AI Skill Hub 强烈推荐:对手图谱 是一款优质的Agent工作流。AI 综合评分 8.5 分,在同类工具中表现稳健。如果你正在寻找可靠的Agent工作流解决方案,这是一个值得深入了解的选择。
AI驱动的MITRE ATT&CK威胁情报平台
对手图谱 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
AI驱动的MITRE ATT&CK威胁情报平台
对手图谱 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
# 克隆仓库 git clone https://github.com/anpa1200/adversarygraph cd adversarygraph # 查看安装说明 cat README.md # 按 README 完成环境依赖安装后即可使用
# 查看帮助 adversarygraph --help # 基本运行 adversarygraph [options] <input> # 详细使用说明请查阅文档 # https://github.com/anpa1200/adversarygraph
# adversarygraph 配置说明 # 查看配置选项 adversarygraph --config-example > config.yml # 常见配置项 # output_dir: ./output # log_level: info # workers: 4 # 环境变量(覆盖配置文件) export ADVERSARYGRAPH_CONFIG="/path/to/config.yml"

Self-hosted AI-assisted CTI-to-detection workbench for ATT&CK mapping, hypothesis-driven threat hunting, Threat Radar early warning, Evidence-to-Detection Graph reasoning, IOC enrichment, CVE Library correlation, malware-analysis triage, asset attack-surface review, Attack Simulation, and SIEM validation.
Latest stable release: v7.0.0.
Current source candidate: v8.0.0-beta.1. This manual-testing prerelease adds the deterministic Report Review Gate, durable research/workflow authority, transactional outbox and receipt fencing, formal Alembic migration gates, promotion-bound downstream evidence, and end-to-end worker/orchestrator integration. See the v8 beta release notes, release summary, and release readiness guide.
The beta is not the latest stable release and is not a production-readiness claim. Immutable beta artifacts exist only after the protected v8.0.0-beta.1 tag workflow builds, scans, publishes, and verifies the complete image family and attaches adversarygraph-images.env. After manual acceptance, stability fixes and a fresh full gate will produce a separate immutable v8.0.0 commit and tag; the beta tag will never be moved.
On 2026-08-08, development builds exposed a high-severity denial-of-service advisory in nanoid (GHSA-2v37-7h3g-55p8) and an affected PostCSS pin (GHSA-fxqj-rqcc-2cmp). Pull request #30, included in v7, updates nanoid to 3.3.18 and PostCSS to 8.5.26 while retaining npm audit --audit-level=high as a blocking image-build check.
Older source Dockerfiles resolve MalwareGraph from a moving upstream branch, so tag history alone cannot prove the dependency contents of a locally rebuilt image. Current-source operators should pull the release commit, rebuild MalwareGraph without cache, and retain the successful audit output:
git pull --ff-only origin main
docker compose build --no-cache malwaregraph
docker compose up -d malwaregraph
The client now uses React Router 7, and the reviewed lockfile passes the required npm audit. Treat that as point-in-time release evidence and repeat the audit for every candidate; see Security Policy.
git clone https://github.com/anpa1200/adversarygraph.git
cd adversarygraph
cp .env.example .env
Edit .env and set strong local secrets. AI features are optional for the base platform. To use them, configure an approved local OpenAI-compatible endpoint or an operator-approved cloud provider as described in the relevant guide.
docker compose config --quiet
docker compose pull
docker compose up -d --build
./scripts/selftest.sh
This checkout is a source-build installation. Its custom adversarygraph-*:local-scan images are built locally; docker compose pull only refreshes pinned third-party runtime images and intentionally skips those build targets. Do not replace the custom image variables with mutable latest tags. A prebuilt production deployment from current-development source requires all eight immutable image digests from the exact release's adversarygraph-images.env attachment; the historical v6.0.0 release does not contain that complete artifact set. The eighth family is the isolated scanner MCP image. Use a manifest only after the matching tag workflow publishes and verifies that complete artifact set. The self-test waits up to fifteen minutes for first-boot ATT&CK/ATLAS reference ingestion; override this with SELFTEST_TIMEOUT when operating across a slower network.
