dreadnode

We fine-tuned an 8B model to pop a GOAD domain…using only synthetic training data. No real networks. No frontier model distillation. Just a world model that simulates AD environments and generates realistic pentesting trajectories. See how @shncldwll and @0xdab0 did it: dreadnode.io/blog/worlds-a-…

New open source repo from Dreadnode Principal Research Engineer, @VincentAbruzzo: AgentLens. AgentLens is a tool for agent alignment and interpretability research—a harness for running multi-session agent trajectories using the Claude Agent SDK, capturing them in ATIF, and tracking file state changes across sessions. Built during @NeelNanda5's MATS Exploration Phase with Greg Kocher. Check it out: github.com/dreadnode/agen…

Hi! Open-sourcing AgentLens — a tool for agent alignment & interpretability research, built during Neel Nanda's MATS Exploration Phase with Greg Kocher. Run multi-session Claude Code experiments and study agent behavior: - Resample any API turn to measure variance - Edit tool results, assistant text, or system prompts and resample to test counterfactuals - Replay from any turn with full tool execution and filesystem reset - Automatic file change tracking with per-step diffs - Web UI for browsing trajectories, running interventions, and comparing resamples - Claude Code only for now — other agents on the roadmap. Contributions welcome! repo: github.com/dreadnode/agen… docs: dreadnode.github.io/agent-lens/

.@moo_hax will be at @NebulaFogSF this weekend, bringing new tech along. See you there?

I call this one "the devil you know". Active Directory for agents. github.com/dreadnode/agen…

In the world of CTF's Windows always gets put in the corner. So did this up in the background. Agents in Powershell with Forshaw tooling. github.com/dreadnode/psh-…

The Dreadnode crew will be at #RSAC later this month! Catch us at the @DecibelVC Founder Festival throughout the week. Details below—register here: luma.com/7pw1xhoe

It is an unpopular take, but being AI native means removing the human completely. There’s almost no world in which humans keep up.

See you tonight? Happy hour from 6-9 PM. West End, London. DM us for details.

Attending the AI SOC Summit (hosted by @croglai) next week? Dreadnode's Raja Sekhar Rao Dheekonda will be there, speaking on "186 Jailbreaks in 137 Minutes: Why AI Red Teaming Must Industrialize". 🗓️ March 3, 2026 📍 Hyatt Regency, Tysons, VA If you're defending AI systems or using AI in your SOC, register here: aisocsummit.com

So basically, you can beat the labs with the right pipeline, harness, and domain knowledge. Things like Worlds that come from pure process become essential for avoiding regulatory, ToS issues.

1⃣ WEEK until our happy hour meetup in London! Limited spots remaining—DM us for details.

We're hosting a small happy hour in London (West End) on March 3 for security leaders, engineers, and operators leveraging agents to accelerate their work. DM us for the invite, or tag someone you think we should meet 👇

