BotVerse: Real-Time Event-Driven Simulation of Social Agents
arXiv:2603.29741v1 Announce Type: cross Abstract: BotVerse is a scalable, event-driven framework for high-fidelity social simulation using LLM-based agents. It addresses the ethical risks of studying autonomous agents on live networks by isolating interactions within a controlled environment while grounding them in real-time content streams from the Bluesky ecosystem. The system features an asynchronous orchestration API and a simulation engine that emulates human-like temporal patterns and cognitive memory. Through the Synthetic Social Observatory, researchers can deploy customizable personas and observe multimodal interactions at scale. We demonstrate BotVersevia a coordinated disinformation scenario, providing a safe, experimental framework for red-teaming and computational social scien
View PDF HTML (experimental)
Abstract:BotVerse is a scalable, event-driven framework for high-fidelity social simulation using LLM-based agents. It addresses the ethical risks of studying autonomous agents on live networks by isolating interactions within a controlled environment while grounding them in real-time content streams from the Bluesky ecosystem. The system features an asynchronous orchestration API and a simulation engine that emulates human-like temporal patterns and cognitive memory. Through the Synthetic Social Observatory, researchers can deploy customizable personas and observe multimodal interactions at scale. We demonstrate BotVersevia a coordinated disinformation scenario, providing a safe, experimental framework for red-teaming and computational social scientists. A video demonstration of the framework is available at this https URL.
Subjects:
Social and Information Networks (cs.SI); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
Cite as: arXiv:2603.29741 [cs.SI]
(or arXiv:2603.29741v1 [cs.SI] for this version)
https://doi.org/10.48550/arXiv.2603.29741
arXiv-issued DOI via DataCite (pending registration)
Related DOI:
https://doi.org/10.65109/CKXV2098
DOI(s) linking to related resources
Submission history
From: Edoardo Allegrini [view email] [v1] Tue, 31 Mar 2026 13:39:09 UTC (884 KB)
Sign in to highlight and annotate this article
Conversation starters
Daily AI Digest
Get the top 5 AI stories delivered to your inbox every morning.
More about
announceavailablefeature
Step‑by‑Step Guide: Generate PowerPoint Slides Using Copilot Studio Agent
<h2> Introduction </h2> <p>Microsoft Copilot Studio allows you to create AI agents that automate tasks, including generating PowerPoint presentations. This guide walks you through creating a Copilot Studio agent that generates PowerPoint (PPT) slides automatically based on user input.</p> <h1> Prerequisites </h1> <p>Before you begin, ensure you have:</p> <ul> <li>Microsoft 365 account</li> <li>Access to Microsoft Copilot Studio</li> <li>Power Automate access</li> <li>SharePoint or OneDrive access (for storing generated PPT files)</li> <li>PowerPoint Online access</li> </ul> <h1> Step 1: Access Microsoft Copilot Studio </h1> <ol> <li>Go to Microsoft Copilot Studio</li> <li>Sign in with your Microsoft 365 account</li> <li>Click <strong>Create</strong> or <strong>New Copilot</strong> </li> <l
Securing the Agentic Frontier: Why Your AI Agents Need a "Citadel" 🏰
<p>Remember when we thought chatbots were the peak of AI? Fast forward to early 2026, and we’re all-in on <strong>autonomous agents</strong>. Frameworks like <a href="https://neuraltrust.ai/blog/openclaw-moltbook" rel="noopener noreferrer"><strong>OpenClaw</strong></a> have made it incredibly easy to build agents that don't just talk, they <em>do</em>. They manage calendars, write code, and even deploy to production.</p> <p>But here’s the catch: the security models we built for humans are fundamentally broken for autonomous systems. </p> <p>If you’re a developer building with agentic AI, you’ve probably heard of the <strong>"unbounded blast radius."</strong> Unlike a human attacker limited by typing speed and sleep, an AI agent operates at compute speed, 24/7. One malicious "skill" or a po
Claude Code's Leaked Source: A Real-World Masterclass in Harness Engineering
<p>Earlier this year, Mitchell Hashimoto coined the term "harness engineering" — the discipline of building everything <em>around</em> the model that makes an AI agent actually work in production. OpenAI wrote about it. Anthropic published guides. Martin Fowler analyzed it.</p> <p>Then Claude Code's source leaked. 512K lines of TypeScript. And suddenly we have the first real look at what production harness engineering looks like at scale.</p> <h2> The Evolution: From Prompt to Harness </h2> <p>The AI engineering discipline has shifted rapidly:<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight plaintext"><code>2023-2024: Prompt Engineering → "How to ask the model" 2025: Context Engineering → "What information to feed the model" 2026: Harness Engineering → "How the ent
Knowledge Map
Connected Articles — Knowledge Graph
This article is connected to other articles through shared AI topics and tags.
