AgentInfra Radar
Free directory

AI agent security tools and adjacent guardrail projects

A public directory of AI agent security platforms, LLM security tools, guardrail libraries, and red-team projects that matter for agentic systems.

View sample leadsRequest full radar
Public project examples
Lasso Security
AI agent security platform
Public-source checked
Website
Why it matters

Positions around securing AI agents, AI applications, discovery, posture management, red teaming, and runtime protection.

Pillar Security
AI agent security platform
Public-source checked
Website
Why it matters

Focuses on visibility, risk assessment, runtime protection, and governance for AI agents, making it a direct category signal.

Lakera Guard
AI-native runtime protection
Public-source checked
Website
Why it matters

Provides GenAI protection patterns and agent-focused security signals such as prompt attack defense and adversarial testing.

Zenity
AI agent security and governance
Public-source checked
Website
Why it matters

Focuses on securing AI agents across buildtime and runtime, with posture management and governance across SaaS and custom agents.

HiddenLayer
AI security platform
Needs verification for agent-specific fit
Website
Why it matters

Covers AI discovery, model security, attack simulation, and runtime defense; agent-specific fit should be checked by use case.

Protect AI
AI security platform
Needs verification for agent-specific fit
Website
Why it matters

Secures AI applications from model selection and testing to runtime, useful as an adjacent AI lifecycle security reference.

NVIDIA garak
LLM vulnerability scanner
Needs verification for agent-specific fit
WebsiteGitHub
Why it matters

Open-source scanner for LLM weaknesses such as prompt injection, leakage, hallucination, toxicity, and jailbreak behavior.

NVIDIA NeMo Guardrails
LLM guardrails library
Needs verification for agent-specific fit
WebsiteGitHub
Why it matters

Programmable guardrails can sit between an application and LLM, relevant to safer agent workflows even when not agent-only.

Microsoft PyRIT
Generative AI red-team framework
Needs verification for agent-specific fit
WebsiteGitHub
Why it matters

Open-source red-team framework for identifying risks in generative AI systems, useful for security teams testing agentic apps.

Promptfoo
LLM red teaming and evals
Needs verification for agent-specific fit
WebsiteGitHub
Why it matters

Combines prompt/model evals with red teaming and vulnerability scanning, useful for testing LLM and agent application behavior.

Giskard
AI agent evaluation and red teaming
Public-source checked
WebsiteGitHub
Why it matters

Provides open-source evaluation and testing for LLM agents plus red-team positioning, making it a useful safety research signal.

Research boundary

Public-source research, not a private contact database

This directory is based on public websites, documentation, GitHub repositories, and public product pages. It is meant for research and market mapping, not automated outreach.

Manual verification

Manually verify before outreach

Categories and URLs can change. Before using any project for sales, partnership, investment, or security research, review the current public source and confirm the contact angle is accurate.

Next step

Want the lead version of this directory?

The free directory is only a starting point. AgentInfra Radar can turn a category into a reviewed lead pack with possibleNeed, leadScore, contactAngle, source, and verificationStatus.

View sample leadsRequest full radar