llm-security-compliance-prompt-injection 开源项目推荐（25 个）

requie /LLMSecurityGuide

大型语言模型（LLMs）安全保护的综合参考资料。涵盖OWASP生成式AI十大风险、即时注入、对抗性攻击、现实世界事件及实用防御。包含红队工具、防护措施和缓解策略目录，帮助开发者、研究人员和安全团队负责任地部署人工智能。

ErdemOzgen /RedAiRange

A professional AI security range for red teaming, vulnerability research, defensive validation, and hands-on AI/ML security training.

regaan /basilisk

Basilisk — Open-source AI red teaming framework with genetic prompt evolution. Automated LLM security testing for GPT-4, Claude, Grok, Gemini. OWASP LLM Top 10 coverage. 32 attack modules.

techlab-innov /llmtrace

Zero-code LLM security & observability proxy. Real-time prompt injection detection, PII scanning, and cost control for OpenAI-compatible APIs. Built in Rust.

HASHIRU-AI /NAAMSE

Neural Adversarial Agent Mutation-based Security Evaluator

0x-Professor /VeilArmor

Veil Armor is an enterprise-grade security framework for Large Language Models (LLMs) that provides multi-layered protection against prompt injections, jailbreaks, PII leakage, and sophisticated attack vectors.

giggsoinc /PatronAI

Apache 2.0 AI endpoint monitoring for shadow AI, ghost AI assets, and unmanaged LLM usage.

Lukentony /AI-guardian-lab

Security middleware for AI Agents. Intercepts shell commands before execution using a multi-layer pipeline: binary allowlist, regex patterns, deterministic intent coherence mapping, and LLM semantic check as last resort.

kevinl05 /promptmenot

Detect and sanitize prompt injection attacks in Rails apps. Protects against direct injection (users hacking your LLMs via form inputs) and indirect injection (malicious prompts stored for other LLMs to scrape). ~70 detection patterns across 7 attack categories with configurable sensitivity levels. Now includes resource extraction detection pattern

d12o6aa /arabguard

A high-performance Python SDK for protecting LLMs against Prompt Injection attacks in Arabic, Egyptian Dialect, and Franco. Combining Rule-based and AI-driven security layers.

Httpslakshya /AgentPrahari

A fail-closed runtime security layer for AI agents with prompt injection detection, tool-call protection, PII/secret sanitization, HITL controls, and output validation.

abood879 /zero-data-intent-gateway

Sub-millisecond Deterministic AI Governance Middleware & Zero-Data Intent Architecture.

yedinrumba-eng /Secure-Vibe

Advanced configuration of toolings, skills, and markdowns for automated code review and security auditing in vibe-coded applications.

blackshadowteamnet-netizen /AI-Security-Research

A technical research repository focusing on AI jailbreak mitigation, prompt injection vulnerabilities, and system-level security guardrails for LLMs.⁠

Kartavyasonar /InvokeCordon

Security scanner and runtime policy gateway for MCP servers

Mohamed-Gomaa30 /Arabic_llm_red_teaming

Robustness and jailbreak evaluation framework for Arabic LLMs (e.g., Jais, AceGPT, ALLaM, Gemini). Evaluates cross-format safety alignment across Standard Arabic, Arabizi, and English transliteration with automated LLM judging and scalable Modal GPU execution.

ScottColeSW /Agent-Defense-Theater

A governed range for testing AI agent defenses: Agent Theater’s earned-trust tiers and human oversight, plus an adversarial arena that attacks each OWASP Top 10 for Agentic Applications risk. Measured results, honest gaps, and what it would take to close them.

Miraj-Rahman-AI /RAI-RAG

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.

TharaneshA /LMNinja

Automated LLM Red Teaming

jerzy99jerzy /cve-digest

Reference implementation of a guardrailed, EPSS/KEV-driven CVE triage pipeline that threat-models its own LLM against prompt injection.