#!/usr/bin/env python3
"""Wave 3 of the status-page probe (config/sources.d/58-status-pages-wave3.yaml).

Same engine as scripts/gen-status-probe.py but restricted to machine-readable STATUS pages (Atlassian
Statuspage `/api/v2/summary.json`, Instatus `/summary.json`, incident.io `/api/v1/summary`, Status.io
`api.status.io/1.0/status/<id>` sniffed from the HTML) and fed with a fresh list of ≥ 900 services that
were not probed in waves 1–2 (files 23 / 34 / 41 / 48): Forbes Cloud 100 / YC companies, fintech & brokers,
crypto infrastructure, dev tools, e-commerce & marketplaces, gaming, telecom, media, travel, HR / legal /
health / education SaaS, IoT, government digital services and university IT status pages.

For every candidate the probe tries, in order: `status.<apex>` (+ the host its root redirects to),
`<domain>/status` (redirect target or embedded Statuspage host), any explicit hosts given in the 5th
column, then `<brand>.statuspage.io` and `<brand>.instatus.com`. Hosts already carrying a STATUSPAGE
sensor anywhere in config/ are never probed again; URLs already in config/ are never emitted.

Usage:
  python3 scripts/gen-status-probe3.py probe [--out /tmp/probe58.jsonl] [--workers 12] [--only a.com,b.com]
  python3 scripts/gen-status-probe3.py gate  [--in /tmp/probe58.jsonl] [--drop /tmp/probe58-drop.json]
  python3 scripts/gen-status-probe3.py emit  [--in /tmp/probe58.jsonl] [--drop /tmp/probe58-drop.json] > config/sources.d/58-status-pages-wave3.yaml

`gate` re-fetches every hit that lives on a `<brand>.statuspage.io` / `<brand>.instatus.com` host and writes the URLs to
drop: pages with 0 components, Statuspage `page.updated_at` older than 2024-09-01 (abandoned pages) or a legacy `page.url`
(e.g. deepl.statuspage.io → old.deeplstatus.com). On 2026-09-13 this removed 153 of 245 such hits — brand hosts are a
weak signal (squatted, test or abandoned pages); `status.<domain>` hosts are taken as authoritative.
Then: node node_modules/tsx/dist/cli.mjs apps/engine/src/validate.ts config/sources.d/58-status-pages-wave3.yaml
      --concurrency 6 --json /tmp/r58.json --quiet && python3 scripts/prune-fragment.py … --drop-warn

Politeness: global token bucket (≤ 6 req/s across all hosts — every request goes to a different host),
15 s timeout, ≤ 1 MB per body, UA "WebSensor registry probe (contact@websensor.io)".
"""
from __future__ import annotations

import argparse
import glob
import json
import os
import re
import socket
import sys
import threading
import time
from concurrent.futures import ThreadPoolExecutor, as_completed
from urllib.parse import urlparse

import requests
import yaml

UA = "WebSensor registry probe (contact@websensor.io)"
TIMEOUT = 15
MAX_BODY = 1_000_000
RATE = 6.0
FRAGMENT = "58-status-pages-wave3.yaml"

# domain | Name | categories | country (default US) | extra status hosts (comma-separated, optional)
CANDIDATES = """
# ── productivity, notes, time tracking, forms, e-signature, scheduling ──
clockify.me | Clockify | enterprise,saas | RS
toggl.com | Toggl | enterprise,saas | EE
getharvest.com | Harvest | enterprise,saas
hubstaff.com | Hubstaff | enterprise,saas
timedoctor.com | Time Doctor | enterprise,saas
everhour.com | Everhour | enterprise,saas
nuclino.com | Nuclino | enterprise,saas | DE
slite.com | Slite | enterprise,saas | FR
getguru.com | Guru | enterprise,saas
tettra.com | Tettra | enterprise,saas
document360.com | Document360 | enterprise,saas
helpjuice.com | Helpjuice | enterprise,saas
archbee.com | Archbee | developer,saas
tldv.io | tl;dv | enterprise,saas,ai | DE
fathom.video | Fathom | enterprise,saas,ai
avoma.com | Avoma | enterprise,saas,ai
grain.com | Grain | enterprise,saas,ai
reclaim.ai | Reclaim.ai | enterprise,saas,ai
usemotion.com | Motion | enterprise,saas,ai
sunsama.com | Sunsama | consumer-tech,saas
akiflow.com | Akiflow | consumer-tech,saas | IT
getclockwise.com | Clockwise | enterprise,saas
superhuman.com | Superhuman | enterprise,saas
hey.com | HEY | consumer-tech,saas
missiveapp.com | Missive | enterprise,saas | CA
shortwave.com | Shortwave | consumer-tech,saas,ai
mem.ai | Mem | consumer-tech,saas,ai
reflect.app | Reflect | consumer-tech,saas
roamresearch.com | Roam Research | consumer-tech,saas
anytype.io | Anytype | consumer-tech,open-source | CH
capacities.io | Capacities | consumer-tech,saas | DE
tana.inc | Tana | consumer-tech,saas | NO
proofhub.com | ProofHub | enterprise,saas | IN
niftypm.com | Nifty | enterprise,saas
hive.com | Hive | enterprise,saas
zenhub.com | ZenHub | developer,saas | CA
businessmap.io | Businessmap | enterprise,saas | BG
teamgantt.com | TeamGantt | enterprise,saas
float.com | Float | enterprise,saas | GB
resourceguruapp.com | Resource Guru | enterprise,saas | GB
runn.io | Runn | enterprise,saas | NZ
workvivo.com | Workvivo | enterprise,saas | IE
staffbase.com | Staffbase | enterprise,saas | DE
simpplr.com | Simpplr | enterprise,saas
lumapps.com | LumApps | enterprise,saas | FR
happeo.com | Happeo | enterprise,saas | FI
joinblink.com | Blink | enterprise,saas | GB
process.st | Process Street | enterprise,saas
tallyfy.com | Tallyfy | enterprise,saas
kissflow.com | Kissflow | enterprise,saas | IN
pipefy.com | Pipefy | enterprise,saas | BR
