RustScan and NSE Pipeline

Definition

The RustScan + NSE pipeline is a two-stage workflow in which RustScan performs fast asynchronous port discovery on a target and pipes the discovered port list directly into Nmap, which then drives the Nmap Scripting Engine (NSE) for service detection, version probing, and category-scoped vulnerability checks.

Why it matters

Running nmap -A -p- target against a single host can take 10–30 minutes — and most of that time is spent on ports that turned out to be closed. Splitting the work into "discover ports fast → enumerate ports deeply" cuts a single-host full-range scan from minutes to seconds without losing any of Nmap's enumeration power, because the deep stage runs only on the open ports.

The senior framing is that this is the same two-phase pattern as Masscan + Nmap, just sized down: RustScan replaces the bulk-discovery layer for single hosts and small subnets, while Masscan owns the same role for AS-scale work. Understanding NSE script categories then determines whether the second phase is safe, useful, or accidentally noisy/destructive.

How it works

The pipeline is 3 stages:

1. Discovery (RustScan). Async Tokio runtime opens up to --ulimit concurrent half-open sockets, sweeps the configured port set, and emits an open-port list. Adaptive learning adjusts batch size based on early response patterns.
2. Handoff. Everything after -- on the RustScan command line is passed verbatim to Nmap. RustScan invokes Nmap with -p <discovered_ports> and the user's flags.
3. Enumeration (Nmap + NSE). Nmap runs version detection, OS fingerprinting, and any selected NSE scripts on the open ports only.

Example:

rustscan -a target --ulimit 5000 -- -sV -sC --script "default,safe,vuln"

Interpretation: - --ulimit 5000 raises the open-file-descriptor limit so 5000 ports can be probed concurrently. - Everything after -- is Nmap's: -sV (version), -sC (default NSE), --script "default,safe,vuln" (category Boolean). - RustScan exec()s Nmap with -p <ports it found> automatically. Nmap then ignores all the closed ports RustScan already ruled out.

The bug RustScan is solving is not in Nmap; it is in how Nmap is typically run (-p- against a full host where 65,500 ports were always going to be closed). The pipeline is an optimization, not a replacement.

NSE script categories are the second part of the senior story:

• auth, broadcast, brute, default, discovery, dos, exploit, external, fuzzer, intrusive, malware, safe, version, vuln.
• Safe engagement default: --script "default,safe" plus selectively discovery and version.
• Add vuln only with explicit authorization.
• intrusive, exploit, brute, dos require explicit consent and a documented blast-radius assessment.

Techniques / patterns

• Master the -- divider. Everything after -- is Nmap. If you can't articulate what -A -sC --script vuln does at the Nmap level, RustScan is hiding bugs from you, not saving you time.
• --ulimit ≥ batch size, always. Default failure mode: RustScan throttles to ulimit and silently reports false filtered. ulimit -n in the shell and --ulimit on the CLI must agree.
• Use --scripts None when you want only the port list (e.g., feeding into a custom pipeline).
• Boolean script selection beats long script lists: --script "(default or safe) and not (brute or dos)".
• Per-port NSE selectivity. --script "http-*" -p 80,443 runs HTTP-only scripts; avoid pulling SMB scripts at HTTP ports.
• Script arguments are mandatory for vuln checks that need credentials or paths: --script-args 'http.useragent="custom",userdb=users.txt'.
• NSE script update before any serious engagement: nmap --script-updatedb.
• Stdin chaining for pipelines: subfinder -d target.com | dnsx -a -resp-only | rustscan -a - -- -sV -sC.

Variants and bypasses

1. Single-host deep scan

rustscan -a host -r 1-65535 --ulimit 65535 -- -A -sC --script vuln. The original use case — full port range, full enumeration, in roughly the time Nmap version detection alone would have taken.

2. Subnet sweep with selective NSE

rustscan -a 10.0.0.0/24 --ulimit 5000 -- -sV --script "(default or safe) and not (brute or intrusive)". Subnet-scale discovery with conservative NSE.

3. Port-list-only mode

rustscan -a host --scripts None -g. Greppable port list, no Nmap. For custom pipelines (Nuclei, custom scanners) that don't want Nmap's overhead.

4. HTTP-focused enumeration

rustscan -a host -p 80,443,8080,8443 -- -sV --script "http-title,http-headers,http-methods,http-enum,http-server-header". Targeted HTTP NSE without pulling unrelated script categories.

5. Authenticated vuln category

rustscan -a host -- -sV --script vuln --script-args "creds.global=user:pass". NSE vuln scripts that check authenticated-only conditions. Requires explicit engagement scope authorizing credentials.

6. Containerized scan

docker run --rm rustscan/rustscan:latest -a target -- -sV -sC. The official deployment — avoids host ulimit fights, reproducible across operators.

