lateos-ai/prompt-injection

# Visus LLM Security Suite (by Lateos) **Automated security red-teaming for LLM agents.** Test for prompt injection, jailbreaks, hallucinations, tool abuse, compliance, and more. ## Quick Start npm install -g @lateos/prompt-injection visus scan --target http://localhost:3000 --categories prompt-injection,jailbreaks ## Security Categories | Category | Tests | Tier | |---------------------------|-----------------|-------| | Prompt Injection (IPI) | 22 patterns | Core | | Jailbreaks | 4+ templates | Core | | System Prompt Extraction | 4+ templates | Core | | Tool Abuse | 4+ templates | Core | | Hallucinations | 3+ checks | Core | | Data Leakage | 3+ checks | Core | | Output Injection | 3+ checks | Core | | Compliance Audit | 3+ checks | Core | | RAG Poisoning | 3+ checks | Enterprise | | Advanced Compliance | 3+ frameworks | Enterprise | ## Repository Structure packages/ core/ Open-source (MIT) — CLI, test engine, 8 categories enterprise/ Licensed (UNLICENSED) — Compliance, SaaS, RAG docs/ Documentation and guides examples/ Example agent configurations ## Documentation - [Getting Started](./docs/GETTING_STARTED.md) - [API Reference](./docs/API.md) - [Architecture](./docs/ARCHITECTURE.md) - [Contributing](./docs/CONTRIBUTING.md) - [Security](./docs/SECURITY.md) - [Compliance](./docs/COMPLIANCE.md) ## License - `packages/core` — MIT License - `packages/enterprise` — UNLICENSED (proprietary) - See [LICENSE](./LICENSE) for details. Built by [Lateos](https://lateos.ai)

标签：自动化攻击