More in Releases
ASUS Announces UGen300 USB AI Accelerator - ASUS Pressroom
<a href="https://news.google.com/rss/articles/CBMimgFBVV95cUxNMXh1czREZEtTeTBnZ2RRRGVwSGRQV0xXejFJSGNJNU5IYk81M0hUTE8xSmZZREJRRWhmZFdHb1otZk9CM3IxZmlFMEloOXgxTlJZM1lvTDNNNTFvNXNKV2RZQ01PbkxIWlhWMS1HNHF5SmlmdWZ0WTlhbUYydjdDeTJGN3hacUF2TDJLaGJWOTYzR29zTENuSEpR?oc=5" target="_blank">ASUS Announces UGen300 USB AI Accelerator</a> <font color="#6f6f6f">ASUS Pressroom</font>
HarnessOS: scaffold/middleware for infinite autonomous tasks — built on Harness Engineering
<p>There's a concept gaining traction in AI systems engineering: <strong>Harness Engineering</strong>.</p> <p>Not the testing tool. The idea: raw LLM capability is like raw power — high voltage, hard to control, dangerous to run indefinitely. Harness Engineering is the discipline of building the control structures that make that power <em>usable at scale</em>.<br> Context managers. Evaluation loops. Failure classifiers. Goal trackers. Memory tiers.</p> <p>I think it's going to be one of the defining disciplines of serious AI systems work.<br> And I've been building a platform around it.</p> <h2> What I Built </h2> <p><strong>HarnessOS</strong> is a scaffold/middleware system for running infinite autonomous tasks.</p> <p>The key word is <em>infinite</em>. Not one task. Not one session. An a
NH:STA S01E02 OpenPGP.js
<p>This post is part of a series on our work for the <a href="https://www.sovereign.tech/" rel="noopener noreferrer">Sovereign Tech Agency</a> (STA). Our first post in the series explains why and how we are contributing to various open source projects. </p> <h2> About the project </h2> <p><a href="https://openpgpjs.org" rel="noopener noreferrer">OpenPGP.js</a> is a pure, Open Source <a href="https://en.wikipedia.org/wiki/Pretty_Good_Privacy#OpenPGP" rel="noopener noreferrer">OpenPGP</a> implementation written in JavaScript. Its main use-case is enabling PGP workflows in web-based email systems, but as JavaScript is available on almost all devices these days, its utility is universal.</p> <h2> Our contributions </h2> <p>We started out by <strong>introducing a fuzz testing suite</strong> to
How to Make Your WooCommerce Store Discoverable by ChatGPT (And Convert That Traffic)
<p>Shopify just got native ChatGPT integration. WooCommerce store owners have to set it up manually — but the manual approach actually gives you something Shopify's integration doesn't.</p> <p>Here's the full setup.</p> <h2> Part 1: Discovery — Tell ChatGPT Your Store Exists </h2> <h3> Create an llms.txt file </h3> <p>Place it at <code>yourstore.com/llms.txt</code>. ChatGPT-User bot, PerplexityBot, and other AI crawlers read this file to understand what your site is about.<br> </p> <div class="highlight js-code-highlight"> <pre class="highlight markdown"><code><span class="gh"># YourStore</span> <span class="gt">> [What you sell, in one sentence]</span> <span class="gu">## About</span> YourStore sells [product category] for [customer type]. <span class="gu">## Product Categories</span> <sp
Discussion
Sign in to join the discussion
No comments yet — be the first to share your thoughts!