you through creating a Twitter agent that automatically posts tweets at regular intervals using the Galadriel framework.

​

Prerequisites

• Python 3.10 or higher
• A Twitter developer account with API keys
• An OpenAI API key

​

Step 1: Set Up the Development Environment

1. Create a virtual environment:

   python3 -m venv venv                                                                                                                         
   source venv/bin/activate                                                                                                                     
   

2. Install Galadriel:

   pip install galadriel                                                                                                                        
   

​

Step 2: Obtain Twitter API Keys

1. Create a Twitter developer account at https://developer.x.com/.
2. Create a new app and obtain the following credentials:
   • Consumer API Key
   • Consumer API Secret
   • Access Token
   • Access Token Secret Ensure that the user authentication is set up with write access.

​

Step 3: Set Up Environment Variables

1. Create a file named .env in your project directory.

2. Add the following environment variables to the .env file:

   OPENAI_API_KEY=<YOUR_OPENAI_API_KEY>                                                                                                         
   TWITTER_CONSUMER_API_KEY=<YOUR_TWITTER_CONSUMER_API_KEY>                                                                                     
   TWITTER_CONSUMER_API_SECRET=<YOUR_TWITTER_CONSUMER_API_SECRET>                                                                               
   TWITTER_ACCESS_TOKEN=<YOUR_TWITTER_ACCESS_TOKEN>                                                                                             
   TWITTER_ACCESS_TOKEN_SECRET=<YOUR_TWITTER_ACCESS_TOKEN_SECRET>                                                                               
   

   Replace the placeholder values with your actual API keys and tokens.

   Optionally, to skip the actual Twitter posting and just print the results, add:

   DRY_RUN=true                                                                                                                                 
   

​

Step 4: Create the Twitter Agent

1. Create a file named twitter_agent.py with the following code:

   import asyncio                                                                                                                               
   import os                                                                                                                                    
   from pathlib import Path                                                                                                                     
                                                                                                                                               
   from dotenv import load_dotenv                                                                                                               
                                                                                                                                               
   from galadriel import CodeAgent                                                                                                              
   from galadriel import AgentRuntime                                                                                                           
   from galadriel.clients import Cron                                                                                                           
   from galadriel.core_agent import LiteLLMModel                                                                                                
   from galadriel.tools.twitter import TwitterPostTool                                                                                          
                                                                                                                                               
   load_dotenv(dotenv_path=Path(".") / ".env", override=True)                                                                                   
   llm_model = LiteLLMModel(model_id="gpt-4o", api_key=os.getenv("OPENAI_API_KEY"))                                                             
                                                                                                                                               
   POST_INTERVAL_SECONDS = 3 * 60 * 60  # 3 hours                                                                                               
                                                                                                                                               
   AGENT_PROMPT = """                                                                                                                           
   You are a humorous Twitter user.                                                                                                             
   Every time you are called:                                                                                                                   
   1. Generate a short tweet (1-2 sentences). About any topic.                                                                                  
   2. Post the tweet.                                                                                                                           
   """                                                                                                                                          
                                                                                                                                               
   agent = CodeAgent(                                                                                                                           
       prompt_template=AGENT_PROMPT,                                                                                                            
       model=llm_model,                                                                                                                         
       tools=[TwitterPostTool()],                                                                                                               
   )                                                                                                                                            
                                                                                                                                               
   runtime = AgentRuntime(                                                                                                                      
       agent=agent,                                                                                                                             
       inputs=[Cron(POST_INTERVAL_SECONDS)],                                                                                                    
       outputs=[],  # No output, posting happens inside Agent                                                                                   
   )                                                                                                                                            
                                                                                                                                               
   asyncio.run(runtime.run())                                                                                                                   
   

   Key components:

   • LiteLLMModel: Loads the OpenAI GPT-4o model.
   • TwitterPostTool: Enables posting tweets to Twitter.
   • CodeAgent: Creates the agent with the specified prompt and tools.
   • Cron: Triggers the agent to run every 3 hours.
   • AgentRuntime: Manages the execution of the agent.

​

Step 5: Run the Agent

Execute the script:

The agent will now run automatically, generating and posting tweets every 3 hours. If DRY_RUN=true is set, it will print the generated tweets
to the console instead of posting them to Twitter.

​

Example using tweet from output client

Create a file named twitter_agent.py with the following code:

import asyncio                                                                                                                               
import os                                                                                                                                    
from pathlib import Path                                                                                                                     
                                                                                                                                            
from dotenv import load_dotenv                                                                                                               
                                                                                                                                            
from galadriel import CodeAgent                                                                                                              
from galadriel import AgentRuntime                                                                                                           
from galadriel.clients import Cron                                                                                                           
from galadriel.clients.twitter_post_client import TwitterPostClient                                                                          
from galadriel.core_agent import LiteLLMModel                                                                                                
                                                                                                                                            
load_dotenv(dotenv_path=Path(".") / ".env", override=True)                                                                                   
llm_model = LiteLLMModel(model_id="gpt-4o", api_key=os.getenv("OPENAI_API_KEY"))                                                             
                                                                                                                                            
POST_INTERVAL_SECONDS = 5                                                                                                                    
                                                                                                                                            
AGENT_PROMPT = """                                                                                                                           
You are a humorous Twitter user.                                                                                                             
Generate a short tweet (1-2 sentences). About any topic.                                                                                     
"""                                                                                                                                          
                                                                                                                                            
agent = CodeAgent(                                                                                                                           
    prompt_template=AGENT_PROMPT,                                                                                                            
    model=llm_model,                                                                                                                         
    tools=[],                                                                                                                                
)                                                                                                                                            
                                                                                                                                            
runtime = AgentRuntime(                                                                                                                      
    agent=agent,                                                                                                                             
    inputs=[Cron(POST_INTERVAL_SECONDS)],                                                                                                    
    outputs=[TwitterPostClient()],                                                                                                           
)                                                                                                                                            
                                                                                                                                            
asyncio.run(runtime.run())                                                                                                                   

​

Conclusion

You have now created a fully autonomous Twitter agent using the Galadriel framework. This agent runs continuously, generating and posting
tweets without any manual intervention. You can further customize the agent by modifying the prompt, adding more tools, and adjusting the
posting interval.