Throne

OpenAPPA is an open, vendor-agnostic, MIT-licensed deterministic AI guardrail engine designed to prevent data exfiltration from LLM agents by tracking data flows and enforcing algebraic security labels without breaking agent utility.

Xalgorix is an autonomous AI pentesting platform that runs exploit-verified security tests against web apps and repos, reproduces findings with working exploits, and provides remediation guidance, CI gating, and auditor-ready reports.

ModelFuzz provides runtime guardrails for LLM agents: a red-team scanner that exposes prompt-injection vulnerabilities and a lightweight Python decorator that intercepts and blocks unsafe tool calls at execution time.

Keydris is an authorization infrastructure that enforces per-action authority for AI agents by evaluating agent-presented authority against versioned policies before actions run, returning ALLOW, APPROVAL REQUIRED, or REJECT decisions and recording decision evidence.

Redthread powers MILLENNIUMS.AI, an agentless-first security platform that maps applications, AI agents, code, cloud assets, identities, and data into a shared risk graph. Its modules assess security posture, autonomously test AI applications, and provide proven findings with remediation guidance.

GPT Guard is an enterprise data-loss-prevention (DLP) platform that enables secure, privacy-preserving chat with LLMs by masking sensitive PII/PHI while preserving context, offered as SaaS or on‑premises for regulated industries.

Anomify is an observability and monitoring platform that uses AI and multi-stage machine learning to deliver real-time insights, event detection, and early warning for critical infrastructure and operations teams.

Backmesh is an open-source backend that protects LLM API keys and acts as an API Gatekeeper, providing JWT-based authentication, per-user rate limits, resource access controls, and instrumentation for LLM usage analytics to help apps safely call LLM APIs and reduce costs.

Autogon (Nemesis Labs) provides a suite of security products—Nemesis Shield (runtime & SecOps), Omniguard (fraud, AML & sanctions screening), Nemesis Blue (endpoint/EDR) and Nemesis Red (autonomous pentest)—that learn normal behavior, block deviations, and produce provable evidence for every action. It's aimed at financial institutions, fintechs, enterprises and infra teams.