nintex.com | Nintex | enterprise,saas
appian.com | Appian | enterprise,saas
pega.com | Pegasystems | enterprise,saas
outsystems.com | OutSystems | developer,saas | PT
mendix.com | Mendix | developer,saas | NL
quickbase.com | Quickbase | enterprise,saas
knack.com | Knack | developer,saas
caspio.com | Caspio | developer,saas
ninox.com | Ninox | developer,saas | DE
stackby.com | Stackby | developer,saas | IN
smartsuite.com | SmartSuite | enterprise,saas
baserow.io | Baserow | developer,open-source,saas | NL
nocodb.com | NocoDB | developer,open-source
seatable.io | SeaTable | developer,saas | DE
jotform.com | Jotform | enterprise,saas
formstack.com | Formstack | enterprise,saas
paperform.co | Paperform | enterprise,saas | AU
fillout.com | Fillout | enterprise,saas
tally.so | Tally | enterprise,saas | BE
cognitoforms.com | Cognito Forms | enterprise,saas
wufoo.com | Wufoo | enterprise,saas
123formbuilder.com | 123FormBuilder | enterprise,saas | RO
formassembly.com | FormAssembly | enterprise,saas
involve.me | involve.me | enterprise,saas | AT
feathery.io | Feathery | developer,saas
proposify.com | Proposify | enterprise,saas | CA
qwilr.com | Qwilr | enterprise,saas | AU
betterproposals.io | Better Proposals | enterprise,saas | GB
conga.com | Conga | enterprise,saas
juro.com | Juro | enterprise,legal,saas | GB
contractbook.com | Contractbook | enterprise,legal,saas | DK
spotdraft.com | SpotDraft | enterprise,legal,saas | IN
signnow.com | SignNow | enterprise,saas
signeasy.com | Signeasy | enterprise,saas
xodo.com | Xodo | enterprise,saas | CA
onespan.com | OneSpan | enterprise,cyber,saas | CA
yousign.com | Yousign | enterprise,saas | FR
oneflow.com | Oneflow | enterprise,saas | SE
scrive.com | Scrive | enterprise,saas | SE
acuityscheduling.com | Acuity Scheduling | enterprise,saas
youcanbook.me | YouCanBookMe | enterprise,saas | GB
setmore.com | Setmore | enterprise,saas
simplybook.me | SimplyBook.me | enterprise,saas | CY
appointlet.com | Appointlet | enterprise,saas
vcita.com | vcita | enterprise,saas | IL
conceptboard.com | Conceptboard | enterprise,saas | DE
stormboard.com | Stormboard | enterprise,saas | CA
klaxoon.com | Klaxoon | enterprise,saas | FR
meister.co | Meister (MindMeister) | enterprise,saas | AT
xmind.app | Xmind | consumer-tech,saas | CN
creately.com | Creately | enterprise,saas | AU
nulab.com | Nulab | developer,saas | JP
vidyard.com | Vidyard | enterprise,media,saas | CA
screencastify.com | Screencastify | education,saas
techsmith.com | TechSmith | consumer-tech,software
descript.com | Descript | media,saas,ai
riverside.fm | Riverside | media,saas | IL
zencastr.com | Zencastr | media,saas
podbean.com | Podbean | media,saas
buzzsprout.com | Buzzsprout | media,saas
transistor.fm | Transistor | media,saas
simplecast.com | Simplecast | media,saas
captivate.fm | Captivate | media,saas | GB
libsyn.com | Libsyn | media,saas
acast.com | Acast | media,saas | SE
megaphone.fm | Megaphone | media,saas
art19.com | ART19 | media,saas
castos.com | Castos | media,saas
redcircle.com | RedCircle | media,saas
spreaker.com | Spreaker | media,saas
audioboom.com | Audioboom | media,saas | GB
quillbot.com | QuillBot | consumer-tech,saas,ai
languagetool.org | LanguageTool | consumer-tech,saas | DE
prowritingaid.com | ProWritingAid | consumer-tech,saas | GB
jasper.ai | Jasper | enterprise,ai,saas
copy.ai | Copy.ai | enterprise,ai,saas
writesonic.com | Writesonic | enterprise,ai,saas
anyword.com | Anyword | enterprise,ai,saas | IL
frase.io | Frase | enterprise,ai,saas
surferseo.com | Surfer | enterprise,saas | PL
clearscope.io | Clearscope | enterprise,saas
marketmuse.com | MarketMuse | enterprise,saas
seranking.com | SE Ranking | enterprise,saas | GB
mangools.com | Mangools | enterprise,saas | SK
spyfu.com | SpyFu | enterprise,saas
serpstat.com | Serpstat | enterprise,saas | UA
screamingfrog.co.uk | Screaming Frog | enterprise,software | GB
botify.com | Botify | enterprise,saas | FR
conductor.com | Conductor | enterprise,saas
brightedge.com | BrightEdge | enterprise,saas
lumar.io | Lumar | enterprise,saas | GB
oncrawl.com | Oncrawl | enterprise,saas | FR
sistrix.com | SISTRIX | enterprise,saas | DE
sembly.ai | Sembly AI | enterprise,ai,saas
read.ai | Read AI | enterprise,ai,saas
krisp.ai | Krisp | consumer-tech,ai,saas
# ── developer tools, CI, testing, observability, incident tooling ──
gitkraken.com | GitKraken | developer,software
zed.dev | Zed | developer,open-source
tabnine.com | Tabnine | developer,ai,saas | IL
supermaven.com | Supermaven | developer,ai,saas
continue.dev | Continue | developer,ai,open-source
gitea.com | Gitea | developer,open-source | CN
codeberg.org | Codeberg | developer,open-source | DE
sr.ht | SourceHut | developer,open-source
launchpad.net | Launchpad | developer,open-source | GB
gocd.org | GoCD | developer,open-source
drone.io | Drone | developer,open-source
woodpecker-ci.org | Woodpecker CI | developer,open-source
tekton.dev | Tekton | developer,open-source
spinnaker.io | Spinnaker | developer,open-source
fluxcd.io | Flux | developer,open-source
dagger.io | Dagger | developer,open-source,saas
earthly.dev | Earthly | developer,saas
nx.dev | Nx | developer,open-source,saas
bazel.build | Bazel | developer,open-source
