Welcome to The Network Pentester's Field Manual. This is not a beginner primer: it is a working reference for people who test internal and external networks for a living, or who want to. Every chapter is theory and practice: the why, the exact commands, the tricks that save hours, and the detections that prove you understand both sides.
One rule above all. Everything here is for systems you own or are explicitly authorised in writing to test, within an agreed scope and window. Unauthorised testing is a crime. Keep the signed authorisation with your notes.
What a network pentest actually delivers
Clients don't pay for a pile of nmap output. They pay for an answer to one question: what can an attacker who reaches this network actually do, and how do we stop them? Your job is to find the shortest realistic path from "foothold" to "impact", prove it safely, and hand back the fix.
External or assumed-breach foothold
│
▼
recon ─► scan ─► enumerate ─► validate
│
▼
exploit ─► privesc ─► pivot ─► lateral
│
▼
objective (DA, crown-jewel data, impact)
│
▼
report + detections + retestThe engagement lifecycle
| Phase | What happens | Your deliverable |
|---|---|---|
| Pre-engagement | Scope, rules of engagement, authorisation, timing, contacts | Signed RoE and scope doc |
| Intelligence gathering | Passive and active recon | Target map |
| Threat modelling | What matters to this client, what an attacker would want | Prioritised test plan |
| Vulnerability analysis | Scan, enumerate, validate | Verified findings list |
| Exploitation | Prove impact safely | Evidence per finding |
| Post-exploitation | Privesc, pivot, reach objectives | Attack path narrative |
| Reporting | Write it up for four audiences | The report and retest |
This is the PTES (Penetration Testing Execution Standard) shape. NIST SP 800-115 and OSSTMM describe the same arc with different emphasis. Map your findings to MITRE ATT&CK technique IDs so blue teams can turn each one into a detection.
Engagement types you'll be asked for
- External: test from the internet against the perimeter. Small attack surface, high stakes.
- Internal: you're dropped on the LAN (or a VM on it). The classic "assumed breach" test, where most real damage happens.
- Assumed breach / insider: you start with a low-privileged domain user or a workstation, and model a phished employee.
- Segmentation test: prove whether VLANs and firewalls actually contain an attacker (PCI DSS asks for this).
- Red team: objective-based, stealthy, often with detection evasion in scope and a blue team watching.
The techniques overlap; what changes is starting point, stealth and objective. Agree all three in the RoE.
OPSEC and safety on the job
Even on an authorised test, you can cause outages and raise alarms. Professionals manage that:
- Know your blast radius. Avoid
nmapscripts and exploits tagged intrusive or DoS on production unless the RoE allows them. Fragile targets (printers, ICS, medical, legacy) get gentle treatment or stay out of scope. - Throttle by default. Start slow, speed up where it's safe. A scan that knocks over a domain controller ends your engagement.
- Announce your source IPs. Give the client the addresses you'll test from so their SOC can tell you apart from a real attacker (unless stealth is the point).
- Have a stop procedure. Know who to call if something breaks, and stop immediately if you hit data you shouldn't (PII, PHI, card data) and document it.
- Don't be the attacker you're simulating. Never exfiltrate real data, never destroy, view the minimum needed to prove impact.
Evidence discipline
Your report is only as good as your notes. Capture as you go, not from memory at the end.
- Timestamped command log.
script,tmuxlogging, or a tool like Pwndoc/Sysreptor / a simple markdown file per host. - Screenshots with the command, output and a clock visible.
- Per-host notes: IP, hostname, OS, open ports, creds found, what worked.
- Scope ledger: what you touched and when, in case something breaks and you need to show it wasn't you.
# Record a whole terminal session with timing
script -q -a engagement-$(date +%F).log
# Organised output dir per engagement
mkdir -p loot/{nmap,creds,screenshots,notes}The network tester's toolkit
You'll meet all of these in later chapters. Kali Linux or Parrot ships most; keep a VM you can revert.
| Job | Primary | Alternatives |
|---|---|---|
| Discovery / scanning | Nmap | masscan, RustScan, fping |
| SMB / AD enum | NetExec (nxc, the maintained CrackMapExec) | enum4linux-ng, smbclient |
| Vuln scanning | OpenVAS / Greenbone, Nuclei | Nessus (commercial) |
| Web | Burp Suite, ffuf | nikto, gobuster |
| Exploitation | Metasploit | manual PoCs, searchsploit |
| Creds / cracking | hashcat, John | hydra (online) |
| Pivoting | ligolo-ng, Chisel | sshuttle, proxychains |
| AD enum | BloodHound, ldapsearch | PowerView, adidnsdump |
| C2 / post-ex | Sliver, Metasploit | (red team scope only) |
| Packet work | Wireshark, tcpdump | tshark, Responder (lab) |
| Notes / reporting | Sysreptor, Obsidian | CherryTree, markdown |
How to use this manual
Each chapter stands alone as a reference but builds a complete workflow. Commands are copy-ready; replace TARGET, 10.10.10.0/24 and $IP with your own authorised targets. The 🧪 Lab at the end of each chapter runs on free, legal targets: your own VMs, HackTheBox, TryHackMe, or VulnHub boxes.
🧪 Lab
- Build a lab: a Kali VM plus two or three vulnerable targets (Metasploitable 2/3, a Windows eval VM, a vulnerable VulnHub box) on a host-only network.
- Create your engagement folder structure and start a
scriptlog. - Write a one-page mock rules-of-engagement for your lab: scope, window, stop procedure.
- Install and verify the toolkit table above (
nmap --version,nxc --version, etc.).
Checkpoint
- What single question does a network pentest report answer?
- Name the seven PTES phases in order.
- Why do you give the client your source IPs before testing?
- What three things change between engagement types, while the techniques stay the same?
- Why capture evidence continuously instead of at the end?