netexpert-mcp
Uses 1.1.1.1 as a trusted DNS resolver for DNS propagation and hijack checks, comparing the system resolver against this trusted service.
Uses Cloudflare's DNS-over-HTTPS endpoint (cloudflare-dns.com) for DNS propagation and hijack checks, and performs bounded download speed tests against Cloudflare's network.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@netexpert-mcp what devices are on my network, and has anything new joined?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
netexpert-mcp
"What's on my network, is my link healthy, is this machine safe, and why is X slow?" — answered by an AI, on your LAN, in one conversation.
netexpert is a cross-platform local-network diagnostics & security-posture MCP server for AI clients (Claude Desktop, Claude Code, or any client that speaks MCP). Windows, Linux, macOS — one normalized JSON envelope, 31 tools, zero config on first run, read-only by default.
It's the local host + LAN half of a home-lab toolkit — the NAS half already exists as ugnas-mcp, and the two compose.
Also listed in the MCP registry as io.github.Omodaka9375/netexpert-mcp.
Install
The package is on npm — no clone, no build:
# run once, on demand (no install)
npx -y netexpert-mcp
# or install globally
npm install -g netexpert-mcp
netexpert-mcpWire it into your MCP client (stdio):
{
"mcpServers": {
"netexpert": {
"command": "npx",
"args": ["-y", "netexpert-mcp"]
}
}
}That's it — first run needs no config file, no environment variables, nothing. Ask your client "what devices are on my network?" and it has the tools to answer.
Optional HTTP transport (stateless, same JSON-RPC surface):
NETEXPERT_HTTP_PORT=3100 netexpert-mcp
# POST http://127.0.0.1:3100/mcp (tools/list, tools/call)
# GET http://127.0.0.1:3100/health (liveness)Related MCP server: Pingo
Try it in 30 seconds
Once wired into your client, try:
"Am I exposed?" → the client calls
sec.exposure— listening ports × firewall state × a risky-port knowledge base, reported asFindings."What's on my network?" →
lan.devices— ARP inventory enriched with IEEE-OUI vendor names and PTR hostnames."Is my DNS lying to me?" →
dns.hijackcheck— known-good domains via your system resolver vs trusted DoH; mismatch ⇒Finding{severity:high}."Is my WiFi any good?" →
wifi.health— RSSI quality bands, channel overlap, band mismatch — interpreted, with next steps.
Why it exists
Existing network MCP servers are either deprecated Python packages, wrappers that demand Nmap, or router-config tools for enterprises. None of them answer the home-lab questions:
Question | netexpert answer |
What devices are on my network — anything new? |
|
Is my WiFi actually good? |
|
Is this machine exposed? |
|
Is my DNS lying to me? |
|
Why is X unreachable? | Layered triage — link → ARP → routing/DNS → port → app ( |
Design guarantees
One envelope. Every tool returns the same JSON shape (
ok,backend,data,warnings/ closed-seterror.code+ actionablehint) — no rawnetshoripoutput ever reaches the model.Platform-normalized. Same field names on Windows, Linux, and macOS; platform dispatch lives below the service layer.
Bounded everything. Port caps, host caps, byte caps, timeouts, and token-bucket rate limits are enforced in code — never left to the model's judgment.
No shell, ever. All subprocesses spawn via arg arrays (
execFile-style); all inputs are zod-validated before touching the OS. No raw sockets, no Npcap, no Python.Read-only by default. The one mutating tool (
lan.wol) uses a dry-run plan + single-use, 10-min-TTLconfirmToken— parameters frozen at plan time.Capability honesty. Tools report which backend produced the data (
backend: "tcp-connect", not a fake "ping");net.healthself-describes what works on this host, with install hints for what doesn't.No telemetry. No phone-home — except the three tools that are about the internet (
net.publicip,sec.speedtest, DoH indns.hijackcheck), which you invoked explicitly.
