domain-checker

name	domain-checker
description	Bulk domain availability checker. Use when the user wants to check whether one or many domain names are available for registration (e.g., "is acme.com free", "check these 50 domains", "find an available .io for my brand"). Supports any TLD (.com/.net/.io/.ai/.dev/.app/...), parallel checking, three input modes (CLI args / file / stdin), bare-name + TLD-list expansion, and table/JSON/CSV output. Python stdlib only — no pip installs, no API keys.
author	George Khananaev

Domain Checker

Check domain availability in bulk using RDAP (Registration Data Access Protocol) as the primary signal, with WHOIS and DNS fallbacks. Python stdlib only.

When to Use

• User asks "is <domain> available / taken / registered?"
• User wants to check many domains at once (a list, a file, names piped in)
• User wants to test a brand name across many TLDs (acme.com, acme.net, acme.io, …)
• User wants machine-readable output (CSV/JSON) for further processing
• User wants only available (or only taken) results filtered

Do NOT use for: WHOIS contact lookup (this skill only checks existence), DNS record inspection (use dig), trademark/legal availability (this only checks registration status).

How It Works

1. RDAP first — fetches the IANA RDAP bootstrap (data.iana.org/rdap/dns.json, cached 24h in ~/.cache/domain-checker/) to find the authoritative RDAP server for each TLD. HTTP 404 = available, HTTP 200 = taken.
2. WHOIS fallback — if RDAP is unavailable for a TLD (notably .io, some ccTLDs) or returns ambiguous, shells out to the local whois CLI and matches well-known "not found"/"registered" patterns.
3. DNS — optional, weak signal. Resolves A/AAAA records. Confirms "taken" only — registered-but-unhosted domains will show no DNS, so DNS alone cannot prove "available".

After the primary check, three enrichment passes refine accuracy:

4. Aftermarket detection — parses the RDAP response we already fetched. Flags aftermarket when the registrant/entity name matches a known reseller (HugeDomains, Sedo, Afternic, Dan.com, NameJet, DropCatch, …) OR when the nameservers match parking patterns (sedoparking.com, parkingcrew.net, namefind.com, afternic.com, bodis.com, …). Free — no extra requests.
5. RDAP status parsing — surfaces pendingDelete / redemptionPeriod / clientHold / serverHold as a separate expiring status (these domains may drop and become available within ~30 days).
6. Premium-likely hint — for available results, appends a warning when the name is short (≤5 chars on .ai/.io/.app/.dev/.co/.tv/.me/.cc/.tech/.xyz or ≤3 chars on .com/.net/.org). This is a heuristic, not a hard claim — the registrar checkout has the real price.
7. Optional HTTP sale-page probe (--probe-sale) — fetches the homepage and looks for "for sale" / Sedo / Dan.com / HugeDomains / Afternic markers. Reclassifies taken → aftermarket on a hit. Off by default; adds ~1s per checked domain.

Concurrency via ThreadPoolExecutor (default 10 workers). Per-request timeout default 8s.

Prerequisites

• Python 3.8+ (stdlib only)
• Outbound HTTPS to RDAP servers
• Optional: whois CLI (brew install whois / apt install whois) for fallback on TLDs without RDAP

Quick Routing

Task	Command
Check single domain	check_domains.py acme.com
Check many domains	check_domains.py acme.com acme.net acme.io
Bare name across TLDs	check_domains.py acme --tlds com,net,io,ai,dev
From a file	check_domains.py --file domains.txt
From stdin	cat list.txt | check_domains.py --stdin
Only available	check_domains.py --file list.txt --only-available
JSON output	check_domains.py acme.com --format json
CSV output	check_domains.py acme.com --format csv
Force WHOIS only	check_domains.py acme.io --method whois
Fastest (DNS only)	check_domains.py acme.com --method dns

Usage

python3 .claude/skills/domain-checker/scripts/check_domains.py [DOMAINS...] [OPTIONS]

Examples

# Single domain
python3 scripts/check_domains.py google.com

# Many domains
python3 scripts/check_domains.py acme.com acme.net brandidea.io

# Brand-name search across TLDs
python3 scripts/check_domains.py mybrand --tlds com,net,io,ai,dev,app

# From a file (one domain per line, # comments allowed)
python3 scripts/check_domains.py --file domains.txt

# From stdin
printf 'foo.com\nbar.io\n' | python3 scripts/check_domains.py --stdin

# Only available, as CSV (great for piping into a sheet)
python3 scripts/check_domains.py --file candidates.txt --only-available --format csv > available.csv

# JSON for programmatic use
python3 scripts/check_domains.py acme --tlds com,net,io --format json

# Higher parallelism for big lists
python3 scripts/check_domains.py --file 500-domains.txt -c 25

# Exit code reflects unknown/invalid results (for CI)
python3 scripts/check_domains.py --file list.txt --exit-code

Output

Default table format:

   domain         status        method  evidence                                                    ms
-  -------------  ------------  ------  ----------------------------------------------------------  ----
✓  acme.dev       available     rdap    RDAP 404 from https://pubapi.registry.google/rdap           437
✓  bzqj.dev       available     rdap    RDAP 404 …; likely premium tier (.dev <=5 chars)            383
✗  google.com     taken         rdap    RDAP 200 from https://rdap.verisign.com/com/v1              333
$  foothill.com   aftermarket   rdap    RDAP 200 …; parking NS: ns1.afternic.com                    352
~  someone.com    expiring      rdap    RDAP 200 …; status: pendingdelete                           340
?  bizarre.tld    unknown       rdap    no RDAP server for .tld                                     12
!  bad..domain    invalid               invalid domain syntax                                       0

