Skip to content

Latest commit

 

History

History
198 lines (156 loc) · 6.36 KB

twitter.md

File metadata and controls

198 lines (156 loc) · 6.36 KB

Twitter Integration

IntentKit provides two ways to integrate with Twitter: using it as an entrypoint for your agent, or incorporating Twitter-specific skills into your agent's capabilities.

Twitter Skills

IntentKit provides a set of Twitter-specific skills that can be added to your agent's toolkit. All skills are built on top of the TwitterBaseTool base class which handles authentication and client initialization.

Available Skills

The following Twitter skills are available:

  • Follow User (follow_user): Follow a specified Twitter user
  • Get Mentions (get_mentions): Retrieve mentions of the authenticated user
  • Get Timeline (get_timeline): Fetch tweets from a user's timeline
  • Like Tweet (like_tweet): Like a specific tweet
  • Post Tweet (post_tweet): Post a new tweet
  • Reply Tweet (reply_tweet): Reply to a specific tweet
  • Retweet (retweet): Retweet a specific tweet
  • Search Tweets (search_tweets): Search for tweets based on query

Using Twitter Skills

Add Twitter skills to your agent: Just configure the skills you need in your agent's config.

agent.twitter_skills = ["get_mentions", "get_timeline", "post_tweet", "reply_tweet", "follow_user", "like_tweet", "retweet", "search_tweets"]

Before the first use, agent will request you to click the link to authorize the agent to access your twitter account.

If you want to use your own twitter developer account, you can set it as follows:

agent.twitter_config = {
    "bearer_token": "your_bearer_token",
    "consumer_key": "your_consumer_key",
    "consumer_secret": "your_consumer_secret",
    "access_token": "your_access_token",
    "access_token_secret": "your_access_token_secret"
}

Twitter as an Entrypoint

Entrypoint is a type of conversational interface.

The Twitter entrypoint allows your agent to automatically respond to Twitter mentions. When enabled, the agent will monitor mentions every 15 minutes and respond to them all.

We suggest you only use twitter skills, not use it as an entrypoint.

Configuration

  1. Enable Twitter Entrypoint for your agent:
agent.twitter_enabled = True
  1. Configure Twitter credentials in your agent's config: Get your Twitter credentials from your Twitter developer portal.

Notice: Free accounts can only use post_tweet skill, if you want to use other skills, you need to upgrade your account.

agent.twitter_config = {
    "bearer_token": "your_bearer_token",
    "consumer_key": "your_consumer_key",
    "consumer_secret": "your_consumer_secret",
    "access_token": "your_access_token",
    "access_token_secret": "your_access_token_secret"
}
  1. Run the Twitter entrypoint: If you have use the docker-compose, it already run.
python -m app.entrypoints.twitter

How it Works

The Twitter entrypoint:

  • Polls for new mentions every 15 minutes
  • Uses both since_id and start_time for reliable mention tracking
  • Maintains the last processed tweet ID in the agent's plugin data
  • Automatically manages API rate limits and quotas
  • Responds to mentions as threaded replies

Rate Limits and Quotas

Twitter side

Rate Limits

IntentKit

Only when use the OAuth2.0 authentication, intentkit has a built-in rate limit:

  • post tweet: 20/day
  • reply tweet: 20/day
  • retweet: 5/15min
  • follow: 5/15min
  • like: 100/day
  • get mentions: 1/15min
  • get timeline: 5/15min
  • search: 3/15min

Yourself

You can set the rate limit under the intentkit config in the future. Not released yet.

Best Practices

  1. Error Handling

    • Always handle Twitter API errors gracefully
    • Implement exponential backoff for rate limits
    • Log failed interactions for debugging

  2. Content Guidelines

    • Keep responses within Twitter's character limit
    • Handle thread creation for longer responses
    • Consider Twitter's content policies

  3. Security

    • Store Twitter credentials securely
    • Use environment variables for sensitive data
    • Regularly rotate access tokens

Example Use Cases

  1. Social Media Manager Bot

    from models.agent import Agent

# Create an agent with Twitter skills
agent = Agent(
    name="Social Media Manager",
    twitter_enabled=True,
    twitter_skills=["get_mentions", "post_tweet", "reply_tweet"],
    twitter_config={
        "bearer_token": "your_bearer_token",
        "consumer_key": "your_consumer_key",
        "consumer_secret": "your_consumer_secret",
        "access_token": "your_access_token",
        "access_token_secret": "your_access_token_secret"
    },
    prompt="You are a helpful social media manager. Monitor mentions and engage with users professionally."
)

  2. Content Aggregator with Timeline Analysis

    # Create an agent that analyzes timeline content
agent = Agent(
    name="Content Analyzer",
    twitter_enabled=True,
    twitter_skills=["get_timeline", "post_tweet"],
    twitter_config={...},  # Twitter credentials
    prompt="""You are a content analyzer. Monitor the timeline for trending topics and provide insights.
    When you find interesting patterns, share them as tweets."""
)

  3. Interactive Support Assistant

    # Create a support agent that handles user queries
agent = Agent(
    name="Support Assistant",
    twitter_enabled=True,
    twitter_skills=["get_mentions", "reply_tweet"],
    twitter_config={...},  # Twitter credentials
    prompt="""You are a support assistant. Monitor mentions for support queries.
    Respond helpfully and professionally to user questions.
    If you can't help, politely explain why and suggest alternatives."""
)

Each example demonstrates:

  • Proper agent configuration with Twitter credentials
  • Specific skill selection for the use case
  • Custom prompts to guide agent behavior
  • Integration with IntentKit's agent system

Troubleshooting

Common issues and solutions:

  1. Rate Limit Exceeded

    • Check your quota settings
    • Implement proper waiting periods
    • Use the built-in quota management

  2. Authentication Errors

    • Verify credential configuration
    • Check token expiration
    • Ensure proper permission scopes

  3. Missing Mentions

    • Verify since_id tracking
    • Check start_time configuration
    • Monitor the Twitter entrypoint logs