Run External Recon Scan Pipeline

Goal

Map the open ports, services, and version exposure of an authorized target (single host, subnet, or BGP AS) using a two-phase Masscan → RustScan/Nmap-NSE pipeline, with reproducible timing, an enforced exclude-list, and structured output that diffs cleanly across runs.

Assumptions

a written authorization (engagement scope, bug-bounty policy, or owned-asset list) exists and covers every IP/CIDR that will appear in the include list
the scan box has a stable source IP and a clean upstream link (no shared NAT with sensitive workloads)
the goal is breadth-then-depth recon, not exploitation; vulnerability validation happens in follow-up playbooks
Nmap, Masscan, RustScan, jq, and whois are installed and current

Prerequisites

INCLUDE.txt — authorized targets (one CIDR/IP/host per line)
EXCLUDE.txt — out-of-scope IPs, partner integrations, monitoring infra, RFC 5737 ranges if relevant
--adapter-ip value matching the scan box's actual routable source IP
nmap --script-updatedb run within the last 30 days
shell ulimit -n 65535 available (or use the RustScan Docker image)

Recon steps

Runnable reference script: cybersecurity/security-playbooks/run-scan-pipeline.sh packages Phases 1–2 below into one shell script. Use it with INCLUDE_FILE, EXCLUDE_FILE, and ADAPTER_IP env vars; the script refuses to run if EXCLUDE_FILE is empty and prompts for an explicit YES before scanning unless --yes is passed. Phase 0 (authorization gate) and Phase 3 (diagnostic packet-shape matrix) are kept as human-in-the-loop steps and are not automated.

Defender pair: detect-external-scan-pipeline is the note-by-note defender-side mirror of this playbook — read both together per Attacker-Defender Duality.

Phase 0 — Authorization gate

Print the scope and exclude list verbatim and confirm against the engagement document.
cat INCLUDE.txt EXCLUDE.txt — visually verify before any packets go out.
If targets are AS-level, expand the AS to prefixes first: bash whois -h whois.radb.net -- "-i origin AS<NUMBER>" | awk '/^route:/ {print $2}' | sort -u > INCLUDE.txt

Phase 1 — Breadth (Masscan)

Echo-validate the Masscan config before scanning: bash sudo masscan --excludefile EXCLUDE.txt -iL INCLUDE.txt \ -p1-65535 --rate 1000 --adapter-ip <YOUR_IP> --echo > masscan.conf less masscan.conf # eyeball the merged include/exclude
Calibrate --rate on a tiny slice first (-p80 on one /24) and watch for ICMP-unreachable / loss spikes. Set the real rate to ~½ of the highest clean rate.
Run the scan with binary output for lossless re-parsing: bash sudo masscan -c masscan.conf -oB scan.bin
Convert to JSON for downstream tooling: bash sudo masscan --readscan scan.bin -oJ scan.json jq -r '.[] | "\(.ip):\(.ports[0].port)"' scan.json | sort -u > openports.txt
If interrupted, resume with sudo masscan --resume paused.conf — never restart from zero.

Phase 2 — Depth (RustScan → Nmap NSE)

Group Masscan output by host so Nmap runs once per host with that host's port list: bash awk -F: '{ports[$1]=ports[$1]","$2} END {for (h in ports) print h, substr(ports[h],2)}' openports.txt > hostports.txt
Run Nmap with version detection and a conservative NSE category Boolean: bash while read host ports; do sudo nmap -Pn -sS -sV -sC \ --script "(default or vuln) and not (intrusive or dos or brute)" \ --min-rate 100 --max-rate 300 --max-retries 3 \ -p "$ports" -oA "nmap/$host" "$host" done < hostports.txt
For single-host deep scans where Masscan-scale isn't needed, substitute RustScan as the discovery layer: bash rustscan -a <HOST> --ulimit 5000 -- -sV -sC \ --script "(default or safe or vuln) and not (intrusive or dos or brute)" \ --min-rate 100
If a target appears rate-limited (high filtered count at high rate, lower at low rate), pin --min-rate 50 --max-rate 100 and re-run only that host.

Phase 3 — Diagnostic packet-shape matrix (optional)

Run only when Phase 2 reports everything as filtered despite expected exposure. Pick one suspected-open port and probe with each evasion primitive separately to fingerprint the inspection layer:

PORT=443; HOST=<TARGET>
for FLAG in "" "-f" "--mtu 24" "--data-length 200" "--badsum" "--source-port 53"; do
  echo "===== $FLAG"
  sudo nmap -Pn -sS -p $PORT $FLAG "$HOST"
done

Whichever invocation returns open reveals what the middlebox does and does not inspect — that intel guides the next phase.

Exploit / test steps

This playbook intentionally stops at enumeration. Validated findings route to: - exploitation playbooks per service class (exploit-sqli, break-jwt-validation, inspect-file-upload-surface, etc.) - cybersecurity/offensive-security/service-validation for manual service inspection - cybersecurity/offensive-security/recon-to-testing-handoff for engagement handoff

Validation clues

open ports appear identically across Masscan and Nmap runs (Masscan ground-truth on discovery, Nmap on protocol)
Version banners are consistent across NSE scripts (-sV, http-server-header, ssl-cert) — divergence indicates a load balancer with mixed backends
filtered counts drop when --min-rate drops — confirms rate limiting, not firewall block
A --badsum probe returning RST/ACK confirms an inline IDS not validating checksums (real OSes drop)
Diff against the previous run shows new ports/services — the actual engagement output

Mitigation

On the defender side, the right controls for the pipeline being run against you: - per-source-IP SYN-rate alerting at the perimeter (defeats Masscan in seconds) - IDS rules tied to fan-out behavior (one source → many destinations) and TCP fingerprint (defeats decoys) - IP fragment reassembly enabled at the IDS (defeats -f/--mtu) - honeyports/tarpits on never-used ports (asymmetric high-confidence detection) - version-banner shaping (raises NSE vuln false-positive rate, lowers attacker signal)

Logging / detection

one source IP hitting > N distinct destination IPs on a single port within 60 s — Masscan signature
one source IP probing > 100 distinct ports on one host within 60 s — RustScan/Nmap signature
NSE-specific probe patterns in IDS rule packs (Snort/Suricata community sets)
per-source TCP-fingerprint stability across "different" source IPs — decoy detection
EDR process-network joins that tie scanner behavior to nmap, masscan, rustscan, python, powershell, or renamed binaries

Operational safety

never use real third-party IPs as -D decoys; prefer RFC 5737 ranges (192.0.2.0/24, 198.51.100.0/24, 203.0.113.0/24)
never scan above --rate 1000 on a shared upstream link without measuring loss first
never run NSE intrusive, exploit, dos, or brute categories without explicit written authorization
always keep EXCLUDE.txt non-overridable — Masscan honors it before INCLUDE.txt, which is the safety property
always write binary output for any scan that takes more than five minutes — Ctrl-C survival matters

References

Official Tool Docs: Nmap Reference Guide — Timing and Performance — https://nmap.org/book/man-performance.html
Official Tool Docs: Masscan README — https://github.com/robertdavidgraham/masscan
Official Tool Docs: RustScan repository and docs — https://github.com/RustScan/RustScan
Research / Deep Dive: ProjectDiscovery Reconnaissance 103 — Host and Port Discovery — https://projectdiscovery.io/blog/reconnaissance-series-3-host-and-port-discovery