The Self-Evolving Agent Ecosystem — Trading agents that evolve through Darwinian selection and adversarial self-play

bili-core is an open-source framework for LLM benchmarking using LangChain, LangGraph, Streamlit, and Flask. It enables effective LLM model comparisons, Retrieval-Augmented Generation (RAG), and customizable decision workflows. Part of MSU Denver’s Sustainability Hub, bili-core promotes data democracy and transparent, reproducible AI research. 🚀

LLM Adversarial Security Arena — Jailbreak → Detect → Defend → Verify

Elenchus MCP Server - Adversarial verification system for code review

A marketplace of Claude Code plugins for adversarial security and architectural code review.

Official cli of humanbound platform.

AI safety evaluation framework testing LLM epistemic robustness under adversarial self-history manipulation

Agent-driven adversarial paper audit framework

Benchmark LLM jailbreak resilience across providers with standardized tests, adversarial mode, rich analytics, and a clean Web UI.

Context engineering toolkit for LLMs — pack, cache, debug, red-team, and orchestrate context windows. Council of Experts, adversarial testing, immune system, context compiler, drift detection, multi-agent entanglement. TypeScript + Python.

Mechanism-grounded taxonomy of 40 LLM jailbreak patterns across 10 categories. Full evaluation harness for 4 frontier models. AI safety research with responsible disclosure.

Systematic exploration of LLM alignment boundaries through logical stress testing

Adversarial eval harness for any LLM agent pipeline — Claude, OpenAI, or your own. CLI + REST API + MCP server for Cursor/Antigravity.

9-stage enterprise development pipeline for Claude Code. TDD, adversarial testing, mechanical verification. Any stack.

Adversarial testing of LLMs on constraint satisfaction deadlocks

Adversarial MCP server benchmark suite for testing tool-calling security, drift detection, and proxy defenses

Multi-agent adversarial API testing CLI. 3 agents, 3 oracle layers, every bug gets a curl command.

Red team toolkit for stress-testing MCP security scanners — find detection gaps before attackers do

Identified critical AI governance gaps: no adversarial testing, undocumented third-party models, and missing incident response. Delivered roadmap to secure high-risk KYC and transaction monitoring systems against evolving threats.

Description URF Application Stress Test — adversarial and scalability tests for Unified Rigidity Framework applications, validating limits under load, noise, and edge cases.