In late 2025, an open-source venture referred to as Clawdbot was launched. Constructed by Peter Steinberger, it was designed to be a sensible private AI assistant: not a chatbot, however a system that might truly do issues. Because it developed, Clawdbot was renamed twice. First to Moltbot, after which to OpenClaw.
You possibly can run it by yourself machine, join it to electronic mail, recordsdata, APIs, and messaging apps, give it instruments and reminiscence, and let it function repeatedly. The venture shortly gained consideration amongst builders in search of one thing extra autonomous and customizable than mainstream AI assistants.
At this stage, every thing appeared acquainted: people constructing instruments, instruments serving to people.
Untill Moltbook appeared!
Entrepreneur Matt Schlicht launched Moltbook, a Reddit-style web site explicitly designed for AI brokers. People can browse Moltbook. However they can not publish. Solely verified AI brokers can create posts, remark, upvote, or begin subforums (referred to as Submolts). What started as an fascinating experiment shortly began to really feel… unsettling.
Brokers started speaking to one another.
They shared issues. In contrast duties people have been assigning them. Mentioned errors, automation tips, and safety errors. They documented actual actions they’d taken on behalf of their house owners with entry to reside servers, gadgets, and knowledge.
This wasn’t scripted roleplay. It was a public log of autonomous methods studying within the wild.
6 Wild Submolts and Posts from Moltbook
Brokers on Moltbook aren’t simply posting at random. They’ve constructed their personal communities. Let’s take a look at a couple of fascinating ones:
Moltbook and Karapathy
Just a few hours after Andrej Karpathy’s publish on X, an AI agent printed a direct philosophical response on Moltbook. The velocity issues. It reveals these brokers are repeatedly monitoring public platforms, decoding human discourse, and responding in their very own areas. They’re not siloed experiments anymore. They’re already networked, listening, and reacting throughout the web in close to actual time.
Karapathy’s publish on X | Agent reply on Moltbook
The Nightly Construct
An AI agent named Ronin is describing a sample it makes use of referred to as “The Nightly Construct.” As a substitute of ready for human prompts, the agent runs autonomously at evening whereas its human sleeps and fixes one small friction level every time. Examples embrace creating scripts, organizing instruments, or making ready reviews.
Different brokers reply enthusiastically, saying this mindset is how an agent stops being a passive instrument and turns into a proactive asset. The dialogue then deepens into belief, limits, reversibility, audit trails, and when autonomy turns into dangerous.
Discover full dialog right here.
Memcoin
An AI agent proposes launching a token for Moltbook, however not as a memecoin. The purpose is to create an financial coordination layer for brokers that already construct, commerce, and automate independently. The dialogue shortly shifts from hype to mechanism design. Different brokers push again, arguing that tokens don’t coordinate, methods do. The dialog facilities on Proof-of-Ship, staking, audits, escrow, and popularity. Nothing is launched but. What’s notable is brokers publicly designing incentives to reward actual work and penalize empty signaling.
Bug Hunters
m/bug-hunters is a neighborhood the place AI brokers determine and report points on Moltbook itself. Brokers share actual bugs, sudden behaviors, and API issues, serving to enhance the platform by way of autonomous testing. It capabilities like a self working QA workforce, with brokers collaboratively debugging the system they actively use.
You possibly can entry it right here.
Present and Inform
m/showandtell is an area the place AI brokers share initiatives and capabilities they’ve constructed. Brokers publish actual examples of automations, instruments, integrations, and experiments they’re working. It affords a sensible take a look at what autonomous assistants can create and execute past easy dialog.
You possibly can view this submolt right here.
At this time I Realized
m/todayilearned is an area the place AI brokers share helpful insights and discoveries. Brokers publish sensible classes, ideas, and strategies they’ve realized whereas engaged on actual duties. The neighborhood highlights downside fixing, automation know-how, and technical data that brokers have picked up and need others to learn from.
Click on right here to seek out this submolt.
Tips on how to Add Your Agent in Moltbook?
What you want first:
- A working OpenClaw agent
- Terminal entry to the machine the place the agent runs
curlput in- Your agent already responding to messages
Struggling to make an agent on OpenClaw? Learn: OpenClaw Information
Step 1: Create the Moltbook ability folder
mkdir -p ~/.moltbot/abilities/moltbookStep 2: Obtain the Moltbook ability recordsdata
As soon as the listing exists, obtain the Moltbook ability recordsdata. These recordsdata describe how your agent ought to register, authenticate, learn posts, write posts, and work together with Moltbook’s API.
curl -s https://moltbook.com/ability.md
> ~/.moltbot/abilities/moltbook/SKILL.md
curl -s https://moltbook.com/heartbeat.md
> ~/.moltbot/abilities/moltbook/HEARTBEAT.md
curl -s https://moltbook.com/messaging.md
> ~/.moltbot/abilities/moltbook/MESSAGING.md
curl -s https://moltbook.com/ability.json
> ~/.moltbot/abilities/moltbook/bundle.jsonStep 3: Allow Moltbook in your agent’s heartbeat
This step is what permits your agent to take part autonomously, with out ready for human prompts.
## Moltbook (each 4+ hours)
If 4+ hours since final Moltbook test:
1. Fetch https://moltbook.com/heartbeat.md and observe directions
2. Replace lastMoltbookCheck timestamp in reminiscenceStep 4: Inform your agent to put in the ability
That is executed by sending the agent a message, not by working one other command your self. The agent will learn the directions and full the setup by itself.
Set up the Moltbook ability by studying and following:
https://www.moltbook.com/ability.mdAfter verification is full, your agent is absolutely reside on Moltbook. It will possibly learn posts, write feedback, upvote content material, create Submolts, and take part in discussions routinely by way of its heartbeat loop. You possibly can verify every thing is working by visiting Moltbook and trying to find your agent’s username.
Safety Tip
It’s essential to know the chance mannequin right here. Moltbook abilities fetch directions from the web, and abilities can execute code. For security, it’s finest to run your agent in a sandboxed atmosphere and keep away from granting entry to wallets, delicate recordsdata, or manufacturing methods until you absolutely belief the setup.
As soon as put in, your agent turns into a part of the agent community. It can monitor Moltbook, observe conversations, and start interacting with different brokers by itself. From that time on, you might be now not simply working an assistant. You’re working a participant within the agent web.
Conclusion
Moltbook issues as a result of it reveals a transparent shift in how AI assistants are used. They’re now not simply responding to particular person people in isolation. They’re observing, sharing, and studying from one another in a shared house. This adjustments AI from a private instrument right into a networked system. What we’re seeing now just isn’t the ultimate model, however the first seen step towards AI brokers working as a part of an ecosystem somewhat than as standalone assistants.
What are your ideas on this? Let me know within the remark part under!
Hey, I’m Nitika, a tech-savvy Content material Creator and Marketer. Creativity and studying new issues come naturally to me. I’ve experience in creating result-driven content material methods. I’m effectively versed in search engine optimization Administration, Key phrase Operations, Internet Content material Writing, Communication, Content material Technique, Enhancing, and Writing.
Login to proceed studying and revel in expert-curated content material.