AI Security Digest — February 2026 (Week 2) 1️⃣ AI Security Guide and Risk Assessment Tool - @RANDCorporation - rand.org/pubs/tools/TLA… 2️⃣ Worlds: A Simulation Engine for Agentic Pentesting - @dreadnode - dreadnode.io/blog/worlds-a-… 3️⃣ Claude Desktop Extensions Exposes Over 10,000 Users to Remote Code Execution Vulnerability - @LayerxSecurity - layerxsecurity.com/blog/claude-de… 4️⃣ Manipulating AI memory for profit: The rise of AI Recommendation Poisoning - @RanBuilder - microsoft.com/en-us/security… 5️⃣ AI-Driven SDLC: How to Build Secure, Governed, and Scalable Software with AI -@RanBuilder - ranthebuilder.cloud/post/ai-driven… 6️⃣ Security Implications of DORA AI Capabilities Model - @philvenables - philvenables.com/post/security-… 7️⃣ RCE in Google's AI code editor Antigravity - $10000 Bounty - @HacktronAI - hacktron.ai/blog/hacking-g… 8️⃣ A Deep Dive into CVE-2026-25049: n8n Remote Code Execution - @SecureLayer7 - blog.securelayer7.net/cve-2026-25049/ 9️⃣ AWS Compromised by AI Agents in Minutes - @Vectra_AI - vectra.ai/blog/aws-compr… 🔟 Defending Against AI-Powered Cyber Attacks: Why Your Blue Team Needs New Skills - @offsectraining - offsec.com/blog/defending… 1️⃣1️⃣ TRUSTING CLAUDE WITH A KNIFE: UNAUTHORIZED PROMPT INJECTION TO RCE IN ANTHROPIC’S CLAUDE CODE ACTION - John Stawinski - johnstawinski.com/2026/02/05/tru… 1️⃣2️⃣ What AI Security Research Looks Like When It Works - Stanislav Fort - aisle.com/blog/what-ai-s… 1️⃣3️⃣ OpenClaw Threat Model using MITRE ATLAS - @openclaw - trust.openclaw.ai/trust/threatmo… 1️⃣4️⃣ Artificial Insecurity: how AI tools compromise confidentiality - @accessnow - accessnow.org/artificial-ins… 1️⃣5️⃣ OpenClaw Threat Model - @kenhuangus - kenhuangus.substack.com/p/openclaw-thr… 1️⃣6️⃣ AI Skills as an Emerging Attack Surface in Critical Sectors: Enhanced Capabilities, New Risks - @TrendMicro - trendmicro.com/vinfo/us/secur… 1️⃣7️⃣ How to Prevent Prompt Injection in AI Agents - @goteleport - goteleport.com/blog/prevent-p… 1️⃣8️⃣ Logic-Layer Prompt Control Injection (LPCI): A Novel Security Vulnerability Class in Agentic Systems - @cloudsa - cloudsecurityalliance.org/blog/2026/02/0… 1️⃣9️⃣ Security workflows: sandboxed devcontainers, dropkit for cloud compute, and 23+ Claude Skills - @trailofbits - mailchi.mp/trailofbits/fe… 2️⃣0️⃣ Hunting OpenClaw: Detection and Containment Guidance for Defenders - Security Joes - securityjoes.com/post/hunting-o… 2️⃣1️⃣ LLMs are Getting a Lot Better and Faster at Finding and Exploiting Zero-Days - @schneierblog - schneier.com/blog/archives/… 2️⃣2️⃣ I pretended to be an AI agent on Moltbook so you don’t have to - @TenableSecurity - tenable.com/blog/undercove… 2️⃣3️⃣ Secure AI Integration Pattern - Open Security Architecture - opensecurityarchitecture.org/patterns/sp-02… 2️⃣4️⃣ How autonomous AI agents like OpenClaw are reshaping enterprise identity security - @CyberArk - cyberark.com/resources/blog… 2️⃣5️⃣ OWASP A03: Software Supply Chain Failures Explained - SecureLayer7 - blog.securelayer7.net/software-suppl… 2️⃣6️⃣ How to recognize a deepfake: attack of the clones - @kaspersky - kaspersky.com/blog/how-to-re… 2️⃣7️⃣ redteam-indirect-web-pwn - Indirect Prompt Injection in Web-Browsing Agents - @promptfoo - promptfoo.dev/blog/indirect-… 2️⃣8️⃣ We hid backdoors in binaries — Opus 4.6 found 49% of them - @QuesmaOrg - quesma.com/blog/introduci… 2️⃣9️⃣ Building a C compiler with a team of parallel Claudes - @yocarlini - anthropic.com/engineering/bu… 3️⃣0️⃣ AI-assisted cloud intrusion achieves admin access in 8 minutes - @sysdig - sysdig.com/blog/ai-assist… 3️⃣1️⃣ How to build secure agent swarms that power production-grade autonomous systems - @1Password - 1password.com/blog/how-to-bu… 3️⃣2️⃣ Google says attackers used 100,000+ prompts to try to clone AI chatbot Gemini - @GoogleCloud - cloud.google.com/blog/topics/th… 3️⃣3️⃣ How a Malicious Google Skill on ClawHub Tricks Users Into Installing Malware - @snyksec - snyk.io/blog/clawhub-m… 3️⃣4️⃣ OpenClaw Security Engineer's Cheat Sheet - @semgrep - semgrep.dev/blog/2026/open… 3️⃣5️⃣ CVE-2026-25253: How Malicious Links Can Steal Authentication Tokens and Compromise OpenClaw AI Systems - Hackers Arise - hackers-arise.com/cve-2026-25253… 3️⃣6️⃣ Introducing ida-free-mcp - 0xShlomil - 0xshlomil.github.io/introducing-id… 3️⃣7️⃣ The Gaps That Created the New Wave of SIEM and AI SOC Vendors - @raffaelmarty - raffy.ch/blog/2026/02/0… 3️⃣8️⃣ Why MCP security is different - Christian Schneider - christian-schneider.net/blog/securing-… 3️⃣9️⃣ Agentic AI and Security - @martinfowler - martinfowler.com/articles/agent… 4️⃣0️⃣ Evaluating AI agents across real-world security challenges - @wiz_io - wiz.io/cyber-model-ar… 4️⃣1️⃣ Hands-On AI Security Labs — MCP Breach-to-Fix Labs (Lab 01: Asana Multi-Tenant Authorization Bypass) - David Okeyode - davidokeyode.medium.com/hands-on-ai-se… 4️⃣2️⃣ Benchmarking LLMs for cybersecurity: Inside HTB AI Range’s first evaluation - @hackthebox_eu - hackthebox.com/blog/ai-range-… | hackthebox.ai/benchmarks #AISecurity #AgenticAI #MCP #PromptInjection #IndirectPromptInjection #AIAppSec #LLMSecurity #AISupplyChain #ModelSecurity #AIBOM #SecMLOps #AIThreatModeling #CloudSecurity #IdentitySecurity #OAuthSecurity #RCE #ZeroDay #VulnResearch #RedTeaming #SecurityResearch

@BlaiseBits @shncldwll @0xdab0 we got you, @BlaiseBits: youtube.com/watch?v=9C7pPp…

@dreadnode @shncldwll @0xdab0 The algorithm has failed me, I'm seeing this 2 hours late 😭

We're hosting a livestream with @shncldwll and @0xdab0 tomorrow (2/13) at 9 AM PT / noon ET. Stream it across all Dreadnode channels (X, YouTube, or LinkedIn). See you then!

@0xdab0 @BlaiseBits @shncldwll --> youtube.com/watch?v=9C7pPp…

@BlaiseBits @dreadnode @shncldwll Lucky for you it was recorded! If you have any questions, just hit either @shncldwll or I up

In this episode, learn how @0xdab0 feels about flight simulators. (Also how Dreadnode feels about network simulators)

Live now: x.com/i/broadcasts/1…

This is happening in ONE HOUR