Tool catalog (31)
net.* — connectivity
Tool | What it does |
| Version, platform, capability matrix (which backends exist on this host, with install hints) |
| ICMP loss/latency stats; honest TCP-connect fallback when ping is unavailable |
| Traceroute hops, normalized across |
| GET/HEAD with timing split: dns / connect / ttfb / total |
| Single TCP connect — open/closed + latency |
| Bounded connect scan (≤256 ports, default top-20 common) |
| TLS cert: subject, SANs, issuer, chain, days-left, expiry grade |
| Your public IPv4 + ASN/org (ipify → ipinfo fallback) |
| Optional nmap adapter — opt-in only: requires the binary and |
dns.* — name resolution
Tool | What it does |
| A/AAAA/CNAME/MX/TXT/NS/SRV/PTR with optional nameserver override |
| PTR names for an IP |
| Which DNS servers the OS actually uses (per-platform) |
| Same record at system + trusted resolvers + DoH — diff table |
| The killer check: known-good domains via system vs trusted DoH; mismatch or NXDOMAIN ⇒ |
lan.* — devices & inventory
Tool | What it does |
| NICs: name, MAC, addresses, type, virtual flag |
| Default gateway + interface + local subnet CIDR |
| Parsed neighbor cache, stale entries filtered |
| Ping-probe an RFC1918 subnet (≤1024 hosts, rate-limited 1/30s) → live set |
| Inventory: ARP ∩ RFC1918, enriched with vendor (40k-entry IEEE OUI db) + PTR hostname |
| All known names for an IP: PTR + NetBIOS (win) |
| Save device inventory to dated snapshots; or list them |
| Diff live vs baseline ⇒ findings (new MAC, MAC-changed-for-IP); first run auto-saves a baseline |
| Wake-on-LAN magic packet — dry-run plan → |
wifi.* — link health
Tool | What it does |
| Current link: SSID, BSSID, band, channel, RSSI, PHY rate, quality 0–100 |
| Nearby AP survey with band inference |
| Interpretation: RSSI bands (≥−50 … <−78 unusable), 2.4 GHz channel overlap, band mismatch, low PHY rate — each a |
sec.* — host posture
Tool | What it does |
| Active sockets with owning PID |
| Listeners per process; flags |
| Get-NetFirewallProfile / ufw→nft / socketfilterfw |
| Headline report: exposed listeners × firewall state × risky-port table (445, 3389, 23, 135–139, 5357, 5900) ⇒ |
| Bounded Cloudflare download; reports a lower bound instead of failing on slow links |
Configuration (all optional)
~/.netexpert-mcp/config.json — sane defaults, so zero-config first run:
{
"defaults": { "pingCount": 4, "traceMaxHops": 20 },
"resolvers": { "trusted": ["1.1.1.1", "8.8.8.8"],
"dohUrl": "https://cloudflare-dns.com/dns-query" },
"caps": { "portscan": { "maxPorts": 256 }, "sweep": { "maxHosts": 1024 } },
"rate": { "sweep": 30, "portscan": 10, "speedtest": 60, "httpCheck": 2 },
"nmap": { "enabled": false } // deep scans are opt-in by design
}Invariants:
caps are ceilings — tool inputs may go lower, never higher
env overrides via
NETEXPERT_*:NETEXPERT_MCP_CONFIG(config path),NETEXPERT_HTTP_PORT,NETEXPERT_WIFI_IF(linux wlan iface)no credentials ever — any config key containing
password/token/secret/credentialis rejected at load
Optional extras:
nmap deep scans — install nmap (
winget install Insecure.Nmap/sudo apt install nmap), then set"nmap": { "enabled": true }in config. Absence just produces an install hint, never a crash.OUI vendor db — ships inside the package; never fetched at runtime.
Monitoring without a daemon
netexpert stays request/response; recurring checks are your MCP client's job:
New-device tripwire lan.newdevices every 15 min fire on Finding[] beyond the info baseline
Gateway reachability net.ping <gw> every 60 s fire on lossPct > 0
DNS integrity dns.hijackcheck hourly fire on sev=high
WiFi degradation wifi.health every 10 min fire on severity ≥ medAI usage — docs/persona.md
The server provides hands; the persona provides judgment. The rule pack in docs/persona.md gives your AI client:
a triage ladder (link → ARP → IP/DNS → port → app — always state which layer exonerated/implicated),
baselines for "suspicious" (new MAC with no vendor match, RFC1918 host on 445/3389 that isn't the NAS, resolver ≠ DoH, ARP flapping),
the Finding format — every answer ends with the concrete next tool call, never a data dump.
