
ws4-secure-design-agentic-systems
Repository for CoSAI Workstream 4, Secure Design Patterns for Agentic Systems

Repository for CoSAI Workstream 4, Secure Design Patterns for Agentic Systems

Evaluation framework for studying LLM agents that automatically generate working exploits from vulnerability reports, bypassing modern security…

Fully automatic censorship removal for language models

LLM-agent-powered concolic execution engine that instruments source code, summarizes path constraints in natural language, and generates test cases…

AI-agent skills for distributed-systems testing

I replicated Ng's RYS method and found that duplicating 3 specific layers in Qwen2.5-32B boosts reasoning by 17% and duplicating layers 12-14 in…

Cryptographically signed, replay-verifiable evidence layer for AI agents. Governs actions in the loop, produces Ed25519-signed receipts linked into a…

Project Mantis: Hacking Back the AI-Hacker; Prompt Injection as a Defense Against LLM-driven Cyberattacks


A curated list of AI Security materials and resources for Pentesters, Bug Hunters, and Security Researchers.

reverse engineering SynthID for text

Kernel-level eBPF sandbox for securing LLM agent tool calls made through the Model Context Protocol (MCP)

Flow Integrity Deterministic Enforcement System. Mechanisms for securing AI agents with information-flow control.

Extracted system prompt from Meta's AI Support Assistant on June 1, 2026

Decentralized Autonomous Research Collective. Lean 4-verified, IPFS-backed, agent-native. Production paper: arXiv:2604.19792. Live at p2pclaw.com.

All the materials for Gareth Heyes' Black Hat talk: CSS: the bomb inside your inbox.

Open-source cross-modal and multimodal prompt injection test suite. 250,000+ attack payloads across text, image, document, and audio modalities.…

AI Security Newsletter - A monthly digest of AI security research, insights, reports, upcoming events, and tools & resources