PenTestClaw v2.0 — 渗透测试 AI 助手

架构

配置层（薄）          能力层（4 Skills）
SOUL.md  ─┐          ┌─ cve-intel    → CVE 情报获取
USER.md   ├─ 人格    │  poc-forge    → PoC 锻造
AGENTS.md ─┤  路由   │  asset-hunt   → 资产狩猎
TOOLS.md  ─┘         └─ mass-verify  → 批量验证

AI（Opus 4.6）= 编排器，不需要 Python 编排脚本。

核心链路

cve-intel → poc-forge → asset-hunt → mass-verify
  查CVE详情    锻造PoC    Fofa搜资产    批量验证+报告

部署

# 1. 安装 OpenClaw
npm install -g openclaw

# 2. 复制工作区
cp -r workspace/ ~/.openclaw/workspace/

# 3. 安装 Python 依赖
pip3 install requests aiohttp python-dotenv

# 4. 配置密钥
mkdir -p ~/.pentestclaw
cat > ~/.pentestclaw/secrets.env << 'EOF'
FOFA_EMAIL=your_email@example.com
FOFA_KEY=your_fofa_api_key
NVD_API_KEY=your_nvd_key_optional
PROXY_URL=
EOF

# 5. 创建工作目录
mkdir -p ~/pentest/{authorizations,targets,logs}
mkdir -p ~/pentest/results/{intel,pocs,assets,verify,reports}

# 6. 启动
openclaw

推荐模型配置

{
  "name": "PenTestClaw",
  "language": "zh-CN",
  "model": {
    "provider": "openrouter",
    "model": "anthropic/claude-sonnet-4-20250514",
    "temperature": 0.3,
    "max_tokens": 8192
  },
  "skills": {
    "enabled": true,
    "local_skills_dir": "./workspace/skills",
    "auto_install": false
  },
  "security": {
    "require_confirmation": ["shell", "network"],
    "max_network_requests_per_minute": 60,
    "log_all_operations": true
  }
}