netdoctor
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
netdoctor runs safe, read-only network checks on the user's own computer to help diagnose slow or broken internet. Suggested order when someone says "my internet is slow/broken": ping the router (often 192.168.1.1 or 192.168.0.1; the first traceroute hop shows it), then ping 1.1.1.1, then dns_lookup a well-known name, then traceroute to whatever is slow. check_port tells you whether one specific service (e.g. port 443 on a website) can be reached. Calls are rate-limited (1 per second, 20 per minute), so plan checks rather than retrying in a loop. Never use check_port to look for open ports: it allows only a few ports per host.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| pingA | Ping one host (hostname or IP) to see if it is reachable, with latency, jitter and packet loss. count: 1-10 pings (default 4). |
| dns_lookupA | Look up one DNS name and time the lookup. record_type: A, AAAA, CNAME, MX, NS or TXT. server: optional resolver to ask instead of the system one (e.g. 1.1.1.1), useful to tell whether the local DNS is the problem. |
| tracerouteA | Trace the network path to one host and show the delay at each hop, to find where traffic slows down or stops. max_hops: 1-30 (default 15). Takes up to a minute or two. |
| check_portA | Check whether ONE TCP port on ONE host accepts connections (e.g. 443 for a website). Opens one normal connection and closes it; sends no data. Not a port scanner: limited to a few different ports per host every 10 minutes. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 4 tools
Each tool targets a distinct diagnostic layer: ICMP reachability (ping), name resolution (dns_lookup), path latency (traceroute), and TCP reachability (check_port). There is no plausible case where an agent would confuse these, and check_port's description explicitly disclaims port scanning.
Two tools are bare command names (ping, traceroute) while the others use differing orders: noun_verb (dns_lookup) and verb_noun (check_port). The names are all readable and snake_case, but the underlying convention is mixed rather than a single predictable pattern.
Four tools is exactly the right scope for a focused network diagnostic server, covering the core checks without padding. Each tool has a clear, non-redundant job.
Reachability, DNS, path tracing, and port checks cover the main triage workflow, with useful options like custom resolvers and hop limits. Minor gaps remain: no reverse DNS/PTR lookup, no HTTP/TLS handshake verification, and no batch or multi-host variant for comparing targets.