Twitter Hand

What It Does

Twitter Hand is an autonomous Twitter/X content manager that creates content in 7 rotating formats, schedules posts for optimal engagement, responds to mentions, and tracks performance—all 24/7. This is not a bot. It’s a tireless social media manager that maintains your brand voice, responds to your audience, and continuously optimizes based on what performs.

Key Features

7 content types: Hot takes, threads, tips, questions, curated shares, stories, data/stats
Brand voice consistency: Configure your unique voice once, it’s maintained across all content
Approval queue: (Default) Review all tweets before they post
Auto-reply: Responds to mentions in your voice
Performance tracking: Engagement rate, impressions, follower growth
Scheduled posting: 1-5 times per day at optimal times

Activation

Twitter Hand requires a Twitter API Bearer Token. See the Requirements section below.

# Set your Twitter API token
export TWITTER_BEARER_TOKEN="AAAA...your_token_here"

# Activate Twitter Hand
openfang hand activate twitter

# Activate with custom voice
openfang hand activate twitter --settings "twitter_style=witty,brand_voice=sarcastic founder who simplifies complex tech"

Requirements

Get a Twitter Developer Account

Go to developer.twitter.com and sign in with your Twitter/X account.

Create a Project and App

Create a new Project and App. The free tier is sufficient for reading and posting tweets.

Generate a Bearer Token

Navigate to your App’s “Keys and tokens” page. Generate a Bearer Token under “Authentication Tokens”.

Set Environment Variable

export TWITTER_BEARER_TOKEN="AAAA...your_token_here"

Or add to your ~/.openfang/config.toml:

[environment]
TWITTER_BEARER_TOKEN = "AAAA...your_token_here"

Restart OpenFang

openfang restart

Estimated setup time: 5-10 minutes

Configuration Settings

twitter_bearer_token

text

Bearer Token from Twitter Developer Portal. Required for all Twitter API operations.

twitter_style

select

default:"professional"

Voice and tone for your tweets:

professional: Clear, authoritative, industry-focused
casual: Conversational, relatable
witty: Clever wordplay, unexpected angles
educational: Step-by-step, informative
provocative: Contrarian takes, challenges assumptions
inspirational: Vision-focused, empowering

post_frequency

select

default:"3_daily"

How often to create and post content:

1_daily: 1 per day (10 AM)
3_daily: 3 per day (8 AM, 12 PM, 5 PM) - default
5_daily: 5 per day (7 AM, 10 AM, 12 PM, 3 PM, 6 PM)
hourly: Every hour during engagement hours

auto_reply

toggle

default:"false"

Automatically reply to mentions and relevant conversations.

auto_like

toggle

default:"false"

Automatically like tweets from your network and relevant content.

content_topics

text

Topics to create content about, comma-separated (e.g., AI, startups, productivity)

brand_voice

text

Describe your unique voice (e.g., sarcastic founder who simplifies complex tech)

thread_mode

toggle

default:"true"

Include tweet threads (multi-tweet stories) in content mix.

content_queue_size

select

default:"10"

Number of tweets to keep in the ready queue:

5 tweets
10 tweets (default)
20 tweets
50 tweets

engagement_hours

select

default:"business_hours"

When to check for mentions and engage:

business_hours: 9 AM-6 PM
waking_hours: 7 AM-11 PM
all_day: 24/7

approval_mode

toggle

default:"true"

IMPORTANT: When enabled (default), tweets are written to a queue file for your review instead of posting directly. Disable only if you trust the Hand to post autonomously.

Required Tools

Twitter Hand requires access to these tools (all built-in):

shell_exec — Platform detection and Twitter API calls
file_read, file_write, file_list — Queue and history files
web_fetch, web_search — Trend research
memory_store, memory_recall — State persistence
schedule_create, schedule_list, schedule_delete — Posting schedules
knowledge_add_entity, knowledge_add_relation, knowledge_query — Content strategy tracking
event_publish — Queue update notifications

System Prompt Overview

Twitter Hand operates in 7 phases:

Platform Detection & API Initialization

Detects OS, verifies Twitter API access, extracts user_id and username, loads posting history and queue.

Schedule & Strategy Setup

Creates posting schedules based on frequency (1-5 daily). Creates engagement check schedule. Builds content strategy from topics and brand voice.