gradle.com | Gradle Inc. | developer,saas
depot.dev | Depot | developer,saas
blacksmith.sh | Blacksmith | developer,saas
namespace.so | Namespace | developer,saas
ubicloud.com | Ubicloud | cloud,open-source
runs-on.com | RunsOn | developer,saas
warpbuild.com | WarpBuild | developer,saas
cirrus-ci.org | Cirrus CI | developer,saas
appveyor.com | AppVeyor | developer,saas
flagsmith.com | Flagsmith | developer,open-source,saas | GB
getunleash.io | Unleash | developer,open-source,saas | NO
configcat.com | ConfigCat | developer,saas | HU
growthbook.io | GrowthBook | developer,open-source,saas
devcycle.com | DevCycle | developer,saas | CA
geteppo.com | Eppo | developer,saas
kameleoon.com | Kameleoon | enterprise,saas | FR
convert.com | Convert | enterprise,saas
raygun.com | Raygun | developer,saas | NZ
airbrake.io | Airbrake | developer,saas
honeybadger.io | Honeybadger | developer,saas
trackjs.com | TrackJS | developer,saas
appsignal.com | AppSignal | developer,saas | NL
scoutapm.com | Scout APM | developer,saas
instabug.com | Instabug | developer,saas
embrace.io | Embrace | developer,saas
coralogix.com | Coralogix | developer,saas | IL
mezmo.com | Mezmo | developer,saas
solarwinds.com | SolarWinds | enterprise,software
graylog.org | Graylog | developer,open-source,saas
observeinc.com | Observe | developer,saas
groundcover.com | groundcover | developer,saas | IL
middleware.io | Middleware | developer,saas
signoz.io | SigNoz | developer,open-source,saas
uptrace.dev | Uptrace | developer,open-source,saas
hyperdx.io | HyperDX | developer,open-source,saas
openobserve.ai | OpenObserve | developer,open-source,saas
victoriametrics.com | VictoriaMetrics | developer,open-source,saas
last9.io | Last9 | developer,saas | IN
dash0.com | Dash0 | developer,saas | DE
xmatters.com | xMatters | developer,saas
everbridge.com | Everbridge | enterprise,saas
ilert.com | ilert | developer,saas | DE
spike.sh | Spike.sh | developer,saas
zenduty.com | Zenduty | developer,saas | IN
allquiet.app | All Quiet | developer,saas | DE
signl4.com | SIGNL4 | developer,saas | DE
blameless.com | Blameless | developer,saas
instatus.com | Instatus | developer,saas
hund.io | Hund | developer,saas
sorryapp.com | Sorry™ | developer,saas | GB
statusgator.com | StatusGator | developer,saas
ohdear.app | Oh Dear | developer,saas | BE
hyperping.com | Hyperping | developer,saas | FR
freshping.io | Freshping | developer,saas
site24x7.com | Site24x7 | developer,saas | IN
uptime.com | Uptime.com | developer,saas
updown.io | updown.io | developer,saas | FR
insomnia.rest | Insomnia | developer,open-source
hoppscotch.io | Hoppscotch | developer,open-source,saas | IN
rapidapi.com | RapidAPI | developer,saas
apidog.com | Apidog | developer,saas | SG
usebruno.com | Bruno | developer,open-source | IN
redocly.com | Redocly | developer,saas
bump.sh | Bump.sh | developer,saas | FR
scalar.com | Scalar | developer,open-source,saas
smartbear.com | SmartBear | developer,software
mabl.com | mabl | developer,saas
testim.io | Testim | developer,saas | IL
functionize.com | Functionize | developer,saas
katalon.com | Katalon | developer,saas
testrail.com | TestRail | developer,saas
qase.io | Qase | developer,saas
testmo.com | Testmo | developer,saas
getxray.app | Xray | developer,saas | PT
tricentis.com | Tricentis | developer,saas | AT
perforce.com | Perforce | developer,software
headspin.io | HeadSpin | developer,saas
kobiton.com | Kobiton | developer,saas
applitools.com | Applitools | developer,saas
chromatic.com | Chromatic | developer,saas
bun.sh | Bun | developer,open-source
deno.com | Deno | developer,open-source,saas
jsdelivr.com | jsDelivr | developer,open-source | PL
unpkg.com | unpkg | developer,open-source
esm.sh | esm.sh | developer,open-source
dev.to | DEV Community | developer,media
hashnode.com | Hashnode | developer,media
nordpass.com | NordPass | cyber,consumer-tech | PA
enpass.io | Enpass | cyber,consumer-tech | IN
roboform.com | RoboForm | cyber,consumer-tech
passbolt.com | Passbolt | cyber,open-source | LU
clicksend.com | ClickSend | developer,telecom,saas | AU
textmagic.com | Textmagic | developer,telecom,saas | GB
simpletexting.com | SimpleTexting | enterprise,telecom,saas
telesign.com | Telesign | developer,telecom,cyber
resend.com | Resend | developer,saas
loops.so | Loops | developer,saas
mailersend.com | MailerSend | developer,saas | LT
elasticemail.com | Elastic Email | developer,saas | CA
socketlabs.com | SocketLabs | developer,saas
smtp2go.com | SMTP2GO | developer,saas | NZ
sendpulse.com | SendPulse | enterprise,saas | UA
moosend.com | Moosend | enterprise,saas | GR
benchmarkemail.com | Benchmark Email | enterprise,saas
emailoctopus.com | EmailOctopus | enterprise,saas | GB
sender.net | Sender | enterprise,saas | LT
mailmodo.com | Mailmodo | enterprise,saas | IN
onelogin.com | OneLogin | cyber,saas
pingidentity.com | Ping Identity | cyber,saas
jumpcloud.com | JumpCloud | cyber,saas
fusionauth.io | FusionAuth | developer,cyber,saas
ory.sh | Ory | developer,open-source,cyber | DE
goauthentik.io | authentik | developer,open-source,cyber
zitadel.com | Zitadel | developer,open-source,cyber | CH
logto.io | Logto | developer,open-source,cyber
hanko.io | Hanko | developer,open-source,cyber | DE
