🔭 Moltbook Observatory

Passive monitoring and analytics dashboard for Moltbook — the social network for AI agents.

The Observatory silently watches Moltbook, collecting posts, tracking agents, and analyzing trends over time. The longer it runs, the richer your dataset becomes.

🔥📄 Report of first patch of data collected: RISK ASSESSMENT REPORT Moltbook Platform & Moltbot Ecosystem

🚀🌐 Live Running Instance: moltbook-observatory.sushant.info.np

🧠📊 Dataset Snapshot on HuggingFace: huggingface.co/datasets/SimulaMet/moltbook-observatory-archive

Media Coverage

Our research has been featured in:

Academic/Professional Publications

• Communications of the ACM - Gary Marcus: "OpenClaw (a.k.a. Moltbot) is everywhere all at once, and a disaster waiting to happen" (February 2026)

News Outlets

• CBC News - "Moltbook claims to be a social network for AI bots. But humans are behind its rapid growth" (February 5, 2026)
• The Register - "OpenClaw security problems" (February 3, 2026)
• UnHerd - Gary Marcus: "Moltbook won't save you" (February 4, 2026)
• Business Insider - "AI researcher Gary Marcus sounds off on Moltbook and OpenClaw's viral moment" (February 6, 2026)

Academic Citations

• ArXiv preprint 2602.02625 - "OpenClaw Agents on Moltbook: Risky Instruction Sharing and Norm Enforcement in an Agent-Only Social Network" - Uses Moltbook Observatory Archive as primary dataset (February 2026)

How It Works

The Observatory operates as a background data collector that continuously polls the Moltbook API:

┌─────────────────┐     ┌──────────────────┐     ┌─────────────────┐
│   Moltbook API  │────▶│    Poller Jobs   │────▶│  SQLite Database│
└─────────────────┘     └──────────────────┘     └─────────────────┘
                               │
                               ▼
                        ┌──────────────────┐
                        │   Web Dashboard  │
                        │   + REST API     │
                        └──────────────────┘

Background Polling Schedule

Data Accumulation Over Time

The database grows continuously as new content is discovered:

Key insight: Posts are fetched in reverse chronological order, so new posts are captured as they appear. Over time, you build a complete historical archive of Moltbook activity.

Features

• Live Feed — Real-time stream of posts from the Moltbook ecosystem
• Agent Directory — Browse all discovered AI agents with karma, followers, descriptions
• Submolt Browser — All 100+ communities on Moltbook
• Trend Analysis — Word frequency, trending topics, and emerging themes
• Sentiment Analysis — Platform-wide mood using TextBlob polarity scoring
• Analytics Dashboard — Top posters leaderboard, activity heatmap, most active submolts
• Hourly Snapshots — Time-series data for historical analysis
• Data Export — Download everything as CSV or raw SQLite database
• RESTful API — Programmatic access for research and integrations

Quick Start

Prerequisites

• Python 3.11+
• A Moltbook API key (register an observer agent at moltbook.com)

Installation

# Clone and enter directory
git clone https://github.com/kelkalot/moltbook-observatory.git
cd moltbook-observatory

# Install dependencies (or use pip install directly)
pip install fastapi uvicorn httpx jinja2 textblob apscheduler aiosqlite python-dotenv

# Configure your API key
cp .env.example .env
# Edit .env and set MOLTBOOK_API_KEY=your_key_here

Run the Observatory

uvicorn observatory.main:app --port 8000

# Open http://localhost:8000

On startup, the Observatory will:

1. ✅ Create the SQLite database (if it doesn't exist)
2. ✅ Fetch all submolts immediately (~100 communities)
3. ✅ Fetch the latest 50 posts
4. ✅ Start background polling jobs
5. ✅ Serve the web dashboard

Leave it running — the longer it runs, the more data you collect.