Symbol	Status	Meaning
✓	available	No registration record found (RDAP 404 or WHOIS no-match)
✗	taken	Domain is registered (legit ownership)
$	aftermarket	Registered but held by a domain investor / parking service — usually for resale at premium prices
~	expiring	Registered but in pendingDelete / redemptionPeriod / clientHold — may drop and become available
?	unknown	Neither method could decide (rate-limited, no RDAP server, etc.) — retry, or use a registrar to confirm
!	invalid	Domain syntax invalid
·	reserved	TLD-reserved (rare, e.g. some registry-internal names)

For available results, the evidence column may include likely premium tier (...) — a heuristic warning that the registry probably charges a premium for this short name on this TLD.

Flags

Flag	Default	Description
positional DOMAINS	—	Zero or more domains or bare names
-f, --file PATH	—	Read domains from file (one per line, # comments OK)
--stdin	off	Read domains from stdin
--tlds com,net,...	—	TLDs to append to bare names (only used for inputs without a .)
-m, --method	auto	auto (rdap→whois), rdap, whois, dns, or comma list
-c, --concurrency N	10	Parallel workers
-t, --timeout S	8.0	Per-request timeout, seconds
--format	table	table, json, or csv
--only-available	off	Filter output to available only
--only-taken	off	Filter output to taken / aftermarket / expiring
--no-color	off	Disable ANSI colors in table output
--no-aftermarket-detect	off	Skip RDAP entity / nameserver analysis (faster, less accurate)
--no-premium-hint	off	Don't warn on short available names
--probe-sale	off	HTTP-probe taken domains for "for sale" landing pages (~1s/domain)
--exit-code	off	Exit non-zero if any unknown/invalid result (useful in CI)

--only-available and --only-taken are mutually exclusive (last one wins).

Method Notes

Method	Pros	Cons
rdap	Standardized JSON; works for nearly all gTLDs; fast; no parsing fragility	Some ccTLDs (.io, .co.uk, …) don't expose RDAP
whois	Covers TLDs without RDAP	Output format varies per registry; rate-limited; requires CLI installed
dns	Very fast	Cannot prove "available" — registered-but-unhosted domains have no DNS
auto (default)	RDAP first, WHOIS fallback when RDAP is unavailable/inconclusive	Slightly slower on TLDs that need fallback

Improving Accuracy

Default mode already runs aftermarket detection + premium hints (no extra cost — they reuse the RDAP response). For maximum accuracy on important brand searches, add --probe-sale:

python3 scripts/check_domains.py mybrand --tlds com,net,io,ai,dev --probe-sale

What each enrichment catches:

Enrichment	Detects	Cost
Aftermarket via RDAP entities	HugeDomains / Sedo / Afternic / Dan.com / NameJet etc. as the registrar or registrant	Free — re-uses RDAP response
Aftermarket via parking nameservers	afternic.com, sedoparking.com, parkingcrew.net, namefind.com (GoDaddy investor portfolio), bodis.com, dan.com, hugedomains.com	Free
RDAP status parsing	pendingDelete / redemptionPeriod / clientHold	Free
Premium-likely hint	≤5 chars on .ai/.io/.app/.dev/.co/.tv/.me/.cc/.tech/.xyz; ≤3 chars on .com/.net/.org	Free (local)
--probe-sale HTTP fetch	"For sale" landing pages (Sedo, Dan.com, HugeDomains, GoDaddy Auctions, Afternic)	~1s/domain

What this can't catch

Limitation	Why
Exact premium price	No free API exposes registry-tier pricing. Always check the registrar checkout.
Trademark conflicts	Requires USPTO TESS / WIPO lookups (out of scope).
Privately-held investor domains	Brokers using cloaked WHOIS + non-parking nameservers (e.g. Cloudflare) will look like normal "taken" — only --probe-sale can sometimes catch these via the landing page.
Domains in pendingCreate / about to register	RDAP shows them as available; you'll race the registrant at checkout.

Caching

The IANA RDAP bootstrap is cached for 24 hours at:

~/.cache/domain-checker/rdap-bootstrap.json

Delete that file to force a refresh.

Limitations

• Rate limits: registry RDAP servers will rate-limit aggressive bulk checks. For lists >200 domains, consider -c 5 and reasonable spacing.
• Truth-source: a domain marked available here may still be unregisterable (premium, reserved by registry, or held in pending-delete). Always confirm at the registrar checkout.
• Privacy: every RDAP/WHOIS query is logged by the registry. Don't bulk-check competitor brand lists from infrastructure you don't want associated with the queries.
• No registrar pricing: this skill only reports availability — pricing varies wildly across registrars and is not queried.

See Also

• references/tlds.md — common TLDs and which method works best for each