Impact

• 10–100× wall-clock reduction vs. naive nmap -p- -A.
• Operational risk if NSE category is wrong. --script "default" includes some scripts that send authentication attempts (http-auth-finder, pop3-brute is not default but easy to add) — review the category contents before assuming "default = safe".
• NSE vuln false positives. Many vuln scripts check banner versions and miss patches — every alert requires manual validation before reporting.
• exploit and dos categories can crash services. They are excluded from default for a reason; opting in needs change-management approval in any production-touching engagement.
• Detection footprint. RustScan + NSE is louder than Nmap alone — same per-IP fingerprint plus NSE's script-specific probes are easily clustered by behavioral SIEM rules.

Detection and defense

1. NSE-aware IDS rules. Snort/Suricata signatures for the more-noisy NSE scripts (http-enum, smb-vuln-*, ssl-enum-ciphers against a closed cluster) are publicly available and tuned. Defenders should subscribe.

2. Per-port deep-inspection rate limits. The RustScan handoff means Nmap will hit a small port set with many NSE probes in quick succession. Per-port-per-source rate limits at the WAF/IDS catch this distinctly.

3. Version banner shaping. Many NSE vuln scripts decide on version strings. Defenders that strip or randomize version banners reduce false-positive surface in engagements and reduce true-positive value for attackers — both reasons to do it.

4. Honeypots that respond to http-enum. A path table that returns 200 to every directory NSE probes makes the script's output meaningless without affecting real users. Asymmetric defense.

What does not work as a primary defense

• Blocking RustScan's source IP — same trap as Masscan: it finishes its job in seconds.
• Disabling specific Nmap probes — Nmap controls its probe library, not the target.
• Hoping --script vuln reports are accurate — most are version-string heuristics. Defenders should not treat an absence of NSE-vuln alerts as evidence of safety.

Practical labs

# Lab 1 — baseline pipeline against an authorized lab host.
rustscan -a LAB --ulimit 5000 -- -sV -sC --script "default,safe"
# Time it. Compare to `nmap -p- -sV -sC LAB`.
# Expect 5–20× wall-clock reduction.

# Lab 2 — ulimit mismatch failure mode.
ulimit -n 1024
rustscan -a LAB -b 5000 -- -sV 2>&1 | grep -i ulimit
# Reproduces the "Too many open files" error. The fix is `--ulimit 65535` (and matching shell ulimit).

# Lab 3 — NSE category Boolean.
nmap --script "(default or vuln) and not (intrusive or dos)" --script-help | head -40
nmap -sV --script "(default or vuln) and not (intrusive or dos)" -p 22,80,443 LAB
# Inspect which scripts ran. Match the category list against the engagement scope.

# Lab 4 — pipeline from subdomain enum.
subfinder -d target.tld -silent |
  dnsx -a -resp-only -silent |
  rustscan -a - --ulimit 5000 -- -sV --script "default,safe"
# A single chain that goes from a domain to enumerated services.
# Only run against authorized scope.

# Lab 5 — port-list-only mode for custom downstream.
rustscan -a LAB --scripts None -g
# Output is one line per host: "10.0.0.5 -> [22,80,443]".
# Pipe to Nuclei, gowitness, ffuf, or custom tooling instead of Nmap.

# Lab 6 — NSE update and script audit.
sudo nmap --script-updatedb
locate nse_main.lua  # find script path
ls $(nmap --script-help default 2>/dev/null | awk '/^Categories:/{exit} {print $1}' | head -5)
# Always know what `default` contains before claiming "I only ran safe scripts".

Practical examples

• Bug-bounty single-target deep dive. RustScan + NSE against an in-scope host completes full-range version detection in 30–60 seconds instead of 15–30 minutes — letting an operator review 10× more targets in a session.
• CI-driven external attack surface. Nightly RustScan + NSE default,safe against a list of owned subdomains, with results diffed against the previous night. New ports/services or version changes trigger an alert.
• Internal red-team pivot. From a foothold, rustscan -a internal-subnet/24 --ulimit 5000 -- -sV --script "default,safe and not brute" finds high-value services (Jenkins, GitLab, internal Confluence) within minutes.
• Cloud workload audit. From a controlled scanner inside a VPC, RustScan + NSE validates that intended security-group exposure matches actual reachability for every workload.
• Pre-engagement smoke test. Before kicking off the real engagement, a --script default,safe pass confirms tooling works end-to-end and produces a baseline diff target.

Related notes

• Nmap Scanning
• Service Enumeration
• Ports and Services
• Host and Port Discovery
• Service Validation
• Active Recon
• Enumeration
• Scan Anomaly Detection and Fingerprint Analysis
• EDR Network Observability and Process Correlation
• nmap-timing-and-evasion
• packet-fragmentation-and-decoy-scans
• masscan-internet-scale-scanning

Suggested future atomic notes

• nse-vuln-category-audit
• writing-custom-nse-scripts
• naabu-vs-rustscan
• scan-pipeline-orchestration

References

• Official Tool Docs: RustScan repository and docs — https://github.com/RustScan/RustScan
• Official Tool Docs: Nmap Scripting Engine — https://nmap.org/book/nse.html
• Research / Deep Dive: ProjectDiscovery Reconnaissance 103 — Host and Port Discovery — https://projectdiscovery.io/blog/reconnaissance-series-3-host-and-port-discovery