看开源

preambleai/prompt-injector

AI security and prompt injection payload toolkit

4vanish/AI-Pentest-Playbook

🛡 The reference playbook for pentesting AI chatbots & LLM-powered apps in one place. Ready-to-use payloads covering the full OWASP LLM Top 10 plus frontier vectors (MCP · RAG · A2A · computer-use · voice)

humanbound/humanbound-firewall

Multi-tier firewall for AI agents — blocks prompt injections, jailbreaks, and scope violations. Local tiers first; LLM judge only when uncertain.

togg53192-cmd/jailbreaks：32 条提示词覆盖 26+ 模型的越狱提示词合集

GitHub 仓库 togg53192-cmd/jailbreaks 收录 32 条针对 26+ 模型的完整提示词，涵盖 OpenCode Zen 免费隐身模型（LongCat 2.5。

ECC 2.2.3：为 Claude Code 等编码 Agent 提供 71 个 agents、311 个 skills 与 AgentShield 安全扫描的 MIT 开源工程系统

ECC 2.2.3 是 MIT 许可的开源编码 Agent 工程系统，通过 `npx ecc-universal@2.2.3 setup` 为 Claude Code、Codex、Kimi Code 提供引导式安装，原生插件 slug 为 `ecc@ecc`。

nightly-labs/openbot：本地优先的 AI 队友桌面工作区（开发预览）

nightly-labs 推出 OpenBot，一款本地优先的桌面工作区，为每个 AI 代理提供独立工作区与会话，支持本地 Codex App Server、Claude Code，并通过 ACP 接入 Grok CLI、OpenCode、Gemini、Cursor CLI 和 Cline CLI。

NuGuardAI/nuguard

AI red-teaming tool and LLM security framework to evaluate agentic AI applications. Tests prompt injections, handles vulnerability assessment, SBOM generation, and static analysis.

GenAI-Security-Project/GenAI-Red-Team-Lab

Provides a collection of sandboxes, exploitation code, and tutorials that exemplifies GenAI Red Teaming exercises. It aims to help security researchers and developers test, probe, and evaluate the safety and security of GenAI-based applications.

hardw00t/ai-security-arsenal

A collection of skills, agents, commands, and workflows for security researchers. Compatible with Claude Code, Claude Desktop, OpenCode, and other AI coding tools.

secureagentics/Adrian

Open-source runtime AI agent security tool - monitors and controls AI agents, catching malicious tool use, prompt injection, and policy drift in real time, before the agent acts.

corca-ai/awesome-llm-security

A curation of awesome tools, documents and projects about LLM Security.

aws-samples/sample-aiml-security-assessment

AI/ML and Generative AI Security Assessment Framework for AWS. Automatically audit Amazon Bedrock , SageMaker ,AgentCore, and Agent Registry workloads for security best practices

TauricResearch/TradingAgents

AWS 开源 Agent Toolkit，为 Claude Code 等编码 Agent 提供 MCP 与安全护栏

AWS 在 GitHub 开源 Agent Toolkit for AWS，为 Claude Code、Codex、Cursor、Devin、Kiro、fx 等编码 Agent 提供 MCP Server 配置。

luongnv89/claude-howto

wshobson/agents

Ponytail 5 发布：Claude Code 智能体技能减少 53% 代码量并保持 98% 风险逻辑测试覆盖

Ponytail 5 是面向 AI 编程智能体的技能提示词，在 Claude Code 中用同一智能体对比测试 39 个任务（含真实 FastAPI + React 仓库、Opus 5.5、各跑 5 次），代码量减少 53%、耗时减少 41%、成本降低 26%、token 减少 45%。