界面已本地化;技能介绍与说明文档按所选语言由 AI 自动翻译,代码、命令、路径和专业术语保留原文,以原作者文档为准。
探索技能
开发与工程 · SKILL.md

red-team-pentest-methodology

red-team-pentest-methodology

【开发与工程】Understand the theoretical frameworks of offensive security, OSINT, and attack lifecycles.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准

技能说明

GitHub

name: red-team-pentest-methodology description: Understand the theoretical frameworks of offensive security, OSINT, and attack lifecycles.

Red Team & Penetration Testing Methodology

The Cyber Kill Chain & Attack Lifecycle

Offensive security operations model adversarial progression through distinct phases. The Lockheed Martin Cyber Kill Chain establishes the prerequisite sequence for successful exploitation: Reconnaissance, Weaponization, Delivery, Exploitation, Installation, Command and Control (C2), and Actions on Objectives. Disrupting any node in this graph theoretically neutralizes the threat.

MITRE ATT&CK Framework Integration

The MITRE ATT&CK framework provides a standardized taxonomy of adversary Tactics, Techniques, and Procedures (TTPs). Red team engagements map operational steps to these matrices (e.g., Initial Access, Execution, Persistence) to objectively measure the defensive posture and detection efficacy of the target organization.

OSINT Conceptual Framework

Open-Source Intelligence (OSINT) represents the passive intelligence gathering phase. It relies on aggregating public footprints—DNS records, BGP routes, organizational charts, and digital exhaust—to construct a comprehensive attack surface map without direct interaction with the target's infrastructure.

Theoretical Mechanics of Privilege Escalation

Privilege escalation exploits conceptual flaws in system architecture and access control models. It involves transitioning from a constrained security context to an administrative one. Vertically, this relies on abusing kernel vulnerabilities, misconfigured service permissions, or token manipulation. Horizontally, it leverages stolen credentials to move laterally across equivalent trust domains. The focus is strictly on architectural mapping of trust boundaries, devoid of actionable exploitation payloads.

%%{init: {"theme": "default", "flowchart": {"useMaxWidth": true}}}%%
flowchart TD
    A[OSINT & Reconnaissance] --> B[Target Surface Mapping]
    B --> C[Weaponization Strategy]
    C --> D[Delivery Mechanisms]
    D --> E[Exploitation of Vulnerability]
    E --> F[Installation & Persistence]
    F --> G[Command & Control Setup]
    G --> H[Internal Recon & Lateral Movement]
    H --> I[Privilege Escalation Mapping]
    I --> J[Actions on Objectives]
我的收藏