A secure low code honeypot framework, leveraging AI for System Virtualization.

AI Security Platform: Defense (61 Rust engines + Micro-Model Swarm) + Offense (39K+ payloads)

Security working agreements for AI coding agents: hardened AGENTS.md, prompt/tool-injection guardrails, dependency hygiene, Scorecard-ready OSS setup

The dashcam and emergency brake for AI agents. A security proxy that physically blocks rogue LLM commands and generates cryptographically proven audit trails for enterprise compliance.

Agentic AI Security Bootcamp is a hands-on, research-driven training environment for analysing, attacking, and securing autonomous AI systems. The repository provides structured labs, adversarial evaluation frameworks, and red-teaming exercises covering multi-agent observability, prompt injection..

🤖 Test and secure AI systems with advanced techniques for Large Language Models, including jailbreaks and automated vulnerability scanners.

💰 Exocomp™ by the Ferengi Commerce Authority 💰

Formal safety framework for AI agents. Pluggable LLM reasoning constrained by mathematically proven budget, invariant, and termination guarantee. 7 theorems enforced by construction, not by prompting. Includes Bayesian belief tracking, causal dependency graphs, sandboxed attestors, environment reconciliation, and a 155-test adversarial suite.

TypeScript/JavaScript SDK for AI Agent Security - Drop-in security for LangChain, CrewAI, AutoGPT and custom agents

An experiment in backdooring a shell safety classifier by planting a hidden trigger in its training data.

Risk-Aware Introspective RAG (RAI-RAG) is a safety-aligned RAG framework integrating introspective reasoning, risk-aware retrieval gating, and secure evidence filtering to build trustworthy, robust, and secure LLM and agentic AI systems.

A formal specification and reference implementation of the Governed State Machine (GSM) for AI agent governance.

Enforce agent actions with policy checks and cryptographic receipts to prove compliance and enable independent verification.

Signed receipts for agent/tool actions. PolicyGate enforces allow/deny; every decision emits a tamper-evident receipt with hashes, signatures, and optional approvals. Verify in CI, prove what happened, and make agent integrations survivable in regulated environments.