hexstrike-ai MCP — ユースケース, インストール & ライブデモ

なぜ使うのか

主な機能

150+ integrated security tools across network, web, binary, cloud, OSINT, and forensics
12+ autonomous AI agents for intelligent decision-making and workflow management
Real-time dashboard with live monitoring and vulnerability reporting
Smart result caching with LRU eviction for performance
Headless Chrome integration for dynamic web application testing

ライブデモ

実際の動作

hexstrike-ai.replay ▶ 準備完了

0/0

インストール

クライアントを選択

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "hexstrike-ai": {
      "command": "TODO",
      "args": [
        "See README: https://github.com/0x4m4/hexstrike-ai"
      ]
    }
  }
}

Claude Desktop → Settings → Developer → Edit Config を開く。保存後、アプリを再起動。

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "hexstrike-ai": {
      "command": "TODO",
      "args": [
        "See README: https://github.com/0x4m4/hexstrike-ai"
      ]
    }
  }
}

Cursor は Claude Desktop と同じ mcpServers スキーマを使用。プロジェクト設定はグローバルより優先。

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "hexstrike-ai": {
      "command": "TODO",
      "args": [
        "See README: https://github.com/0x4m4/hexstrike-ai"
      ]
    }
  }
}

Cline サイドバーの MCP Servers アイコンをクリックし、"Edit Configuration" を選択。

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "hexstrike-ai": {
      "command": "TODO",
      "args": [
        "See README: https://github.com/0x4m4/hexstrike-ai"
      ]
    }
  }
}

Claude Desktop と同じ形式。Windsurf を再起動して反映。

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "hexstrike-ai",
      "command": "TODO",
      "args": [
        "See README: https://github.com/0x4m4/hexstrike-ai"
      ]
    }
  ]
}

Continue はマップではなくサーバーオブジェクトの配列を使用。

~/.config/zed/settings.json

{
  "context_servers": {
    "hexstrike-ai": {
      "command": {
        "path": "TODO",
        "args": [
          "See README: https://github.com/0x4m4/hexstrike-ai"
        ]
      }
    }
  }
}

context_servers に追加。保存時に Zed がホットリロード。

claude mcp add hexstrike-ai -- TODO 'See README: https://github.com/0x4m4/hexstrike-ai'

ワンライナー。claude mcp list で確認、claude mcp remove で削除。

ユースケース

実用的な使い方： hexstrike-ai

How to solve a multi-stage CTF challenge with HexStrike AI

👤 CTF participants working on security challenges ⏱ ~60 min intermediate

使うタイミング： You have a CTF target and need to enumerate, scan, and exploit in a structured way.

前提条件

HexStrike AI installed in a sandboxed environment — Clone and pip install in a VM or container
Target is a CTF challenge you are authorized to test — Only use on CTF platforms or labs you have permission to test

フロー

Reconnaissance

Run an Nmap scan on the CTF target at 10.10.10.1. Identify open ports and services.✓ コピーしました

→ Port scan results with service versions
Web enumeration

The target has a web server on port 80. Run Gobuster for directory enumeration and Nikto for vulnerability scanning.✓ コピーしました

→ Discovered directories and potential vulnerabilities
Exploitation assistance

Found a login page at /admin. Run Hydra with the top-1000 passwords list against it.✓ コピーしました

→ Login attempt results

結果： Systematic enumeration and exploitation path for the CTF challenge.

注意点

Running aggressive scans that crash the CTF target — Use moderate scan speeds and timing options (Nmap -T3 instead of -T5)

組み合わせ： filesystem

Perform an authorized web application security assessment

👤 Security professionals with written authorization to test a web application ⏱ ~120 min advanced

使うタイミング： You have a signed agreement to test a client's web application.

前提条件

Written authorization for the target — Signed penetration testing agreement from the target owner
HexStrike AI configured — Install and configure in an isolated environment

フロー

Vulnerability scanning

Run Nuclei against https://target.example.com with default templates. Focus on critical and high severity findings.✓ コピーしました

→ List of discovered vulnerabilities with severity ratings
SQL injection testing

Test the login endpoint for SQL injection using SQLMap.✓ コピーしました

→ SQLMap results showing injectable parameters or confirming the endpoint is secure

結果： A documented list of vulnerabilities found during authorized testing.

注意点

Testing outside the agreed scope — Always reference your authorization document. Only test explicitly listed assets.

組み合わせ： filesystem

組み合わせ

他のMCPと組み合わせて10倍の力を

hexstrike-ai + filesystem

Save scan results and generate a penetration testing report

Run a full assessment on the target, save all results, and generate a professional pentest report as a markdown file.✓ コピーしました

ツール

このMCPが提供する機能

ツール	入力	呼び出すタイミング
nmap_scan	target: str, options?: str	Network port scanning and service detection
nuclei_scan	target: str, templates?: str	Automated vulnerability scanning with templates
sqlmap_test	url: str, params?: str	SQL injection detection and exploitation testing
gobuster_dir	url: str, wordlist?: str	Directory and file brute-force enumeration
hydra_attack	target: str, service: str, userlist?: str, passlist?: str	Password brute-force testing on services

コストと制限

運用コスト

APIクォータ: N/A — all tools run locally
呼び出しあたりのトークン: 500–3000 tokens per scan result
金額: Free — all tools are open source. Some tools (Shodan) need their own API keys.
ヒント: Run targeted scans instead of full-spectrum sweeps to reduce noise and time.

セキュリティ

権限、シークレット、影響範囲

認証情報の保管： Individual tool API keys (Shodan, etc.) in environment variables

データ送信先： Network scans connect to target systems. OSINT tools call their respective APIs.

⚠ This tool is designed for authorized security research, CTF competitions, and defensive analysis only. Do not use it against systems you don't own or have written authorization to test.
These are real offensive security tools — unauthorized use is illegal in most jurisdictions.
Always run in an isolated environment (VM, Docker) to prevent accidental exposure.
Some tools (Hydra, Hashcat) perform brute-force attacks — only use on systems you're authorized to test.

トラブルシューティング

よくあるエラーと対処法

Tool not found / command not available

Many tools must be installed separately on your system. Install them via your package manager (apt, brew) or from source.

確認： which nmap && which nuclei && which sqlmap

Scan timeout

Reduce scan scope or increase timeout. Use -T3 for Nmap instead of default timing.

確認： Try a simple ping or single-port scan first

Permission denied on scan

Some scans (SYN scan) require root. Run with sudo or switch to a less privileged scan type.

確認： sudo nmap -sS target

代替案

hexstrike-ai 他との比較

代替案	代わりに使う場面	トレードオフ
mcp-security-hub	You want Dockerized security tools with better isolation and CI/CD integration	Better containerization but may require more setup
mcp-kali-server	You want a lightweight API bridge to a full Kali Linux installation	Full Kali toolkit but less AI agent specialization

その他

リソース

📖 GitHub の公式 README を読む

🐙 オープンな issue を見る

🔍 400以上のMCPサーバーとSkillsを見る