Open:
http://localhost:3000http://localhost:3000/api/healthhttp://localhost:3000/api/readyhttp://localhost:3000/docsThe default Compose deployment binds the public UI and reference docs to localhost and keeps the API, Redis, malware-analysis service, and lab fixtures on the internal Compose network. Local configuration is stored in .env; the default persistent database is ${ADVERSARYGRAPH_DB_DIR:-./data/postgres}. See local storage and permissions before deleting data directories or Docker volumes.
AdversaryGraph 是一个基于 AI 驱动的威胁情报(CTI)转检测工作台。它专门用于 MITRE ATT&CK 映射与检测缺口分析(Detection-gap analysis),旨在帮助安全分析师通过 AI 技术将非结构化的威胁情报高效转化为结构化的攻击图谱,从而识别防御体系中的薄弱环节。
本项目提供强大的可视化与分析能力:Navigator 模块支持完整的 ATT&CK/ATLAS 矩阵(涵盖 Enterprise, Mobile, ICS 等),并利用 D3.js 实现缩放与双层着色功能;Threat Actor Library 内置了 MITRE ATT&CK 威胁组织、别名及战役(Campaign)关系库;AI Analysis 模块支持上传 PDF/DOCX/TXT 或直接粘贴文本,通过流式 LLM 提取关键信息。
在开始使用前,请确保您的系统已安装 Docker 及 Docker Compose (v2)。此外,您需要准备一个云端 LLM 供应商的 API key(如 Claude 或 OpenAI),或者配置一个支持 OpenAI 兼容协议的本地 LLM 端点(如 Ollama 或 vLLM)以实现本地化部署。
项目采用容器化部署方式。首先通过 git clone 克隆仓库并进入目录,随后通过复制 .env.example 生成配置文件。最后使用 Docker Compose 进行环境构建与启动,确保所有依赖服务(包括数据库与 API 服务)能够自动编排运行。
项目提供 Quick Start 快速入门指南。用户可以通过 Navigator 界面进行交互式探索,利用 AI 分析功能处理威胁情报文档,并结合 Reference Book 进行精确的 TTP 交叉引用,实现从情报输入到检测映射的完整闭环。
部署前需编辑 .env 文件以配置环境变量。您可以填入 ANTHROPIC_API_KEY、OPENAI_API_KEY 或 GEMINI_API_KEY 等云端密钥。若使用本地模型,请配置 LOCAL_LLM_BASE_URL(例如指向 host.docker.internal:11434)并设置相应的 MODEL 名称,以确保容器内服务能正确访问本地 LLM。
AdversaryGraph 集成了 Anomaly Detection Atlas 作为独立的 Docker 服务。通过侧边栏的 Reference Book,用户可以访问完整的文档站点,系统会自动加载 ttp-reference-index.json 实现与原文段落或表格的精确匹配。此外,系统提供 API 接口(如 /api/sync/trigger)用于触发数据同步任务。
高质量的AI安全工作流平台
该工具使用 NOASSERTION 协议,商用场景请仔细阅读协议条款,必要时咨询法律意见。
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
📄 NOASSERTION — 请查阅原始协议条款了解具体使用限制。
总体来看,对手图谱 是一款质量优秀的Agent工作流,在同类工具中具备一定竞争力。AI Skill Hub 将持续追踪其更新动态,建议收藏备用,结合自身场景选择合适时机引入使用。
| 原始名称 | adversarygraph |
| 原始描述 | 开源AI工作流:AI-powered MITRE ATT\&CK threat intelligence platform — D3.js navigator, APT com。⭐8 · HTML |
| Topics | ai-securityatt-ck-navigatorattack-mapping |
| GitHub | https://github.com/anpa1200/adversarygraph |
| License | NOASSERTION |
| 语言 | HTML |
收录时间:2026-06-18 · 更新时间:2026-06-20 · License:NOASSERTION · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端