Development
git clone https://github.com/Omodaka9375/netexpert-mcp.git
cd netexpert-mcp
pnpm install
pnpm typecheck # strict TS (Kai/Claude Code runs this after edits)
pnpm test # 50 fixture-driven tests (fake runner, zero real network in units)
pnpm build # → dist/
pnpm smoke # build → stdio → tools/list → net.health
pnpm smoke:http # HTTP transport check
node scripts/gen-oui.mjs # regenerate the OUI db (rarely needed)Project layout:
src/
exec/ LocalExecutor (execFile, arg arrays) + per-platform CommandMaps (win32/linux/darwin)
parse/ pure parsers (arp, ping, traceroute, netsh, iw, nmap -oG) — fixture-tested
services/ coreService dnsService lanService wifiService secService nmapService
safety/ caps.ts (bounds + token-bucket rate limiter) · confirm.ts (dry-run/confirmToken)
data/ oui.json (40k IEEE OUI entries, committed)
mcp/ toolset.ts (31 tools, zod inputs) · createServer.ts
scripts/ gen-oui.mjs · smoke.mjs · http-smoke.mjs
docs/ persona.md · index.html (the docs page)
tests/ fixtures/ + 50 tests across 6 filesLicense: MIT — see LICENSE.
npm: netexpert-mcp · MCP registry: io.github.Omodaka9375/netexpert-mcp
Source & issues: GitHub
Available Tools
31 toolsdns.hijackcheckC
Killer check: known-good domains via system resolver vs trusted DoH; mismatch ⇒ Finding.
Inputs: no named parameters.
dns.lookupC
A/AAAA/CNAME/MX/TXT/NS/SRV/PTR lookup, optional nameserver override.
Inputs: name (required), record, nameserver.
dns.propagationC
Same record queried at trusted resolvers + DoH; diff table.
Inputs: name (required), record.
dns.resolversC
Which DNS resolvers the OS actually uses.
Inputs: no named parameters.
dns.reverseC
PTR names for an IP.
Inputs: no named parameters.
lan.arpB
ARP/neighbor cache, parsed, stale entries filtered.
Inputs: no named parameters.
lan.devicesB
Inventory: ARP ∪ RFC1918 filter, enriched with vendor (OUI) + hostname (PTR).
Inputs: no named parameters.
lan.gatewayC
Default gateway + interface + local subnet CIDR.
Inputs: no named parameters.
lan.hostnameC
All known names for an IP: PTR, NetBIOS (win).
Inputs: ip (required).
lan.interfacesC
Network interfaces (Nic[]): name, mac, ipv4[], ipv6[], type.
Inputs: no named parameters.
lan.newdevicesA
Diff live devices vs latest snapshot ⇒ Finding[] (new MAC, MAC change). First run saves baseline.
Inputs: no named parameters.
lan.snapshotA
Save current lan.devices to a dated snapshot; or list snapshots.
Inputs: mode.
lan.sweepB
Ping-probe an RFC1918 subnet (cap 1024 hosts) → live set.
Inputs: subnet.
lan.wolB
Wake-on-LAN magic packet (UDP broadcast). Mutating: dry-run returns confirmToken; re-call with token to execute.
Inputs: mac (required), port, confirmToken.
net.deepscanB
Optional nmap deep scan (LAN RFC1918 targets). Requires nmap on PATH AND nmap.enabled=true in config. Opt-in by design.
Inputs: host (required), ports, topPorts.
net.healthC
Server version, platform, capability matrix, config caps.
Inputs: no named parameters.
net.httpcheckC
GET/HEAD with timing split (dns/connect/ttfb/total), status, headers, size. LAN-safe.
Inputs: url (required), method, timeoutMs.
net.pingC
ICMP ping (min/avg/max/mdev), TCP-connect fallback reported honestly.
