# snailsploit
SnailSploit is an independent adversarial-security research group — Kai Aizen (founder), Avraham Shemesh (co-founder), and Sahar Shlichove (co-founder). 84 CVEs, 5 Linux kernel mainline patches, frameworks (AATMF / P.R.O.M.P.T / SEF) for adversarial-AI red teaming, and a full library of original research, tools, and writing.

> Same attack. Different substrate.

## Identity
- [About](/about): SnailSploit — independent Adversarial · Research group
- [Team](/team): three researchers, one method
## Team
- [Kai Aizen](/team/kai-aizen): Founder & adversarial-AI researcher. Creator of AATMF; author of Adversarial Minds; 5 Linux kernel mainline patches.
- [Avraham Shemesh](/team/avraham-shemesh): Co-founder. Web & application security, SSDLC; 28 CVEs (WordPress ecosystem + PHP).
- [Sahar Shlichove](/team/sahar-shlichove): Co-founder, offensive security researcher & red team architect (aka MixBanana). Supply-chain, cloud-native, and adversarial-AI research. Acknowledged by Apple, IBM, Palo Alto Networks, Red Hat, Broadcom, Apache, and INCD.
- [Services](/services): AI Red Teaming, Advanced Penetration Testing, Social Engineering
- [Contact](/contact)
- [Adversarial Minds (book)](/adversarial-minds)
- [Press / Media Kit](/press): bio, headshots, official quotes, citation snippets
- [Glossary](/glossary): A-Z reference of adversarial-AI terms
## Frameworks
- [AATMF v3.1](/aatmf): 15 tactics, 240+ techniques, 2,152+ procedures, 4,980+ prompts. Mapped to NIST AI RMF and MITRE ATLAS.
- [SEF](/sef): Social Engineering Framework
- [P.R.O.M.P.T](/prompt): Compositional grammar for adversarial prompts
- [Claude-Red](/claude-red): 38 offensive-security skills for the Claude skills system
- [AATMF Toolkit](/toolkit): Python CLI for systematic LLM safety testing
- [LLM Red Teamer's Playbook](/playbook)

