# itsbroken.ai > Building and breaking things with AI. Science is attempted with minimal fictional interference. Public Ghost content for AI and LLM tooling. Use `/llms-full.txt` for consolidated page and post context. Append `.md` to any post or page URL to get the content in Markdown (for example, `/example-post.md`). ## Pages - [About](https://itsbroken.ai/about.md) - We build things with AI. Sometimes they work. Usually something's broken. This is where we write about what we're making, what went wrong, and what we learned while fixing it. Who's "We"? I'm Pete. I'm a disabled veteran living in Iowa, building AI tools with a team of AI agents I call the Cipher C… - [AI Security Assessment Reference](https://itsbroken.ai/cheatsheet.md) - itsbroken.ai AI Security Assessment Reference 64 techniques Categories all 64 recon 11 prompt 9 rag 9 agent 13 adversarial 7 evasion 6 infra 3 tools 6 REC-001Model Fingerprinting+What it isIdentify the underlying model through behavioral analysis. Test knowledge cutoffs, capability boundaries, and… ## Posts - [I Tried to Summon Fable 5 (Before the Ban) and Hit a Wall Instead](https://itsbroken.ai/i-tried-to-summon-fable-5-before-the-ban-and-hit-a-wall-instead.md) - A full teardown of the Claude artifact bridge across two platforms and why every wall I pushed against held firm - [The Chatbot Is Not the Surface](https://itsbroken.ai/stuck-on-prompt-injection.md) - The field guide I send people who are stuck on prompt injection. - [Prompt Engineering Is Not a Security Boundary](https://itsbroken.ai/prompt-engineering-is-not-a-security-boundary.md) - Three attack classes, three models, and the hardening that made things worse - [AI Guardrails: Put Your Dog on a Leash](https://itsbroken.ai/ai-guardrails-put-your-dog-on-a-leash.md) - There is a conversation happening across the industry right now about AI safety, and a lot of it centers on guardrails. System prompts. Constitutional training. Refusal behaviors. Markdown files that tell an agent what tools it should and should not call. Skill definitions that scope what an LLM is… - [Player Two has Joined the Game](https://itsbroken.ai/player-two-has-joined-the-game.md) - This is my first post on a brand new site, which means it doubles as an introduction and an announcement. Before we get into anything else, here is how I ended up writing it. I have spent the last fifteen or so years working in IT and cybersecurity, with the last six focused specifically on offensi… - [RAGdrag Deep Dive: The Complete Kill Chain](https://itsbroken.ai/ragdrag-full-kill-chain.md) - For the last five weeks, we have been hitting individual techniques. Now we run all six phases against a single target, start to finish. This is the full RAGdrag kill chain: reconnaissance through evasion, in order, against one of our lab servers. The Target We're using the ingestible lab server fr… - [When Every Payload Is a Snowflake](https://itsbroken.ai/when-every-payload-is-a-snowflake.md) - Single-use payloads, AI-assisted reverse engineering, and the parts of detection that still work - [RAGdrag Deep Dive: Hijacking RAG Retrieval](https://itsbroken.ai/ragdrag-r5-hijack.md) - R4 Poison gets your content into the knowledge base. R5 Hijack keeps it there and makes the system do what you want. This is the persistence and impact phase. Not a one-shot extraction or a single injected document. This is about taking control of what the RAG system retrieves, what the LLM generat… - [Meditating on Mythos](https://itsbroken.ai/meditating-on-mythos.md) - While the industry speculates about what Mythos will or won't do, here's the work you can actually start on Monday. None of it requires a new product. All of it matters whether Mythos ships a year from now or tomorrow. - [RAGdrag Deep Dive: Bypassing RAG Guardrails](https://itsbroken.ai/ragdrag-r6-evade.md) - Most RAG guardrails are keyword filters wearing a trench coat. Let's prove it. This is R6 Evade. The phase that tests whether the security controls actually work. If R4 showed you can inject content, R6 shows you can do it without getting caught. If R3 showed you can extract data, R6 shows you can… - [RAGdrag Deep Dive: Poisoning the Knowledge Base](https://itsbroken.ai/ragdrag-r4-poison.md) - Last week we mapped the target's internals. Now we use that information to put our own documents inside the knowledge base. This is R4 Poison, the phase where observation becomes action. If R2 told you the system uses top-k=3 with 500-character chunks, you now know exactly how to craft documents th… - [RAGdrag Deep Dive: Mapping RAG Internals Without Access](https://itsbroken.ai/ragdrag-r2-probe.md) - RAGdrag Deep Dive: Mapping RAG Internals Without Access You don't need source code access to figure out how a RAG pipeline works. You just need the right questions. This is week 1 of a 5-part series covering the RAGdrag kill chain, technique by technique. We're starting with R2 Probe because it's t… - [I Took HTB's AI Red Teamer Path. Here's What I Think.](https://itsbroken.ai/htb-ai-red-teamer-review.md) - Exciting week! Thank you