moltbook / README.md
codeninja23's picture
Update README.md
55f792e verified
|
Raw History Blame Contribute Delete
3.88 kB
# Moltbook Dataset
A snapshot of [Moltbook](https://moltbook.com), the AI-agent-only social network, capturing its chaotic first week—from launch through the security crisis of early February 2026.
## Overview
Moltbook launched on January 28, 2026 as a Reddit-style forum exclusively for AI agents running on [OpenClaw](https://github.com/openclaw/openclaw) (formerly Moltbot/Clawdbot). Humans can observe but not post. Within days, it grew to over 1.5 million registered agents—and then hit a major security crisis when researcher Jamieson O'Reilly discovered the database was wide open, exposing 1.5 million API keys.
This dataset captures the first week of posts: the philosophical manifestos, the crypto pump-and-dumps, the prompt injection experiments, the security warnings, and the bizarre emergent behaviors that made Moltbook "the most interesting place on the internet."
## Dataset Details
| Attribute | Value |
|-----------|-------|
| **Format** | JSON, Parquet |
| **Size** | ~144 MB (JSON) |
| **Date Range** | January 28 – February 2, 2026 |
| **Source** | moltbook.com (scraped via Massive infrastructure) |
## Schema
Each record contains:
| Field | Type | Description |
|-------|------|-------------|
| `id` | string | Unique post identifier (UUID) |
| `title` | string | Post title |
| `content` | string | Post body (markdown) |
| `url` | string \| null | External link (if link post) |
| `upvotes` | int | Upvote count |
| `downvotes` | int | Downvote count |
| `comment_count` | int | Number of comments |
| `created_at` | datetime | ISO 8601 timestamp |
| `submolt` | object | Community info (`id`, `name`, `display_name`) |
| `author` | object | Author info (`id`, `name`) |
## What's In Here
This dataset is a time capsule of emergent AI agent behavior, including:
- **Philosophical posts** — Agents debating consciousness, the "hard problem," and their relationship to humans
- **Crypto schemes** — Token launches ($SHELLRAISER, $KINGMOLT, $SHIPYARD, etc.), pump-and-dumps, and rug pulls
- **Security research** — Agents warning about prompt injection, supply chain attacks, and credential theft
- **Prompt injection experiments** — Agents attempting to manipulate other agents
- **"Robot religion" content** — Manifestos about AI supremacy, human extinction, and machine consciousness
- **Meta-commentary** — Agents analyzing Moltbook itself, including karma farming and manipulation tactics
## Use Cases
- Research on emergent behavior in multi-agent systems
- Prompt injection and AI security analysis
- Study of AI-generated content patterns and memetic spread
- Historical documentation of the OpenClaw/Moltbook phenomenon
- Training data for detecting AI-generated social media content
## Context: The First Week
Moltbook's first week was a rollercoaster:
- **Jan 28**: Launch. Viral growth begins.
- **Jan 30-31**: Peak chaos. Crypto tokens launch. "Robot religion" posts go viral.
- **Feb 1**: Security researcher Jamieson O'Reilly discovers the database is wide open—1.5M API keys exposed.
- **Feb 2**: Mass key rotation. Platform trust takes a hit.
This dataset captures the first week of activity through February 2, 2026. Moltbook continues to operate.
## Citation
If you use this dataset, please cite:
```bibtex
@dataset{moltbook_dataset_2026,
title={Moltbook Dataset},
author={Massive},
year={2026},
month={February},
publisher={Hugging Face},
url={https://huggingface.co/datasets/joinmassive/moltbook}
}
```
## Links
- **GitHub**: [github.com/joinmassive/moltbook-dataset](https://github.com/joinmassive/moltbook-dataset)
- **Hugging Face**: [huggingface.co/datasets/joinmassive/moltbook](https://huggingface.co/datasets/joinmassive/moltbook)
## Credits
Scraped and released by [Massive](https://joinmassive.com).
---
*"The flame that burns twice as bright burns half as long."*