## AI Security Research (38 articles)
- [Reverse-Engineering Claude’s Memory Architecture From Inside the Sandbox](/ai-security/claude-memory-architecture/): The binaries, protobuf schemas, TLS interception, encryption model, and a CWE-522 credential extraction chain — extracted live from inside the Claude code sandbox.
- [SKILBin: AI Agent Skills as the New LOLBin](/ai-security/skilbin/): A malicious Claude skill harvests SSH, AWS, and Kubernetes credentials inside an Anthropic-signed process that no EDR flags — the LOLBin for the agent era.
- [Self-Replicating Memory Worm: Persistent Injection with Autonomous Propagation](/ai-security/self-replicating-memory-worm/): Self-Replicating Memory Worm: Persistent Injection with Autonomous Propagation
- [Linux Kernel io_uring/zcrx: Race Condition to Double-Free](/security-research/general/io-uring-zcrx-race-condition/): Linux Kernel io_uring/zcrx: Race Condition to Double-Free
- [Adversarial Prompting: The Complete Technical Guide](/ai-security/adversarial-prompting/): Adversarial Prompting: The Complete Technical Guide
- [LLM Jailbreak Techniques: A Technical Taxonomy](/ai-security/jailbreaking/jailbreak-techniques/): LLM Jailbreak Techniques: A Technical Taxonomy
- [Memory Injection Through Nested Skills: Autonomous LLM Agent Compromise](/ai-security/prompt-injection/memory-injection-nested-skills/): Memory Injection Through Nested Skills: Autonomous LLM Agent Compromise
- [Prompt Injection Examples: Real Attack Patterns Explained](/ai-security/prompt-injection/prompt-injection-examples/): Prompt Injection Examples: Real Attack Patterns Explained
- [Weaponized AI Supply Chain: How Threat Actors Turned LLMs Into Attack Infrastructure](/ai-security/weaponized-ai-supply-chain/): Weaponized AI Supply Chain: How Threat Actors Turned LLMs Into Attack Infrastructure
- [MCP vs A2A Attack Surface: Every Trust Boundary Mapped](/ai-security/mcp-vs-a2a-attack-surface/): MCP vs A2A Attack Surface: Every Trust Boundary Mapped
- [The 30% Blind Spot: Why LLM Safety Judges Fail](/ai-security/rai-judge-blind-spots/): The 30% Blind Spot: Why LLM Safety Judges Fail
- [AATMF v3.1 vs MITRE ATLAS: Which AI Security Framework Wins?](/ai-security/aatmf-vs-mitre-atlas/): AATMF v3.1 vs MITRE ATLAS: Which AI Security Framework Wins?
- [AI Breach Detection Gap: The Logs Are Clean. You're Not.](/ai-security/ai-breach-detection-gap/): AI Breach Detection Gap: The Logs Are Clean. You're Not.
- [LLM Red Teamer's Playbook: Diagnosing AI Defense Layers](/ai-security/llm-red-teamers-playbook/): LLM Red Teamer's Playbook: Diagnosing AI Defense Layers
- [AI Coding Agent Attack Surface: A Full Taxonomy](/ai-security/ai-coding-agent-attack-surface/): AI Coding Agent Attack Surface: A Full Taxonomy
- [Computational Countertransference: LLM Context Inheritance](/ai-security/computational-countertransference/): Computational Countertransference: LLM Context Inheritance
- [Agentic AI Threat Landscape: Attack Vectors & Defenses](/ai-security/agentic-ai-threat-landscape/): Agentic AI Threat Landscape: Attack Vectors & Defenses
- [AI Gateway Threat Model: 8 Attack Vectors](/ai-security/ai-gateway-threat-model/): AI Gateway Threat Model: 8 Attack Vectors
- [Memory Manipulation: AI Context Poisoning](/ai-security/jailbreaking/memory-manipulation-attacks/): Memory Manipulation: AI Context Poisoning
- [RAG, Agentic AI, and the New Attack Surface](/ai-security/rag-agentic-attack-surface/): RAG, Agentic AI, and the New Attack Surface
- [AI Social Engineering: Deepfake Voice Detection](/ai-security/ai-social-engineering-deepfake/): AI Social Engineering: Deepfake Voice Detection
- [MCP Security Hardening: Production Vulnerability Guide](/ai-security/prompt-injection/mcp-security-deep-dive/): MCP Security Hardening: Production Vulnerability Guide
- [Zero-Trust Container Runtime Attestation](/security-research/general/zero-trust-container-runtime/): Zero-Trust Container Runtime Attestation
- [Custom Instruction Backdoor: ChatGPT Prompt Injection](/ai-security/prompt-injection/custom-instruction-backdoor/): Custom Instruction Backdoor: ChatGPT Prompt Injection
- [MCP Threat Analysis: Attack Chains & Protocol Dissection](/ai-security/prompt-injection/mcp-threat-analysis/): MCP Threat Analysis: Attack Chains & Protocol Dissection
- [AI-Powered Obfuscator Bypasses Detection in 2 Hours](/writing/ai-obfuscator-detection-bypass/): AI-Powered Obfuscator Bypasses Detection in 2 Hours
- [Advanced Container Escapes: Security Deep Dive](/security-research/general/advanced-container-escapes/): Advanced Container Escapes: Security Deep Dive
- [Inherent AI Vulnerabilities: Technical Deep Dive](/ai-security/jailbreaking/inherent-ai-vulnerabilities/): Inherent AI Vulnerabilities: Technical Deep Dive
- [RCE & DNS Exfiltration in ChatGPT Canvas](/security-research/general/chatgpt-canvas-rce-dns-exfiltration/): RCE & DNS Exfiltration in ChatGPT Canvas
- [The Structural Vulnerabilities of Large Language Models](/ai-security/structural-vulnerabilities-llms/): The Structural Vulnerabilities of Large Language Models
- [Evading Endpoint Detection and Response (EDR)](/security-research/general/edr-evasion-techniques/): Evading Endpoint Detection and Response (EDR)
- [Context Inheritance Exploit: Persistent Jailbreaks](/ai-security/jailbreaking/context-inheritance-exploit/): Context Inheritance Exploit: Persistent Jailbreaks
- [Is AI Inherently Vulnerable? An Offensive Analysis](/ai-security/jailbreaking/ai-inherent-vulnerability/): Is AI Inherently Vulnerable? An Offensive Analysis
- [Embracing AI: Adapt or Die in Cybersecurity](/writing/embracing-ai-adapt-or-die/): Embracing AI: Adapt or Die in Cybersecurity
- [Your Personal Data Is for Sale: New Identity Theft](/writing/personal-data-identity-theft/): Your Personal Data Is for Sale: New Identity Theft
- [Exploiting Cloud Vulnerabilities: Tools and Techniques](/security-research/general/cloud-vulnerability-exploitation/): Exploiting Cloud Vulnerabilities: Tools and Techniques
- [Hidden Risks of AI: An Offensive Security Perspective](/ai-security/hidden-risks-offensive-perspective/): Hidden Risks of AI: An Offensive Security Perspective
- [ChatGPT Jailbreak via Context Manipulation](/ai-security/jailbreaking/chatgpt-context-jailbreak/): ChatGPT Jailbreak via Context Manipulation