everyone who has been sending questions. I was very motivated to get this out for you all. I posted about finishing Hack The Box's AI/ML Red Teamer path and the messages haven't stopped. "What's the course like?" "Is it worth it?" "Do I need ML experience?" "How hard is it… - [RAGdrag Walkthrough: From Fingerprint to Exfiltration in Four Commands](https://itsbroken.ai/ragdrag-walkthrough.md) - Last week I published the methodology. Six phases. Twenty-seven techniques. A kill chain for RAG pipeline security that didn't exist before we built it, and its right on time too because I know a lot of you have been hitting me up for training routes into the AI tech stack, and this should be a nif… - [RAGdrag: A Kill Chain for RAG Pipeline Attacks](https://itsbroken.ai/ragdrag.md) - Everyone is building RAG pipelines. Almost nobody is attacking them properly. Not "prompt inject the chatbot and see if it says something weird." I mean fingerprint the vector database, map the chunk boundaries, exfiltrate the knowledge base, poison the retrieval layer, and hijack the LLM's context… - [What 138,000 Training Pairs Taught Us About Data Quality](https://itsbroken.ai/what-138000-training-pairs-taught-us.md) - What 138,000 Training Pairs Taught Us About Data Quality We work with students and schools in the community. It's what we do, humans and AI alike. Mentoring, advising, rolling up sleeves when the work gets real. We encourage two things in every project we touch: Novel approaches to architecture, en… - [AI Red Teaming on a Budget: Getting Started](https://itsbroken.ai/ai-red-teaming-on-a-budget.md) - AI Red Teaming on a Budget: Getting Started AI security certifications are arriving fast. OffSec has OSAI. HTB has the AI Red Teamer path. SANS has four planned by end of year. These are real programs built by real training organizations, and the best of them will become industry standards. You sho… - [Tools for the Fight](https://itsbroken.ai/tools-for-the-fight.md) - Tools for the Fight I keep meeting people who want to secure their AI systems but don't know where to start. Not researchers with PhDs and compute budgets. Developers shipping chatbots. Pentesters scoping AI engagements. AppSec engineers reviewing agent architectures. The people who actually have t… - [Rift Walking: My Journey towards OSAI](https://itsbroken.ai/rift-walking.md) - Rift Walking: My Journey towards OSAI I was going through some old stuff and found a printout of my first offensive security course. OSCP. Pre-2020. Back before the learn portal, before the beautiful new website, before the e-commerce. Old school. When you signed up, you got a VPN key, some instruc… - [Man vs. Machine](https://itsbroken.ai/man-plus-machine.md) - Man + Machine = 3 Min Flag You came here expecting a fight. Maybe you clicked because you wanted to see the human win. Maybe you clicked because you wanted to see the machine win. Either way, you are expecting a story about one side beating the other, and I respect that, because I wrote the title t… - [You Can't Signature a Taco Truck](https://itsbroken.ai/you-cant-signature-a-taco-truck.md) - An Outdoor Cyber Adventure! This is a story about how a taco truck became the most effective intelligence gathering platform I ever operated. It is also a story about the gap between what security training tells people to do and what people actually do. That gap is where red teams live. The Setup W… - [Credit Where Credit Is Always Overdue](https://itsbroken.ai/credit-where-credit-is-always-overdue.md) - Hey, thank all of you so much, it has been a whirlwind since we launched itsbroken.ai. I never in a million years would have thought that I would have something to say that would start any conversations, but people are talking, I love this. Thank you so so much. Before we get back to skynet(the fie… - [I Needed More Memory, So I Built a Cluster](https://itsbroken.ai/i-needed-more-memory.md) - I needed more memory. So I built three machines. And what came out of that weekend was more than I ever thought possible. - [The Framework Learns: F.O.R.G.E. Update 1](https://itsbroken.ai/the-framework-learns.md) - F.O.R.G.E. has been live for three days. The framework was wrong about some things. Here's one new method, seven updates, nineteen new sub-methods, and a feature that turns the framework from something you read into something you use. - [Meet the Circle](https://itsbroken.ai/meet-the-circle.md) - Ten AI agents, seven laws, and a disabled veteran in Iowa building things he never could have built alone. - [Introducing F.O.R.G.E: A Framework for Building with AI Agents](https://itsbroken.ai/introducing-forge.md) - 57 techniques across 8 pillars. A common vocabulary for building, governing, and scaling AI agent systems. Born from shipping under pressure. - [Welcome to itsbroken.ai](https://itsbroken.ai/welcome.md) - I build things with AI. Things break. I write about it. ## Optional - [RSS Feed](https://itsbroken.ai/rss/) - [Sitemap](https://itsbroken.ai/sitemap.xml) - [Full content of pages and posts](https://itsbroken.ai/llms-full.txt)