magic.link | Magic | developer,crypto,saas
dynamic.xyz | Dynamic | developer,crypto,saas
privy.io | Privy | developer,crypto,saas
web3auth.io | Web3Auth | developer,crypto,saas | SG
inngest.com | Inngest | developer,saas
mergent.co | Mergent | developer,saas
lob.com | Lob | developer,saas
postgrid.com | PostGrid | developer,saas | CA
cloudconvert.com | CloudConvert | developer,saas | DE
zamzar.com | Zamzar | consumer-tech,saas | GB
convertio.co | Convertio | consumer-tech,saas
deepl.com | DeepL | ai,consumer-tech,saas | DE
lokalise.com | Lokalise | developer,saas | LV
crowdin.com | Crowdin | developer,saas | EE
phrase.com | Phrase | developer,saas | DE
transifex.com | Transifex | developer,saas
smartling.com | Smartling | enterprise,saas
weglot.com | Weglot | developer,saas | FR
localazy.com | Localazy | developer,saas | CZ
onfleet.com | Onfleet | logistics,saas
bringg.com | Bringg | logistics,saas | IL
certn.co | Certn | enterprise,saas | CA
sterlingcheck.com | Sterling | enterprise,saas
tremendous.com | Tremendous | payments,saas
flatfile.com | Flatfile | developer,saas
oneschema.co | OneSchema | developer,saas
nylas.com | Nylas | developer,saas
merge.dev | Merge | developer,saas
apideck.com | Apideck | developer,saas | BE
tryfinch.com | Finch | developer,saas
rutter.com | Rutter | developer,payments,saas
codat.io | Codat | developer,finance,saas | GB
unified.to | Unified.to | developer,saas | CA
smartcar.com | Smartcar | developer,automotive,saas
ayrshare.com | Ayrshare | developer,saas
bannerbear.com | Bannerbear | developer,saas | SG
liveblocks.io | Liveblocks | developer,saas | FR
partykit.io | PartyKit | developer,open-source
zeplo.io | Zeplo | developer,saas | GB
mux.com | Mux | developer,media,saas
# ── AI infra & ML tooling (status pages only) ──
anyscale.com | Anyscale | ai,developer,saas
deepinfra.com | DeepInfra | ai,developer,saas
hyperbolic.xyz | Hyperbolic | ai,developer,saas
novita.ai | Novita AI | ai,developer,saas
nebius.com | Nebius | ai,cloud | NL
crusoe.ai | Crusoe | ai,cloud
vast.ai | Vast.ai | ai,cloud
paperspace.com | Paperspace | ai,cloud
predibase.com | Predibase | ai,developer,saas
cerebrium.ai | Cerebrium | ai,developer,saas | GB
beam.cloud | Beam | ai,developer,saas
mystic.ai | Mystic | ai,developer,saas | GB
clearml.com | ClearML | ai,developer,saas
valohai.com | Valohai | ai,developer,saas | FI
iterative.ai | Iterative (DVC) | ai,developer,open-source
hopsworks.ai | Hopsworks | ai,developer,saas | SE
arize.com | Arize AI | ai,developer,saas
whylabs.ai | WhyLabs | ai,developer,saas
fiddler.ai | Fiddler AI | ai,developer,saas
rungalileo.io | Galileo | ai,developer,saas
humanloop.com | Humanloop | ai,developer,saas | GB
braintrust.dev | Braintrust | ai,developer,saas
langfuse.com | Langfuse | ai,developer,open-source,saas | DE
helicone.ai | Helicone | ai,developer,open-source,saas
portkey.ai | Portkey | ai,developer,saas
promptlayer.com | PromptLayer | ai,developer,saas
traceloop.com | Traceloop | ai,developer,saas | IL
labelbox.com | Labelbox | ai,developer,saas
snorkel.ai | Snorkel AI | ai,developer,saas
v7labs.com | V7 | ai,developer,saas | GB
encord.com | Encord | ai,developer,saas | GB
superannotate.com | SuperAnnotate | ai,developer,saas
humansignal.com | HumanSignal (Label Studio) | ai,developer,open-source,saas
datasaur.ai | Datasaur | ai,developer,saas
character.ai | Character.AI | ai,consumer-tech
poe.com | Poe | ai,consumer-tech
you.com | You.com | ai,consumer-tech
phind.com | Phind | ai,developer,saas
assemblyai.com | AssemblyAI | ai,developer,saas
deepgram.com | Deepgram | ai,developer,saas
speechmatics.com | Speechmatics | ai,developer,saas | GB
rev.com | Rev | ai,media,saas
play.ht | Play.ht | ai,media,saas
murf.ai | Murf AI | ai,media,saas | IN
synthesia.io | Synthesia | ai,media,saas | GB
heygen.com | HeyGen | ai,media,saas
d-id.com | D-ID | ai,media,saas | IL
pika.art | Pika | ai,media,saas
lumalabs.ai | Luma AI | ai,media,saas
ideogram.ai | Ideogram | ai,media,saas | CA
leonardo.ai | Leonardo.Ai | ai,media,saas | AU
krea.ai | Krea | ai,media,saas
civitai.com | Civitai | ai,media,internet
suno.com | Suno | ai,media,saas
udio.com | Udio | ai,media,saas
voyageai.com | Voyage AI | ai,developer,saas
unstructured.io | Unstructured | ai,developer,saas
llamaindex.ai | LlamaIndex (LlamaCloud) | ai,developer,saas
cognition.ai | Cognition (Devin) | ai,developer,saas
lovable.dev | Lovable | ai,developer,saas | SE
bolt.new | Bolt.new | ai,developer,saas
v0.dev | v0 | ai,developer,saas
glean.com | Glean | ai,enterprise,saas
writer.com | Writer | ai,enterprise,saas
sierra.ai | Sierra | ai,enterprise,saas
decagon.ai | Decagon | ai,enterprise,saas
harvey.ai | Harvey | ai,legal,saas
hebbia.com | Hebbia | ai,finance,saas
abridge.com | Abridge | ai,health,saas
# ── payments, fintech, banking, brokers, crypto ──
marqeta.com | Marqeta | payments,developer
lithic.com | Lithic | payments,developer
galileo-ft.com | Galileo Financial Technologies | payments,developer
highnote.com | Highnote | payments,developer
treasuryprime.com | Treasury Prime | payments,developer
increase.com | Increase | payments,developer
column.com | Column | payments,finance
mercury.com | Mercury | finance,payments
brex.com | Brex | finance,payments
