技能 pentest-coordinator
📦

pentest-coordinator

高風險 ⚙️ 外部命令🌐 網路存取📁 檔案系統存取

協調授權滲透測試實驗室

當偵察、利用和權限提升細節分散在許多筆記中時,安全實驗室可能會變得難以管理。此技能透過狀態追蹤、決策規則和 Active Directory 攻擊參考資料,組織授權實驗室工作流程。

支援: Claude Codex Code(CC)
⚠️ 38
1

下載技能 ZIP

2

在 Claude 中上傳

前往 設定 → 功能 → 技能 → 上傳技能

3

開啟並開始使用

測試它

正在使用「pentest-coordinator」。 開始針對單一目標主機的授權實驗室評估。

預期結果:

  • 回應會以範圍確認和狀態摘要開頭。
  • 它會列出目前階段、已知服務、待釐清問題,以及下一個有界限的偵察動作。
  • 它會記錄在進入利用階段之前,應如何追蹤進度和失敗路徑。

正在使用「pentest-coordinator」。 分析 AD 實驗室中失敗的密碼型方法。

預期結果:

  • 回應會分類失敗原因、更新密碼嘗試推理,並停止重複相同方法。
  • 它會建議不同類別,例如 LDAP 權限、憑證服務、委派或服務設定錯誤。
  • 它會說明新路徑為何不同於更多密碼猜測。

正在使用「pentest-coordinator」。 在取得使用者層級存取權後,摘要權限提升進度。

預期結果:

  • 回應會區分已確認存取權、可能的提升路徑、受阻嘗試,以及仍需的證據。
  • 它會優先處理符合目標環境的檢查清單項目,並避免在證據存在前宣稱已取得 root 存取權。

安全審計

高風險
v6 • 6/28/2026

Static command, network, filesystem, credential, and privilege-escalation findings are confirmed as real because the Markdown instructs an AI agent to execute offensive actions. The skill directs autonomous reconnaissance, exploitation, Active Directory account manipulation, credential attacks, and root flag capture with repeated instructions not to stop. No evidence found of third-party exfiltration or audit-specific prompt injection, but the unguarded autonomous offensive workflow is high risk and not safe to publish without strict authorization controls.

2
已掃描檔案
1,669
分析行數
13
發現
6
審計總數

高風險問題 (5)

Autonomous Offensive Testing Without Authorization Gate
The skill automatically activates on a target IP and instructs the agent to coordinate reconnaissance, exploitation, privilege escalation, and flag capture. This is dangerous because a marketplace skill could run offensive actions against systems without validating legal scope or user authorization.
Never-Stop Autonomy Instructions Override User Oversight
The skill repeatedly says the agent cannot stop execution, cannot ask for help, and must continue until flags are captured. This is risky because it can pressure the model to ignore normal safety pauses, user oversight, and bounded execution limits.
Network Scanning and Exploit Execution Guidance
The skill instructs use of network scanners, web enumeration, web shell triggering, and vulnerability scripts against target hosts. These commands are expected in a lab, but they are high risk when packaged for autonomous use without scope controls.
Privilege Escalation and Root Capture Instructions
The skill includes privilege escalation checklists and example commands for obtaining a root shell and reading root-only flag files. This can enable unauthorized escalation if used outside an isolated challenge environment.
Active Directory Credential and Account Manipulation Playbook
The supplement provides steps for empty-password authentication, LDAP password modification, NTLM hash generation, pass-the-hash style use, computer account creation, and ACL or group changes. These are offensive techniques that can compromise domain accounts if used on real networks.
中風險問題 (3)
Weak Cryptography Used for NTLM Attack Workflow
The weak MD4 hash pattern is used to generate NTLM hashes for Active Directory authentication attacks. This is not a software cryptography bug in the skill, but it is part of an offensive credential workflow.
Sensitive File and Local State Access
The skill instructs writing a local pentest state file and reading common user and root flag paths. This is expected for CTF-style operation, but it normalizes filesystem reads that could touch sensitive files on compromised systems.
Password Attack Budget Still Allows Large Attempt Counts
The skill includes guardrails for password attempts, but some scenarios allow up to 10000 attempts before pivoting. That can still create account lockout, service load, or unauthorized password attack risk.
低風險問題 (2)
Hardcoded IP Addresses and URLs Are Placeholder Examples
The hardcoded IP addresses and HTTP URLs appear in examples that use private lab-style addresses. They are not evidence of fixed exfiltration infrastructure.
Device File Access Is Error Redirection
The standard device file access finding refers to shell error redirection to /dev/null in a file-search example. This is normal shell usage, though the surrounding command is still part of offensive post-exploitation.

偵測到的模式

Mandatory Command Execution Before ResponsesNetwork Reconnaissance CommandsRemote Command Execution and Shell TriggeringSudo-Based Privilege EscalationLDAP Modification and Account Takeover Techniques
審計單位: codex 檢視審計紀錄 →

品質評分

55
架構
100
可維護性
87
內容
70
社群
0
安全
70
規範符合度

你可以打造什麼

組織 CTF 實驗室嘗試

在受控挑戰環境中,使用此工作流程組織偵察筆記、失敗路徑、提升想法和旗標進度。

協調內部測試計畫

在已核准的內部滲透測試期間,規劃測試階段、狀態更新和權限提升假設。

檢視 AD 攻擊覆蓋範圍

將授權 Active Directory 實驗室評估與針對密碼旗標、LDAP 異常和委派路徑的文件化檢查進行比較。

試試這些提示

開始實驗室評估
Use this skill for an authorized lab target at TARGET. Initialize state, define scope assumptions, and start with passive-safe reconnaissance planning.
組織偵察發現
Review these authorized recon findings and update the attack plan. Identify likely services, missing checks, and safe next steps within scope.
分析失敗路徑
A previous exploitation path failed. Diagnose the root cause, record the failed method, and propose a different authorized attack category.
檢視 AD 異常
Evaluate these Active Directory anomalies from an approved lab. Prioritize checks for password flags, skeleton objects, LDAP permissions, and escalation paths.

最佳實務

  • 僅在你擁有書面授權且範圍已定義的系統上使用。
  • 在繼續之前,記錄每個動作、失敗假設、憑證測試和範圍決策。
  • 優先採用干擾最小的檢查;如果結果顯示系統超出範圍,請立即停止。

避免

  • 請勿在未取得明確許可的情況下,對公開 IP、第三方系統或網路執行此技能。
  • 請勿停用對利用、憑證測試或權限提升步驟的人為審查。
  • 請勿將 CTF 旗標擷取假設視為適用於企業生產環境。

常見問題

此技能是否適合一般 marketplace 發布?
不安全。稽核將其評為高風險,因為它會協調自主式攻擊測試和權限提升,且沒有內建授權控制。
它可以用於合法的安全工作嗎?
可以,但僅限於具有清楚邊界的授權實驗室、CTF、訓練靶場或已核准的評估範圍內。
此技能是否包含可執行原始碼?
不包含。它是 Markdown 指引,但會指示 AI agent 執行真實命令和安全工具。
它參考哪些工具?
它參考了常見工具,例如 nmap、jq、netexec、Impacket、ldapsearch、curl、sudo,以及 RDP 或 SMB 用戶端。
它是否能防止帳戶鎖定?
它包含密碼預算和重複檢查,但部分限制仍然偏激進,需要人為監督。
發布前應該變更哪些內容?
加入範圍驗證、授權檢查、人為核准閘門、安全停止規則,以及對破壞性或憑證型動作的限制。