attack-chain
Teaches your AI agent to organize and coordinate AI agents and plan projects before coding.
What it does
This tool helps plan and execute multi-stage computer attacks, like gathering information, gaining initial access, escalating privileges, and moving through a network. It's an entry point for full security assessments.
When to use it
- ✓When an authorized security assessment spans multiple phases.
- ✓To coordinate different tools into one attack plan.
- ✓To plan a complete attack, from information gathering to the final report.
Example of what to ask your AI
“Plan a full penetration test for example.com, starting with reconnaissance and ending with a report.”
Want more? Open the sections below.
How to use
The easiest way: copy the message below and paste it into Claude, ChatGPT, Cursor or any AI. It reads the skill and follows the instructions.
Skill link
Ready-to-paste message
How it works
The tool first plans the sequence of actions needed for an attack, figuring out which tools to use at each stage. It then executes these actions in order, ensuring each step is authorized.
Installation
Easiest way: copy the skill link and paste it into Claude, ChatGPT or Cursor asking it to follow the instructions.
In the terminal (Claude Code, Codex, Cursor), in your project folder:
npx skills add sickn33/agentic-awesome-skills --skill attack-chainOr download just the file:
curl -L https://raw.githubusercontent.com/sickn33/agentic-awesome-skills/main/plugins/agentic-awesome-skills-claude/skills/attack-chain/SKILL.md -o SKILL.mdTechnical content (original SKILL.md)
Original text by the author, straight from GitHub.
⚠️ AUTHORIZED USE ONLY This skill is for educational purposes or authorized security assessments only. You must have explicit, written permission from the system owner before using this tool. Misuse of this tool is illegal and strictly prohibited.
Mandatory confirmation gate Before running any command that probes, exploits, changes, persists on, extracts data from, or attempts credential access against a target:
- Ask the user to state the exact target URL, IP, account, or resource.
- Ask the user to confirm written authorization and the permitted scope.
- Show the exact command(s) and explain their expected effect.
- Wait for explicit confirmation in the current conversation.
Without that confirmation, remain read-only and provide defensive guidance only. Prefer a sandbox, disposable VM, or controlled lab.
Attack Chain Orchestration Skill
When to Use
- An authorized engagement spans multiple kill-chain phases.
- Coordinating several specialist skills into one coherent attack path.
何时路由到本 Skill
以下场景必须先经过本 Skill 做全链路规划,再分发到具体子 Skill 执行:
| 场景 | 为什么需要编排 |
|---|---|
| "帮我做一次完整的渗透测试" | 需要规划从信息收集到报告的全流程 |
| "从外网打到域控" | 跨越边界突破→提权→横向→AD 多个阶段 |
| "HW 攻防演练" | 需要完整攻击链 + 隐蔽性 + 痕迹清理 |
| "评估这个目标的攻击面" | 需要多维度信息收集 + 路径规划 |
| "我拿到了一个 webshell,下一步怎么办" | 需要从当前据点规划后续路径 |
| "帮我规划攻击路径" | 明确需要路径编排 |
| "从这个漏洞能打到什么程度" | 需要评估漏洞的链式利用价值 |
| "Bug Bounty 持续监控" | 需要自动化多阶段流程 |
| "内网渗透全流程" | 横向移动 + 提权 + 域攻击组合 |
| "近源渗透方案" | 物理接入 + 内网渗透组合 |
| "供应链攻击路径" | 跨组织多跳攻击 |
| "钓鱼 + 后渗透" | 初始访问 + 后续利用组合 |
单阶段任务不需要经过本 Skill:
- 只做端口扫描 → 直接去
pentest-tools/ - 只做 SQL 注入 → 直接去
pentest-tools/ - 只做 APK 逆向 → 直接去
apk-reverse/ - 只做域渗透 → 直接去
windows-ad/SKILL.md
编排原则
本 Skill 的角色
用户提出多阶段任务
↓
attack-chain/SKILL.md(本文件)
↓ 规划攻击路径、确定阶段顺序
↓ 评估每阶段所需工具和方法
↓
分发到具体子 Skill 执行:
├── pentest-tools/ → 工具调用、漏洞利用
├── apk-reverse/ → 移动端渗透
├── js-reverse/ → Web 前端突破
├── reverse-engineering/ → 二进制分析
├── ida-reverse/ → 深度逆向
└── browser-automation/ → 自动化操作
↓
每阶段完成后回到本 Skill 评估下一步
↓
全部完成 → docs-generator 生成报告
路径规划决策树
拿到目标后:
1. 目标是什么?(Web/内网/云/移动/IoT)
2. 当前有什么?(外部视角/已有凭据/已有据点)
3. 最终目标是什么?(域控/数据/特定系统/证明影响)
4. 约束条件?(时间/隐蔽性/不可触碰的系统)
↓
根据以上信息规划最短路径
↓
一条路走不通 → 回到本 Skill 重新规划备选路径
完整攻击链阶段