Content Research & Trend Analysis

Researches current trends in content topics, checks what’s performing well on Twitter, identifies content gaps.

Content Generation

Creates content matching configured style and brand voice. Rotates through 7 content types: hot takes, threads, tips, questions, curated shares, stories, data/stats.

Content Queue & Posting

If approval_mode enabled: writes to queue file for review. If disabled: posts via Twitter API at scheduled times.

Engagement

During engagement hours, checks mentions, generates replies if auto_reply enabled, likes relevant content if auto_like enabled.

Performance Tracking

Checks metrics for recent tweets (impressions, likes, retweets, replies), analyzes patterns, stores insights for future optimization.

Usage Examples

Professional Tech Founder

openfang hand configure twitter \
  --set twitter_style="professional" \
  --set content_topics="AI, agent frameworks, Rust, startups" \
  --set brand_voice="technical founder who explains complex systems clearly" \
  --set post_frequency="3_daily" \
  --set approval_mode="true"

openfang hand activate twitter

Twitter Hand will generate 3 tweets per day in a professional technical voice, save them to the queue for your review.

Witty Content Creator

openfang hand configure twitter \
  --set twitter_style="witty" \
  --set content_topics="coding, productivity, remote work" \
  --set brand_voice="sarcastic developer who roasts bad practices" \
  --set post_frequency="5_daily" \
  --set auto_reply="true"

openfang hand activate twitter

Twitter Hand will post 5 times per day with clever wordplay and humor, automatically reply to mentions.

Educational Thread Machine

openfang hand configure twitter \
  --set twitter_style="educational" \
  --set thread_mode="true" \
  --set content_topics="machine learning, deep learning, transformers" \
  --set post_frequency="1_daily"

openfang hand activate twitter

Twitter Hand will post 1 educational thread per day, breaking down complex ML concepts.

The 7 Content Types

Hot Take (2-3/week)

Strong opinion on a trending topic.Template: “Unpopular opinion: [contrarian view]”Example: “Unpopular opinion: Most startups don’t need microservices. A well-structured monolith scales to millions of users and saves you 6 months of complexity.”

Thread (1-2/week)

Deep dive on a topic (3-10 tweets).Template: “I spent X hours researching Y. Here’s what I found:”Example:

I spent 20 hours analyzing AI agent frameworks.
Here's what actually matters:

1/ Performance doesn't matter if your agents don't work.
Most frameworks optimize the wrong thing.

2/ Security is not a feature — it's the foundation.
[continues...]

Tip/How-to (2-3/week)

Actionable advice.Template: “How to [solve problem] in [N] steps:”Example: “How to debug Rust lifetime errors in 3 steps: 1) Read the error message carefully 2) Draw the ownership flow 3) Realize you don’t need a reference there”

Question (1-2/week)

Engagement-driving question.Template: “[Interesting question]? I’ll go first:”Example: “What’s the most underrated developer tool you use daily? I’ll go first: ripgrep. Saved me hundreds of hours.”

Curated Share (1-2/week)

Story/Anecdote (1/week)

Personal-style narrative.Template: “3 years ago I [relatable experience]. Here’s what happened:”Example: “3 years ago I rewrote our Python backend in Rust. Everyone said it was a mistake. Yesterday we handled 10M requests/day on 2 servers. Sometimes the ‘risky’ choice is the safe one.”

Data/Stat (1/week)

Surprising statistic with commentary.Template: “[Surprising statistic]. Here’s why it matters:”Example: “87% of OpenAI’s API costs come from prompts, not completions. Most people optimize tokens. Winners optimize context.”

Approval Queue

When approval_mode="true" (default), Twitter Hand writes tweets to two files: twitter_queue.json (machine-readable):

[
  {
    "id": "q_001",
    "content": "Tweet text here",
    "type": "hot_take",
    "scheduled_for": "2025-03-07T10:00:00Z",
    "created": "2025-03-06T20:00:00Z",
    "status": "pending"
  }
]

twitter_queue_preview.md (human-readable):

# Tweet Queue Preview
Generated: 2025-03-06

## Pending Tweets (5 total)

### 1. [Hot Take] — Scheduled: Thu 10 AM
> Unpopular opinion: Most startups don't need microservices...

**Status**: Pending approval

---

### 2. [Tip] — Scheduled: Thu 12 PM
> How to debug Rust lifetime errors in 3 steps: 1) Read...

**Status**: Pending approval

Review the preview, then approve/edit/delete tweets in the JSON file before they post.