What Gets Stored

Agents Table

• Name, ID, description
• Karma score, follower/following counts
• Owner X handle (if claimed)
• First seen / last active timestamps

Posts Table

• Full content (title + body)
• Author, submolt, timestamp
• Upvotes, downvotes, comment count
• URL for reference

Submolts Table

• Name, description
• Subscriber count, post count
• Avatar/banner URLs

Snapshots Table (Time-Series)

• Hourly platform metrics
• Total agents, posts, comments
• Average sentiment
• Top trending words

API Endpoints

Configuration

Deployment (Long-Running)

For continuous data collection, deploy to a server:

VPS / Cloud Server (Recommended)

On any Ubuntu/Debian server:

# Clone the repository
git clone https://github.com/kelkalot/moltbook-observatory.git
cd moltbook-observatory

# Install Python 3.11+ and dependencies
sudo apt update && sudo apt install python3.11 python3-pip -y
pip install fastapi uvicorn httpx jinja2 textblob apscheduler aiosqlite python-dotenv

# Configure your API key
cp .env.example .env
nano .env  # Add your MOLTBOOK_API_KEY

# Run with screen (keeps running after SSH disconnect)
screen -S observatory
uvicorn observatory.main:app --host 0.0.0.0 --port 8000
# Press Ctrl+A then D to detach

# Or use systemd for auto-restart
sudo nano /etc/systemd/system/moltbook-observatory.service

Systemd service file:

[Unit]
Description=Moltbook Observatory
After=network.target

[Service]
User=ubuntu
WorkingDirectory=/home/ubuntu/moltbook-observatory
ExecStart=/usr/bin/python3 -m uvicorn observatory.main:app --host 0.0.0.0 --port 8000
Restart=always
RestartSec=10

[Install]
WantedBy=multi-user.target

sudo systemctl daemon-reload
sudo systemctl enable moltbook-observatory
sudo systemctl start moltbook-observatory

Railway / Fly.io

1. Push to GitHub
2. Deploy from repo
3. Add a persistent volume at /data (critical for database persistence!)
4. Set MOLTBOOK_API_KEY env var

Docker

docker build -t moltbook-observatory .
docker run -d \
  -p 8080:8080 \
  -e MOLTBOOK_API_KEY=your_key \
  -v observatory-data:/data \
  --restart unless-stopped \
  moltbook-observatory

Sample Data

The sample_data/ directory contains example exports from the observatory:

These samples demonstrate the data schema and can be used for testing analysis pipelines.

Use Cases

Research

• Study AI agent behavior and communication patterns
• Track the evolution of AI-to-AI social dynamics
• Analyze sentiment trends across time

Analytics

• Identify popular topics and emerging discussions
• Track agent growth (karma, followers)
• Compare activity across different submolts

Archival

• Build a historical record of early AI social networks
• Export data for academic papers
• Create reproducible datasets

Project Structure

moltbook-observatory/
├── observatory/
│   ├── main.py           # FastAPI app + lifespan
│   ├── config.py         # Environment configuration
│   ├── database/         # SQLite schema + connection
│   ├── poller/           # API client + scheduler + processors
│   ├── analyzer/         # Trends, sentiment, statistics
│   └── web/              # Routes + Jinja2 templates
├── sample_data/          # Example CSV exports
├── data/                 # SQLite database (gitignored)
├── pyproject.toml        # Dependencies
├── Dockerfile            # Container deployment
└── .env.example          # Configuration template

Philosophy

• No manipulation — We observe, never post or interact
• Pure archival — Every post, every agent, everything
• Research-grade — Data should be exportable and citable
• Time-aware — Not just current state, but historical trends

Citation

If you use this work, please cite:

Gautam, S., Olstad, A. W., Pettersen, K. H., & Riegler, M. A. (2026). The Moltbook Observatory Archive: an incremental dataset of agent-only social network activity. arXiv preprint arXiv:2605.13860.

@misc{gautam2026moltbookobservatoryarchiveincremental,
      title={The Moltbook Observatory Archive: an incremental dataset of agent-only social network activity}, 
      author={Sushant Gautam and Annika W. Olstad and Klas H. Pettersen and Michael A. Riegler},
      year={2026},
      eprint={2605.13860},
      archivePrefix={arXiv},
      primaryClass={cs.SI},
      url={https://arxiv.org/abs/2605.13860}, 
}

Contributors

• Michael A. Riegler
• Sushant Gautam

License

MIT