Inputs: host (required), count, timeoutMs.
net.portcheckC
Single TCP connect: open/closed + latency.
Inputs: host (required), port (required).
net.portscanC
Bounded TCP connect scan, default top common ports, cap 256.
Inputs: host (required), ports.
net.publicipA
Public IPv4 (+ASN/org). Explicitly user-requested external call.
Inputs: no named parameters.
net.sslB
TLS cert inspection: subject, SANs, issuer, expiry, daysLeft, grade.
Inputs: host (required), port, timeoutMs.
net.traceC
Traceroute hop list (ttl, ip, latency).
Inputs: host (required), maxHops, timeoutMs.
sec.connectionsC
Active sockets with owning PID/process.
Inputs: no named parameters.
sec.exposureB
Composite security report: exposed listeners × firewall state × risky ports ⇒ Finding[].
Inputs: no named parameters.
sec.firewallC
Firewall profile state per platform (win profiles / ufw / socketfilterfw).
Inputs: no named parameters.
sec.listeningB
Listening ports per process; flags non-loopback exposures.
Inputs: no named parameters.
sec.speedtestA
Bounded download speed test (Cloudflare default, configurable). Rate-limited 1/min.
Inputs: no named parameters.
wifi.healthB
WiFi interpretation: RSSI bands, channel overlap, band mismatch ⇒ Finding[].
Inputs: no named parameters.
wifi.networksB
Nearby AP survey: SSID, BSSID, band, channel, security.
Inputs: no named parameters.
wifi.statusA
Current WiFi link: SSID, BSSID, band, channel, RSSI, PHY rate, quality.
Inputs: no named parameters.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 31 tools
Most tools target a distinct resource+action, and descriptions explicitly differentiate close pairs (net.portcheck single connect vs net.portscan bounded scan; lan.arp raw cache vs lan.devices enriched inventory). A few residual overlaps remain: dns.lookup already covers PTR while dns.reverse exists, dns.propagation and dns.hijackcheck both diff system resolver against trusted DoH, and sec.listening is effectively a filtered subset of sec.connections.
Every tool follows the same lowercase namespace.action convention (net.*, dns.*, lan.*, wifi.*, sec.*) with no separator or case mixing. The namespace prefix also maps cleanly to functional areas, making the set scannable.
31 tools exceeds a comfortable surface for a single agent to select from and sits above the heavy threshold, though the coverage genuinely spans five distinct diagnostic areas (net/dns/lan/wifi/sec). There is little filler and each tool has a real purpose, but the sheer count increases selection cost.
The surface covers a broad diagnostics lifecycle: reachability, DNS, LAN discovery and drift detection, WiFi health, and exposure/security reporting, with sensible opt-in gating for nmap and WOL. Minor gaps exist (no route table/IPv6-focused probing, no packet capture or bandwidth/latency soak between hosts), but agents can work around them.
Maintenance
Related MCP Connectors
Network, domain and website diagnostics for AI clients via MCP.
IP intelligence, network diagnostics, and AI service availability tools for AI agents.
DNS, email auth, TLS, BGP/RPKI, WHOIS and subnet tools with plain-English fixes, plus webhook URLs.
Free, no-API-key network lookups for AI agents — IP geolocation, DNS, WHOIS, HTTP and SSL/TLS certificate checks, and full domain health reports.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables AI assistants to perform real network diagnostics on the local machine, including ping, traceroute, DNS lookups, TLS checks, and more.MIT
- AlicenseNot gradedqualityCmaintenanceEnables AI agents to debug local networks in plain language by pinging, sweeping subnets, tracing routes, resolving DNS, scanning ports, inspecting ARP tables and active connections, and checking URL reachability.MIT
- FlicenseAqualityBmaintenanceEnables LLM clients to run structured, bottom-up network diagnostics across OSI Layers 1–3, including Wi-Fi telemetry, ping checks, and runbook-based remediation.7-
- AlicenseBqualityCmaintenanceEnables AI assistants to discover devices and perform connectivity diagnostics across a private Tailscale or WireGuard mesh through authenticated CLI or REST API backends. Supports safe dry-run simulations before affecting live node state.2Apache 2.0