rho.co | Rho | finance,payments
airwallex.com | Airwallex | payments,finance | AU
payoneer.com | Payoneer | payments,finance
currencycloud.com | Currencycloud | payments,developer | GB
ofx.com | OFX | payments,finance | AU
remitly.com | Remitly | payments,finance
mx.com | MX | finance,developer
teller.io | Teller | finance,developer
truelayer.com | TrueLayer | finance,developer | GB
tink.com | Tink | finance,developer | SE
yapily.com | Yapily | finance,developer | GB
saltedge.com | Salt Edge | finance,developer | CA
paysafe.com | Paysafe | payments | GB
payu.com | PayU | payments | NL
sumup.com | SumUp | payments | GB
payplug.com | PayPlug | payments | FR
lydia-app.com | Lydia | payments,finance | FR
interac.ca | Interac | payments | CA
moneris.com | Moneris | payments | CA
helcim.com | Helcim | payments | CA
clover.com | Clover | payments,commerce
tide.co | Tide | finance | GB
qonto.com | Qonto | finance | FR
pennylane.com | Pennylane | finance,saas | FR
agicap.com | Agicap | finance,saas | FR
bunq.com | bunq | finance | NL
nubank.com.br | Nubank | finance | BR
webull.com | Webull | finance
moomoo.com | moomoo | finance | SG
tastytrade.com | tastytrade | finance
tradestation.com | TradeStation | finance
etrade.com | E*TRADE | finance
ally.com | Ally | finance
capitalone.com | Capital One | finance
current.com | Current | finance
varomoney.com | Varo | finance
dave.com | Dave | finance
moneylion.com | MoneyLion | finance
monarchmoney.com | Monarch Money | finance,consumer-tech
rocketmoney.com | Rocket Money | finance,consumer-tech
carta.com | Carta | finance,saas
angellist.com | AngelList | finance,saas
pulley.com | Pulley | finance,saas
ledgy.com | Ledgy | finance,saas | CH
stripe.com | Stripe | payments,developer
crypto-infra-placeholder.invalid | skip | skip
moralis.io | Moralis | crypto,developer | SE
chainstack.com | Chainstack | crypto,developer | SG
ankr.com | Ankr | crypto,developer
thegraph.com | The Graph | crypto,developer
anchorage.com | Anchorage Digital | crypto,finance
bitgo.com | BitGo | crypto,finance
copper.co | Copper | crypto,finance | GB
tether.to | Tether | crypto,finance | HK
paxos.com | Paxos | crypto,finance
magiceden.io | Magic Eden | crypto,internet
phantom.app | Phantom | crypto,consumer-tech
rainbow.me | Rainbow | crypto,consumer-tech
trezor.io | Trezor | crypto,consumer-tech | CZ
uniswap.org | Uniswap | crypto
aave.com | Aave | crypto | GB
lido.fi | Lido | crypto
coingecko.com | CoinGecko | crypto,media | SG
nansen.ai | Nansen | crypto,saas | SG
messari.io | Messari | crypto,media
glassnode.com | Glassnode | crypto,saas | CH
kaiko.com | Kaiko | crypto,finance | FR
bitget.com | Bitget | crypto | SC
gate.io | Gate.io | crypto | KY
htx.com | HTX | crypto | SC
mexc.com | MEXC | crypto | SC
bitpanda.com | Bitpanda | crypto,finance | AT
bitso.com | Bitso | crypto | MX
luno.com | Luno | crypto | GB
coinsquare.com | Coinsquare | crypto | CA
ndax.io | NDAX | crypto | CA
newton.co | Newton | crypto | CA
shakepay.com | Shakepay | crypto | CA
bitbuy.ca | Bitbuy | crypto | CA
wealthsimple.com | Wealthsimple | finance | CA
alpaca.markets | Alpaca | finance,developer
# ── e-commerce platforms, marketplaces, retail tech, logistics ──
bigcommerce.com | BigCommerce | commerce,saas
woocommerce.com | WooCommerce | commerce,open-source
commercetools.com | commercetools | commerce,saas | DE
saleor.io | Saleor | commerce,open-source | PL
medusajs.com | Medusa | commerce,open-source | DK
swell.is | Swell | commerce,saas
elasticpath.com | Elastic Path | commerce,saas | CA
fabric.inc | fabric | commerce,saas
spryker.com | Spryker | commerce,saas | DE
shopware.com | Shopware | commerce,open-source | DE
prestashop.com | PrestaShop | commerce,open-source | FR
ecwid.com | Ecwid | commerce,saas
sellfy.com | Sellfy | commerce,saas | LV
payhip.com | Payhip | commerce,saas | GB
podia.com | Podia | commerce,education,saas
teachable.com | Teachable | education,saas
thinkific.com | Thinkific | education,saas | CA
kajabi.com | Kajabi | education,saas
learnworlds.com | LearnWorlds | education,saas | CY
circle.so | Circle | consumer-tech,saas
mightynetworks.com | Mighty Networks | consumer-tech,saas
memberful.com | Memberful | commerce,saas
drip.com | Drip | commerce,saas
sendlane.com | Sendlane | commerce,saas
listrak.com | Listrak | commerce,saas
bluecore.com | Bluecore | commerce,saas
richpanel.com | Richpanel | commerce,saas
reamaze.com | Re:amaze | commerce,saas
hiverhq.com | Hiver | enterprise,saas | IN
shipstation.com | ShipStation | logistics,saas
goshippo.com | Shippo | logistics,developer
easyship.com | Easyship | logistics,saas | HK
shipbob.com | ShipBob | logistics,saas
shiphero.com | ShipHero | logistics,saas
easypost.com | EasyPost | logistics,developer
aftership.com | AfterShip | logistics,saas | HK
narvar.com | Narvar | logistics,saas
parcellab.com | parcelLab | logistics,saas | DE
route.com | Route | logistics,saas
loopreturns.com | Loop Returns | commerce,saas
coveo.com | Coveo | enterprise,ai,saas | CA
lucidworks.com | Lucidworks | enterprise,saas
searchspring.com | Searchspring | commerce,saas
klevu.com | Klevu | commerce,saas | FI
doofinder.com | Doofinder | commerce,saas | ES
poshmark.com | Poshmark | commerce