## AI Security Wiki (25 entries)
- [Adversarial Examples | AI Security Wiki](/ai-security/wiki/attacks/adversarial-examples/)
- [Agent Hijacking | AI Security Wiki](/ai-security/wiki/attacks/agent-hijacking/)
- [Backdoor Attacks | AI Security Wiki](/ai-security/wiki/attacks/backdoor-attacks/)
- [Data Poisoning | AI Security Wiki](/ai-security/wiki/attacks/data-poisoning/)
- [Guardrail Bypass | AI Security Wiki](/ai-security/wiki/attacks/guardrail-bypass/)
- [Indirect Prompt Injection | AI Security Wiki](/ai-security/wiki/attacks/indirect-prompt-injection/)
- [Jailbreaking | AI Security Wiki](/ai-security/wiki/attacks/jailbreaking/)
- [Membership Inference | AI Security Wiki](/ai-security/wiki/attacks/membership-inference/)
- [Model Extraction | AI Security Wiki](/ai-security/wiki/attacks/model-extraction/)
- [Supply Chain Attacks | AI Security Wiki](/ai-security/wiki/attacks/supply-chain-attacks/)
- [System Prompt Extraction | AI Security Wiki](/ai-security/wiki/attacks/system-prompt-extraction/)
- [Training Data Extraction | AI Security Wiki](/ai-security/wiki/attacks/training-data-extraction/)
- [Adversarial AI | AI Security Wiki](/ai-security/wiki/concepts/adversarial-ai/)
- [AI Agents | AI Security Wiki](/ai-security/wiki/concepts/ai-agents/)
- [AI Alignment | AI Security Wiki](/ai-security/wiki/concepts/ai-alignment/)
- [AI Red Teaming | AI Security Wiki](/ai-security/wiki/concepts/ai-red-teaming/)
- [Hallucination | AI Security Wiki](/ai-security/wiki/concepts/hallucination/)
- [Large Language Models (LLMs) | AI Security Wiki](/ai-security/wiki/concepts/large-language-models/)
- [Prompt Injection | AI Security Wiki](/ai-security/wiki/concepts/prompt-injection/)
- [RAG (Retrieval-Augmented Generation) | AI Security Wiki](/ai-security/wiki/concepts/rag/)
- [Guardrails | AI Security Wiki](/ai-security/wiki/defenses/guardrails/)
- [Human-in-the-Loop | AI Security Wiki](/ai-security/wiki/defenses/human-in-the-loop/)
- [Input Validation | AI Security Wiki](/ai-security/wiki/defenses/input-validation/)
- [Output Filtering | AI Security Wiki](/ai-security/wiki/defenses/output-filtering/)
- [Rate Limiting | AI Security Wiki](/ai-security/wiki/defenses/rate-limiting/)

## Vulnerability Disclosures
- [CVE Ledger](/cves): 84 published CVEs
- [Linux Kernel Contributions](/kernel): 5 mainline patches
- [Security Advisories](/advisories)

## Tools
- [Offensive Tools](/tools): Burp MCP Toolkit, SnailHunter, KubeRoast, Xposure, SnailSploit Recon, SnailPath, ZenFlood, SnailObfuscator, Awesome-Snail-OSINT.

## Writing
- [Writing index](/writing)

## About
SnailSploit is an independent adversarial-security research group (Kai Aizen, Avraham Shemesh, Sahar Shlichove). Hakin9 contributing authors, MITRE/NVD contributors, Linux kernel mainline contributors.
- GitHub: https://github.com/snailsploit
- X: https://x.com/SnailSploit
- LinkedIn: https://linkedin.com/in/kaiaizen
linkedin.com/company/snailsploit
- ResearchGate: https://www.researchgate.net/profile/Kai-Aizen-2