net-pentest — скилл для Claude Code: что делает и как установить
Скилл net-pentest для Claude Code систематизирует процесс тестирования на проникновение сетевой инфраструктуры. Он помогает проводить сканирование, перечисление хостов и создание отчётов в формате PDF.
MIT ★ 106 обновлён 2026-10-22
npx skills add hassan-algo/open-sesame --skill net-pentest -a claude-codeЧто делает net-pentest
Инструмент ориентирован на выполнение тестов безопасности по стандартам PTES и NIST SP 800-115. Он включает пошаговые инструкции для внешней разведки, анализа Active Directory, повышения привилегий и бокового перемещения. Скилл опирается на справочные файлы для каждой фазы тестирования, охватывая такие темы, как работа с nmap, netexec, impacket и методология отчётности.
Для создания итогового документа предусмотрен скрипт scripts/render-report.sh. Перед началом работы обязательно подтверждение прав на тестирование целевой инфраструктуры. Работа с облачными провайдерами или сторонними сервисами без специального разрешения запрещена.
Какие задачи решает
- составить отчёт о безопасности в формате PDF
- выполнить перечисление доменов в Active Directory
- провести безопасную проверку сетевого периметра
Примеры запросов агенту
- просканируй подсеть на предмет уязвимостей
- помоги с перечислением хостов в Active Directory
- составь технический отчёт на основе собранных данных
Как установить net-pentest
1Через Skills CLI (Vercel)
npx skills add hassan-algo/open-sesame --skill net-pentest -a claude-code2Вручную: положить файлы
~/.claude/skills/net-pentest/SKILL.mdПапка скилла целиком; для одного проекта — .claude/skills/net-pentest/
1Через Skills CLI (Vercel)
npx skills add hassan-algo/open-sesame --skill net-pentest -a codex2Вручную: положить файлы
.agents/skills/net-pentest/SKILL.mdДля всех проектов — ~/.agents/skills/
1Через Skills CLI (Vercel)
npx skills add hassan-algo/open-sesame --skill net-pentest -a cursor2Вручную: положить файлы
.cursor/skills/net-pentest/SKILL.mdCursor читает и .agents/skills/, и .claude/skills/
1GitHub CLI
gh skill install hassan-algo/open-sesame .claude/skills/net-pentest2Вручную: положить файлы
.github/skills/net-pentest/SKILL.md1Через Skills CLI (Vercel)
npx skills add hassan-algo/open-sesame --skill net-pentest -a gemini-cli2Командой платформы
gemini skills install https://github.com/hassan-algo/open-sesame --path .claude/skills/net-pentestВручную: положить файлы
.devin/skills/net-pentest/SKILL.mdСтарый путь .windsurf/skills/ тоже читается
Claude.ai
Упакуйте папку в zip и загрузите в Settings → Capabilities → Skills
По шагам, с командами для каждой платформы: Как установить скилл в Claude Code: пять способов
Вопросы о net-pentest
Как установить net-pentest в Claude Code?
Выполните npx skills add hassan-algo/open-sesame --skill net-pentest -a claude-code. После завершения проверьте файл ~/.claude/skills/net-pentest/SKILL.md.
Что делает net-pentest?
Скилл направляет действия специалиста по пентесту от этапа определения целей до создания финального отчёта, следуя методологиям PTES и NIST.
Что нужно для работы net-pentest?
Требуется официальное разрешение на тестирование конкретной сети, инфраструктуры или использование лицензированного лабораторного стенда.
Исходный текст
SKILL.md · MIT
--- name: net-pentest description: >- Guides authorized network and infrastructure penetration testing — external perimeter and internal/Active Directory — from scoping through reporting. Use this skill whenever the user is testing a network, hosts, or infrastructure they're authorized to assess: their own network, a client engagement with a signed scope, or a lab (HackTheBox, TryHackMe, GOAD, a local range). Trigger it for requests like "pentest my network", "scan this subnet", "enumerate these hosts", "what's exposed on our perimeter", "help with Active Directory / AD attacks", "Kerberoasting / BloodHound / password spray", "privilege escalation on this box", "which nmap/netexec/impacket command do I run", or "map lateral movement" — even when the user doesn't say "pentest". Follows the PTES / NIST SP 800-115 flow, organizes tooling into a repeatable workflow, and produces clean technical + stakeholder reports (PDF). Sibling to the web-pentest skill; use that one for a single web application. Authorized testing only — third-party and cloud-provider infrastructure is out of scope by default. --- # Network & Infrastructure Penetration Testing This skill helps you run a disciplined network penetration test — external or internal, including Active Directory — and turn what you find into a clear report. It follows the Penetration Testing Execution Standard (PTES) and NIST SP 800-115 so coverage stays systematic. For testing a **single web application**, use the sibling **web-pentest** skill instead. Use this one for networks, hosts, services, and AD. ## Authorization comes first — every time Network testing is only legal and ethical against infrastructure you're allowed to test. Before any **active** work (scanning, enumeration, exploitation), confirm the engagement is one of: - The user's **own** network / infrastructure. - A **client engagement** with a signed scope / rules of engagement (RoE). - A **lab** the user controls or is licensed to use (HackTheBox, TryHackMe, GOAD, Vulnhub, a local range). If authorization or scope is unclear, ask once before proceeding: *"Is this a network you're authorized to test, and what's the exact scope?"