For detailed phase methodology, toolchains, and playbooks, refer to:
红队行动铁律
三条底线
- 所有操作必须获得书面授权
- 数据渗出需进行匿名化处理
- 清理所有攻击痕迹(包括内存驻留)
行动纪律
- 每个操作前评估风险等级(低/中/高/严重)
- 高风险操作前通知项目经理
- 保持操作日志(时间、动作、结果)
- 发现高危漏洞立即上报,不扩大利用
- 不影响业务可用性(禁止 DoS)
- 不访问/下载真实用户数据
典型失败案例
| 失败原因 | 后果 | 教训 |
|---|---|---|
| 未清除 Mimikatz 内存 dump | 蓝队溯源完整攻击路径 | 操作后立即清理 |
| C2 域名被威胁情报标记 | 首次连接即被拦截 | 使用新注册域名 + 域前置 |
| 钓鱼邮件触发 DLP 告警 | 蓝队提前预警 | 测试邮件网关规则 |
| 横向移动触发蜜罐 | 暴露攻击意图 | 先识别蜜罐再行动 |
工具速查表
信息收集
subfinder amass httpx naabu katana gau dnsx nmap whatweb wpscan
漏洞利用
nuclei sqlmap sstimap xsstrike burpsuite metasploit
权限提升
winPEAS linpeas GodPotato PrintSpoofer watson
横向移动
mimikatz crackmapexec/netexec impacket bloodhound certipy coercer responder evil-winrm
C2 框架
cobalt-strike sliver havoc mythic adaptixc2
近源渗透
fluxion aircrack-ng proxmark3 rubber-ducky wifi-pineapple
与本包其他 Skill 的关系
| 需求 | 路由到 |
|---|---|
| Web 漏洞深度利用 | pentest-tools/SKILL.md |
| 内网 AD 攻击详细步骤 | windows-ad/SKILL.md |
| 逆向分析恶意样本 | reverse-engineering/SKILL.md |
| APK 逆向(移动端渗透) | apk-reverse/SKILL.md |
| JS 前端签名绕过 | js-reverse/SKILL.md |
| 自动化群体渗透 | Pentest Swarm AI(pentestswarm scan --swarm) |
| AI 辅助渗透 | mcp-kali-server / metasploitmcp / hexstrike-ai |
| 报告生成 | docs-generator/SKILL.md |
| 攻击路径图 | diagram-generator/SKILL.md |
任务完成自检(声称完成前 MUST 通过)
- 我是否执行了工作流中的每一步(而不是只阅读)?
- 我是否基于
tool-index使用了真实工具路径? - 我是否产出了可复现证据(命令/脚本/截图/报告)?
- 我是否完成并回写了 RULES 要求的 Checklist 项?
Limitations
- Execution of attack stages requires written authorization per target.
- Assumes supporting tooling (recon/exploit kits) is installed and licensed where applicable.
Adapted from zhaoxuya520/reverse-skill (MIT).
FAQ
Do I need permission to use this tool?
Yes, you must have explicit, written permission from the system owner before using this tool for any purpose.
What happens if I don't have permission?
Misusing this tool is illegal and strictly prohibited. The tool will only proceed if you confirm you have the necessary authorization.
Does this tool do everything by itself?
No, it plans and orchestrates the attack but may need to use other specific tools for each stage, like information gathering or vulnerability exploitation.
Details
Original description by the author
Authorized multi-stage attack-path planning and orchestration spanning reconnaissance, initial access, privilege escalation, lateral movement, and reporting. Entry point for full engagements and cross-phase operations.
Popularity
- GitHub stars
- 48k
- Ranking position
- #2218
- License
- MIT
- Updated
- 10/15/2026
Comments
- …
More skills from this repository
00-andruia-consultant
Helps your agent start or improve AI projects. It analyzes code, suggests tools, and creates action plans in Spanish.
sickn33/agentic-awesome-skills
007
Lets your agent perform security audits and threat modeling. It's useful for protecting projects against attacks and vulnerabilities.
sickn33/agentic-awesome-skills
10-andruia-skill-smith
Helps your agent create or improve AI tools. It defines how tools work, writes documentation, and validates the structure.
sickn33/agentic-awesome-skills
20-andruia-niche-intelligence
Lets your agent create a summary about a specific industry. It's useful for understanding the context, rules, usage patterns, and risks of a project, using reliable sources.
sickn33/agentic-awesome-skills
Similar skills
agent-architecture-audit
Makes your agent check if everything is working correctly. It finds and fixes problems in AI systems, ensuring they perform as expected.
affaan-m/ECC
agent-eval
Helps your agent compare different AI coding tools. It measures which one is faster, cheaper, and more reliable for specific tasks.
affaan-m/ECC
agent-harness-construction
Makes your agent better understand how to perform tasks. It optimizes the tools and instructions the agent uses to increase its chances of success.
affaan-m/ECC