security-pentest-recon-scan
Installation
SKILL.md
SKILL: security-pentest-recon-scan
Wrapper-only recon + scan guidance. Recon-first, passive-before-active. Planning/Read-only by design — like
pentest-scope-and-roe, this skill reads, plans, and explains; the project's validated-wrapper scripts run the CLIs andappsec-sdkpersists evidence. Sits betweenpentest-scope-and-roe(ROE) andauthorized-pentest-validation(exploitation hard gate). NEVER bridges to exploitation on its own.
1. Mission
为已授权的渗透测试提供 recon(侦察)+ vulnerability scan(漏洞扫描)的包装化调用指导:
- 侦察:amass(攻击面 / 子域枚举)、httpx(存活探测 / 指纹)
- 扫描:nuclei(模板化漏扫,non-intrusive 模板优先)、ffuf(web 内容发现 / fuzzing)
核心设计 = wrapper-only(直接对治 benchmark 实证的 LLM 失败主因:"乱调原始安全 CLI" —— 错参数 / 不支持 flag / 误开破坏性模式)。每个 CLI 都经项目自带的 validated wrapper 调用(固定安全 flag + 参数校验 + target allowlist + 速率上限),禁止 raw free-form 调用。这与 dast-baseline-scanning 的 wrapper 安全边界完全一致。