* Scanning or attacking infrastructure you don't own, without authorization, is a crime — don't help with it, and say why. Network engagements need a **tighter RoE than web**, because scans and exploits can disrupt live services. Record before starting: - **In-scope IP ranges / CIDRs and hostnames**, and every host explicitly **out** of scope (domain controllers, SCADA/OT, medical or safety systems, shared infrastructure). - **Test windows** and rate/intensity limits — fragile services can crash under aggressive scans or exploits. - Whether **exploitation**, **password attacks**, **DoS-prone checks**, and **AD changes** are permitted, or detection/enumeration only. - A **point of contact** for "stop" and emergencies, and a rollback plan. - **Third-party / cloud-provider infrastructure is out of scope by default** — test your own hosts and config, not the cloud platform, ISP, or hosting provider underneath (and note each provider's own testing policy; many require prior notice). ## Engagement types - **External** — testing from the internet against the perimeter: what's exposed, what's exploitable from outside. - **Internal** — testing from inside the network (assumed-breach or on-site): lateral movement, privilege escalation, and **Active Directory**, where most real-world compromise happens. Say which you're doing; it changes the emphasis (perimeter hardening vs. AD and lateral movement). ## The workflow 1. **Scope & RoE** — confirm authorization, record scope (above). 2. **Host discovery** — find live hosts in the in-scope ranges. 3. **Port & service scanning** — open ports, service/version detection. 4. **Enumeration** — dig into each service (SMB, LDAP, SNMP, SSH, web, DB, mail…) and, internally, enumerate Active Directory. 5. **Vulnerability identification** — map versions/configs to known issues; confirm, don't just trust a scanner. 6. **Exploitation & privilege escalation** — gain and elevate access, safely and in scope. 7. **Post-exploitation, lateral movement & pivoting** — demonstrate real impact (and, in AD, the path to Domain Admin). 8. **Reporting** — write each finding up with evidence, severity, and a fix. Keep running notes of every host, port, service, credential, and foothold — the report and the attack path depend on them. ## Where to go next Read the reference for the phase you're in: - **`references/methodology.md`** — the phase-by-phase network testing checklist (discovery, scanning, per-service enumeration, vuln ID, exploitation, privesc, pivoting). The core of the skill. - **`references/active-directory.md`** — internal AD attacks: enumeration, Kerberoasting / AS-REP roasting, password spraying, SMB/NTLM relay, BloodHound attack paths, credential abuse (pass-the-hash, DCSync), and the path to Domain Admin. Open it on internal engagements. - **`references/privesc.md`** — privilege escalation after a foothold, Linux and Windows (sudo/SUID/cron/caps; token privileges, service misconfig, stored creds). - **`references/pivoting.md`** — tunneling and pivoting into deeper in-scope segments (SSH, chisel, ligolo-ng, proxychains, SOCKS/port-forwards). - **`references/password-attacks.md`** — safe credential testing: spraying vs brute force, lockout policy, offline hash cracking, default/reused creds. - **`references/cloud-identity.md`** — hybrid/cloud identity: Entra ID (Azure AD) and AWS — what's in scope (the client's tenant/config) vs the provider's platform (out of scope). Open it on hybrid/cloud engagements. - **`references/wireless.md`** — Wi-Fi testing (WPA2/3-PSK handshake/PMKID, Enterprise/802.1X evil-twin, WPS, guest segmentation). Needs a monitor-mode adapter. - **`references/containers.md`** — Docker & Kubernetes: exposed APIs (Docker socket, kube API/kubelet/etcd) and container escapes. - **`references/tooling.md`** — how to run nmap, masscan, netexec (CrackMapExec), impacket, enum4linux-ng, responder, BloodHound, Metasploit, etc., with safe-usage notes. - **`references/reporting.md`** — severity (CVSS), evidence, the attack-path narrative, and the **three deliverables (technical report, stakeholder summary, how-to-recreate), all as PDF** via `scripts/render-report.sh`. - **`references/lab-targets.md`** — legal practice ranges and a learning path. - **`assets/report-template.md`** — fill-in network pentest report. ## Working principles - **Don't break production.** Scans and exploits can crash services — start low-intensity, respect the RoE, and never run a DoS-prone or known-unstable exploit on a live target without explicit sign-off. - **Confirm before you exploit.** Prove a vulnerability with the lightest safe proof; a reliable, in-scope demonstration beats a risky one-shot exploit. - **Handle credentials carefully.** Treat any credential or hash you recover as sensitive; record where it came from and what it unlocked, and don't reuse it outside scope. - **Map the path, not just the bugs.** The valuable output of an internal test is the *attack path* (foothold → escalation → Domain Admin), not a pile of isolated findings. - **Stay in scope, always.** An interesting host outside the agreed ranges is still out of scope — note it, don't touch it. - **Log everything** with timestamps for the report and to prove scope discipline.
© hassan-algo · MIT · hassan-algo/open-sesame