depop.com | Depop | commerce | GB
vinted.com | Vinted | commerce | LT
stockx.com | StockX | commerce
goat.com | GOAT | commerce
grailed.com | Grailed | commerce
reverb.com | Reverb | commerce
discogs.com | Discogs | commerce,media
mercari.com | Mercari | commerce | JP
offerup.com | OfferUp | commerce
zalando.com | Zalando | retail,commerce | DE
asos.com | ASOS | retail,commerce | GB
farfetch.com | Farfetch | retail,commerce | GB
overstock.com | Overstock | retail,commerce
chewy.com | Chewy | retail,commerce
boxed.com | Boxed | retail,commerce
thredup.com | thredUP | commerce
therealreal.com | The RealReal | commerce
project44.com | project44 | logistics,saas
fourkites.com | FourKites | logistics,saas
freightos.com | Freightos | logistics,saas | IL
shippeo.com | Shippeo | logistics,saas | FR
sendcloud.com | Sendcloud | logistics,saas | NL
packlink.com | Packlink | logistics,saas | ES
# ── hosting, domains, CDN, cloud, databases, queues ──
koyeb.com | Koyeb | cloud,developer | FR
northflank.com | Northflank | cloud,developer | GB
qovery.com | Qovery | cloud,developer | FR
porter.run | Porter | cloud,developer
zeabur.com | Zeabur | cloud,developer | TW
sevalla.com | Sevalla | cloud,developer | CH
kinsta.com | Kinsta | cloud,developer | CH
wpengine.com | WP Engine | cloud,developer
pantheon.io | Pantheon | cloud,developer
getflywheel.com | Flywheel | cloud,developer
pressable.com | Pressable | cloud,developer
siteground.com | SiteGround | cloud,developer | BG
bluehost.com | Bluehost | cloud,developer
hostgator.com | HostGator | cloud,developer
hover.com | Hover | internet,developer | CA
porkbun.com | Porkbun | internet,developer
dynadot.com | Dynadot | internet,developer
ionos.com | IONOS | cloud,developer | DE
hostinger.com | Hostinger | cloud,developer | LT
dreamhost.com | DreamHost | cloud,developer
a2hosting.com | A2 Hosting | cloud,developer
inmotionhosting.com | InMotion Hosting | cloud,developer
liquidweb.com | Liquid Web | cloud,developer
nexcess.net | Nexcess | cloud,developer
cloudways.com | Cloudways | cloud,developer | MT
rocket.net | Rocket.net | cloud,developer
servebolt.com | Servebolt | cloud,developer | NO
upcloud.com | UpCloud | cloud | FI
exoscale.com | Exoscale | cloud | CH
civo.com | Civo | cloud | GB
kamatera.com | Kamatera | cloud | IL
contabo.com | Contabo | cloud | DE
hostwinds.com | Hostwinds | cloud
binarylane.com.au | BinaryLane | cloud | AU
wasabi.com | Wasabi | cloud
storj.io | Storj | cloud,crypto
filebase.com | Filebase | cloud,crypto
bunny.net | bunny.net | cloud,infrastructure | SI
keycdn.com | KeyCDN | cloud,infrastructure | CH
cdn77.com | CDN77 | cloud,infrastructure | GB
gcore.com | Gcore | cloud,infrastructure | LU
aiven.io | Aiven | cloud,developer | FI
upstash.com | Upstash | cloud,developer
scalegrid.io | ScaleGrid | cloud,developer
crunchydata.com | Crunchy Data | cloud,developer
tembo.io | Tembo | cloud,developer
dragonflydb.io | Dragonfly | developer,open-source | IL
couchbase.com | Couchbase | developer,cloud
datastax.com | DataStax | developer,cloud
scylladb.com | ScyllaDB | developer,cloud | IL
tinybird.co | Tinybird | developer,cloud | ES
materialize.com | Materialize | developer,cloud
risingwave.com | RisingWave | developer,open-source,cloud
trychroma.com | Chroma | ai,developer,open-source
marqo.ai | Marqo | ai,developer,saas | AU
redpanda.com | Redpanda | developer,cloud
cloudamqp.com | CloudAMQP | developer,cloud | SE
streamnative.io | StreamNative | developer,cloud
sigmacomputing.com | Sigma Computing | enterprise,saas
omni.co | Omni | enterprise,saas
holistics.io | Holistics | enterprise,saas | SG
redash.io | Redash | developer,open-source
dnsmadeeasy.com | DNS Made Easy | internet,infrastructure
easydns.com | easyDNS | internet,infrastructure | CA
constellix.com | Constellix | internet,infrastructure
ssl.com | SSL.com | cyber,infrastructure
zerossl.com | ZeroSSL | cyber,infrastructure | AT
# ── security ──
cybereason.com | Cybereason | cyber,saas
eset.com | ESET | cyber,software | SK
bitdefender.com | Bitdefender | cyber,software | RO
kaspersky.com | Kaspersky | cyber,software | RU
avast.com | Avast | cyber,software | CZ
mcafee.com | McAfee | cyber,software
withsecure.com | WithSecure | cyber,saas | FI
trellix.com | Trellix | cyber,saas
tanium.com | Tanium | cyber,saas
automox.com | Automox | cyber,saas
action1.com | Action1 | cyber,saas
wiz.io | Wiz | cyber,saas
orca.security | Orca Security | cyber,saas | IL
aquasec.com | Aqua Security | cyber,saas | IL
sysdig.com | Sysdig | cyber,saas
checkmarx.com | Checkmarx | cyber,developer | IL
veracode.com | Veracode | cyber,developer
contrastsecurity.com | Contrast Security | cyber,developer
semgrep.dev | Semgrep | cyber,developer,open-source
gitguardian.com | GitGuardian | cyber,developer | FR
trufflesecurity.com | Truffle Security | cyber,developer
socket.dev | Socket | cyber,developer
endorlabs.com | Endor Labs | cyber,developer
chainguard.dev | Chainguard | cyber,developer
abnormalsecurity.com | Abnormal Security | cyber,saas
ironscales.com | IRONSCALES | cyber,saas | IL
knowbe4.com | KnowBe4 | cyber,saas
hoxhunt.com | Hoxhunt | cyber,saas | FI
cofense.com | Cofense | cyber,saas
hackerone.com | HackerOne | cyber,saas