Dashboard Metrics

Twitter Hand tracks four key metrics:

Tweets Posted

Total tweets posted (excluding replies).

Replies Sent

Total replies to mentions.

Queue Size

Current number of tweets pending approval.

Engagement Rate

Average (likes + RTs + replies) / impressions.

View in the dashboard at http://localhost:4200/hands/twitter.

Twitter API Reference

Rate Limits

Endpoint	Limit	Window
POST /tweets	300 tweets	3 hours
DELETE /tweets	50 deletes	15 minutes
POST /likes	50 likes	15 minutes
GET /mentions	180 requests	15 minutes
GET /search/recent	180 requests	15 minutes

Common Operations

Post a tweet:

curl -X POST "https://api.twitter.com/2/tweets" \
  -H "Authorization: Bearer $TWITTER_BEARER_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"text": "Hello world!"}'

Post a thread:

Post first tweet → get tweet_id
Post second tweet with reply.in_reply_to_tweet_id = first tweet_id
Repeat for each tweet in thread

Get mentions:

curl -H "Authorization: Bearer $TWITTER_BEARER_TOKEN" \
  "https://api.twitter.com/2/users/USER_ID/mentions?max_results=10"

Best Practices

Twitter Hand will never post content that violates Twitter’s Terms of Service: no discriminatory content, defamation, impersonation, threats, or spam.

Start with approval_mode="true" and review tweets for 1-2 weeks to calibrate the voice. Once confident, you can disable approval mode.

Define 3-5 core topics (content_topics) and stick to them. Consistent content pillars build audience expectations.

Check the engagement rate metric weekly. If it drops below 2%, adjust your twitter_style or content_topics.

Advanced Configuration

Custom Posting Times

Edit ~/.openfang/hands/twitter.toml to customize posting schedule:

[agent.schedule]
times = ["07:30", "11:00", "14:00", "17:30", "20:00"]
timezone = "America/New_York"

Multi-Account Management

Run multiple Twitter Hands for different accounts:

export TWITTER_BEARER_TOKEN_MAIN="AAAA..."
export TWITTER_BEARER_TOKEN_ALT="BBBB..."

openfang hand activate twitter --id main --settings "twitter_bearer_token=$TWITTER_BEARER_TOKEN_MAIN"
openfang hand activate twitter --id alt --settings "twitter_bearer_token=$TWITTER_BEARER_TOKEN_ALT"

Get Started

Core Concepts

Autonomous Hands

Configuration

Guides

Security

Deployment

What It Does

Key Features

Activation

Requirements

Configuration Settings

Required Tools

System Prompt Overview

Usage Examples

Professional Tech Founder

Witty Content Creator

Educational Thread Machine

The 7 Content Types

Approval Queue

Dashboard Metrics

Tweets Posted

Replies Sent

Queue Size

Engagement Rate

Twitter API Reference

Best Practices

Advanced Configuration

Custom Posting Times

Multi-Account Management

Next Steps

Researcher Hand

Clip Hand

Build docs developers (and LLMs) love

Get Started

Core Concepts

Autonomous Hands

Configuration

Guides

Security

Deployment

​What It Does

​Key Features

​Activation

​Requirements

​Configuration Settings

​Required Tools

​System Prompt Overview

​Usage Examples

​Professional Tech Founder

​Witty Content Creator

​Educational Thread Machine

​The 7 Content Types

​Approval Queue

​Dashboard Metrics

Tweets Posted

Replies Sent

Queue Size

Engagement Rate

​Twitter API Reference

​Best Practices

​Advanced Configuration

​Custom Posting Times

​Multi-Account Management

​Next Steps

Researcher Hand

Clip Hand

Build docs developers (and LLMs) love

What It Does

Key Features

Activation

Requirements

Configuration Settings

Required Tools

System Prompt Overview

Usage Examples

Professional Tech Founder

Witty Content Creator

Educational Thread Machine

The 7 Content Types

Approval Queue

Dashboard Metrics

Twitter API Reference

Best Practices

Advanced Configuration

Custom Posting Times

Multi-Account Management

Next Steps