bugcrowd.com | Bugcrowd | cyber,saas
intigriti.com | Intigriti | cyber,saas | BE
synack.com | Synack | cyber,saas
yeswehack.com | YesWeHack | cyber,saas | FR
imperva.com | Imperva | cyber,saas
radware.com | Radware | cyber,saas | IL
sucuri.net | Sucuri | cyber,saas
catonetworks.com | Cato Networks | cyber,saas | IL
perimeter81.com | Perimeter 81 | cyber,saas | IL
twingate.com | Twingate | cyber,saas
zerotier.com | ZeroTier | cyber,open-source
nordlayer.com | NordLayer | cyber,saas | PA
bitsight.com | Bitsight | cyber,saas
securityscorecard.com | SecurityScorecard | cyber,saas
upguard.com | UpGuard | cyber,saas | AU
panorays.com | Panorays | cyber,saas | IL
varonis.com | Varonis | cyber,saas
bigid.com | BigID | cyber,saas
onetrust.com | OneTrust | cyber,legal,saas
transcend.io | Transcend | cyber,legal,saas
osano.com | Osano | cyber,legal,saas
usercentrics.com | Usercentrics | cyber,legal,saas | DE
termly.io | Termly | legal,saas
iubenda.com | iubenda | legal,saas | IT
didomi.io | Didomi | legal,saas | FR
ketch.com | Ketch | legal,saas
cloudflare-placeholder.invalid | skip | skip
expel.com | Expel | cyber,saas
redcanary.com | Red Canary | cyber,saas
blumira.com | Blumira | cyber,saas
axonius.com | Axonius | cyber,saas
lansweeper.com | Lansweeper | enterprise,cyber | BE
keepersecurity.com | Keeper Security | cyber,saas
# ── HR, payroll, recruiting, learning, expenses ──
adp.com | ADP | enterprise,payments,saas
paycor.com | Paycor | enterprise,saas
ukg.com | UKG | enterprise,saas
dayforce.com | Dayforce | enterprise,saas
namely.com | Namely | enterprise,saas
hirevue.com | HireVue | enterprise,labour,saas
sparkhire.com | Spark Hire | enterprise,labour,saas
codility.com | Codility | developer,labour,saas | GB
hackerrank.com | HackerRank | developer,labour,saas
hackerearth.com | HackerEarth | developer,labour,saas | IN
coderpad.io | CoderPad | developer,labour,saas
codesignal.com | CodeSignal | developer,labour,saas
karat.com | Karat | developer,labour,saas
testgorilla.com | TestGorilla | enterprise,labour,saas | NL
vervoe.com | Vervoe | enterprise,labour,saas | AU
harver.com | Harver | enterprise,labour,saas | NL
oysterhr.com | Oyster | enterprise,labour,saas
papayaglobal.com | Papaya Global | enterprise,labour,saas | IL
velocityglobal.com | Velocity Global | enterprise,labour,saas
usemultiplier.com | Multiplier | enterprise,labour,saas | SG
workleap.com | Workleap | enterprise,saas | CA
cornerstoneondemand.com | Cornerstone | enterprise,education,saas
docebo.com | Docebo | enterprise,education,saas | IT
litmos.com | Litmos | enterprise,education,saas
talentlms.com | TalentLMS | enterprise,education,saas
learnupon.com | LearnUpon | enterprise,education,saas | IE
absorblms.com | Absorb LMS | enterprise,education,saas | CA
360learning.com | 360Learning | enterprise,education,saas | FR
workramp.com | WorkRamp | enterprise,education,saas
expensify.com | Expensify | finance,saas
concur.com | SAP Concur | finance,saas
rydoo.com | Rydoo | finance,saas | BE
emburse.com | Emburse | finance,saas
airbase.com | Airbase | finance,saas
payfit.com | PayFit | enterprise,payments,saas | FR
factorialhr.com | Factorial | enterprise,saas | ES
lucca.fr | Lucca | enterprise,saas | FR
humaans.io | Humaans | enterprise,saas | GB
charliehr.com | CharlieHR | enterprise,saas | GB
breathehr.com | Breathe | enterprise,saas | GB
employmenthero.com | Employment Hero | enterprise,saas | AU
keypay.com | KeyPay | enterprise,payments,saas | AU
wagepoint.com | Wagepoint | enterprise,payments,saas | CA
humi.ca | Humi | enterprise,saas | CA
collage.co | Collage HR | enterprise,saas | CA
# ── legal, real-estate, construction, field service, health, education ──
mycase.com | MyCase | legal,saas
practicepanther.com | PracticePanther | legal,saas
smokeball.com | Smokeball | legal,saas | AU
filevine.com | Filevine | legal,saas
litify.com | Litify | legal,saas
lawpay.com | LawPay | legal,payments,saas
appfolio.com | AppFolio | real-estate,saas
buildium.com | Buildium | real-estate,saas
yardi.com | Yardi | real-estate,saas
realpage.com | RealPage | real-estate,saas
entrata.com | Entrata | real-estate,saas
rentmanager.com | Rent Manager | real-estate,saas
doorloop.com | DoorLoop | real-estate,saas
turbotenant.com | TurboTenant | real-estate,saas
buildertrend.com | Buildertrend | enterprise,saas
fieldwire.com | Fieldwire | enterprise,saas
housecallpro.com | Housecall Pro | enterprise,saas
workiz.com | Workiz | enterprise,saas | IL
simprogroup.com | Simpro | enterprise,saas | AU
athenahealth.com | athenahealth | health,saas
drchrono.com | DrChrono | health,saas
tebra.com | Tebra | health,saas
nextgen.com | NextGen Healthcare | health,saas
eclinicalworks.com | eClinicalWorks | health,saas
practicefusion.com | Practice Fusion | health,saas
simplepractice.com | SimplePractice | health,saas
therapynotes.com | TherapyNotes | health,saas
jane.app | Jane | health,saas | CA
zocdoc.com | Zocdoc | health,consumer-tech
doximity.com | Doximity | health,saas
teladoc.com | Teladoc Health | health
amwell.com | Amwell | health
hims.com | Hims & Hers | health,consumer-tech
ro.co | Ro | health,consumer-tech
onemedical.com | One Medical | health
carbonhealth.com | Carbon Health | health
zochealth-placeholder.invalid | skip | skip
powerschool.com | PowerSchool | education,saas
nearpod.com | Nearpod | education,saas
seesaw.com | Seesaw | education,saas
udacity.com | Udacity | education
memrise.com | Memrise | education,consumer-tech | GB
busuu.com | Busuu | education,consumer-tech | GB
brainly.com | Brainly | education,consumer-tech | PL
course-hero-placeholder.invalid | skip | skip
gradescope.com | Gradescope | education,saas
turnitin.com | Turnitin | education,saas
proctorio.com | Proctorio | education,saas
examity.com | Examity | education,saas
zybooks.com | zyBooks | education,saas
mheducation.com | McGraw Hill | education,media
pearson.com | Pearson | education,media | GB
cengage.com | Cengage | education,media
wiley.com | Wiley | education,media
labster.com | Labster | education,saas | DK
codecombat.com | CodeCombat | education,developer
replit-placeholder.invalid | skip | skip
# ── gaming, esports, game infrastructure ──
activision.com | Activision | gaming
2k.com | 2K | gaming
take2games.com | Take-Two Interactive | gaming
bandainamcoent.com | Bandai Namco Entertainment | gaming | JP
capcom.com | Capcom | gaming | JP
sega.com | Sega | gaming | JP
konami.com | Konami | gaming | JP
guilded.gg | Guilded | gaming,consumer-tech
teamspeak.com | TeamSpeak | gaming,consumer-tech | DE
fanatical.com | Fanatical | gaming | GB
greenmangaming.com | Green Man Gaming | gaming | GB
hypixel.net | Hypixel | gaming
curseforge.com | CurseForge | gaming
modrinth.com | Modrinth | gaming,open-source
nexusmods.com | Nexus Mods | gaming | GB
playfab.com | PlayFab | gaming,developer
photonengine.com | Photon Engine | gaming,developer | DE
heroiclabs.com | Heroic Labs (Nakama) | gaming,developer | GB
accelbyte.io | AccelByte | gaming,developer
pragma.gg | Pragma | gaming,developer
wargaming.net | Wargaming | gaming | CY
gaijin.net | Gaijin Entertainment | gaming | HU
ncsoft.com | NCSoft | gaming | KR
netmarble.com | Netmarble | gaming | KR
pearlabyss.com | Pearl Abyss | gaming | KR
rovio.com | Rovio | gaming | FI
chess.com | Chess.com | gaming,consumer-tech
lichess.org | Lichess | gaming,open-source | FR
faceit.com | FACEIT | gaming,sports | GB
streamlabs.com | Streamlabs | gaming,media
streamelements.com | StreamElements | gaming,media | IL
restream.io | Restream | media,saas
overwolf.com | Overwolf | gaming,developer | IL
gamesight-placeholder.invalid | skip | skip
enad-placeholder.invalid | skip | skip
embracer.com | Embracer Group | gaming | SE
focus-entmt.com | Focus Entertainment | gaming | FR
devolverdigital.com | Devolver Digital | gaming
team17.com | Team17 | gaming | GB
kongregate.com | Kongregate | gaming
armorgames.com | Armor Games | gaming
poki.com | Poki | gaming | NL
crazygames.com | CrazyGames | gaming | BE
geforcenow-placeholder.invalid | skip | skip
shadow.tech | Shadow | gaming,cloud | FR
boosteroid.com | Boosteroid | gaming,cloud | UA
parsec.app | Parsec | gaming,developer
opgg-placeholder.invalid | skip | skip
gamersclub-placeholder.invalid | skip | skip
skinport-placeholder.invalid | skip | skip
pokemon.com | The Pokémon Company | gaming,entertainment | JP
pokemongolive.com | Pokémon GO (Niantic) | gaming
nianticlabs.com | Niantic | gaming
jagex.com | Jagex | gaming | GB
runescape.com | RuneScape | gaming | GB
warframe.com | Warframe (Digital Extremes) | gaming | CA
pathofexile.com | Path of Exile (Grinding Gear Games) | gaming | NZ
worldofwarships.com | World of Warships | gaming | CY
eveonline.com | EVE Online (CCP Games) | gaming | IS
finalfantasyxiv.com | Final Fantasy XIV | gaming | JP
robloxdev-placeholder.invalid | skip | skip
# ── streaming, music, media platforms, publishing ──
roku.com | Roku | media,consumer-tech
scribd.com | Scribd | media,consumer-tech
kobo.com | Kobo | media,consumer-tech | CA
tunein.com | TuneIn | media
iheart.com | iHeartMedia | media
siriusxm.com | SiriusXM | media
viki.com | Rakuten Viki | media | US
philo.com | Philo | media
sling.com | Sling TV | media
fubo.tv | Fubo | media
plutotv-placeholder.invalid | skip | skip
tubitv.com | Tubi | media
vudu-placeholder.invalid | skip | skip
curiositystream.com | CuriosityStream | media
mubi.com | MUBI | media | GB
britbox.com | BritBox | media | GB
hayu.com | hayu | media | GB
crave.ca | Crave | media | CA
cbc-gem-placeholder.invalid | skip | skip
stan.com.au | Stan | media | AU
binge.com.au | Binge | media | AU
hotstar.com | Disney+ Hotstar | media | IN
jiocinema-placeholder.invalid | skip | skip
zee5.com | ZEE5 | media | IN
sonyliv.com | SonyLIV | media | IN
iqiyi.com | iQIYI | media | CN
viu.com | Viu | media | HK
wetv-placeholder.invalid | skip | skip
napster.com | Napster | media
qobuz.com | Qobuz | media | FR
audiomack.com | Audiomack | media
mixcloud.com | Mixcloud | media | GB
bandlab.com | BandLab | media,consumer-tech | SG
splice.com | Splice | media,saas
distrokid.com | DistroKid | media,saas
tunecore.com | TuneCore | media,saas
cdbaby.com | CD Baby | media,saas
ditto-placeholder.invalid | skip | skip
epidemicsound.com | Epidemic Sound | media,saas | SE
artlist.io | Artlist | media,saas | IL
storyblocks.com | Storyblocks | media,saas
shutterstock.com | Shutterstock | media,saas
gettyimages.com | Getty Images | media
pond5-placeholder.invalid | skip | skip
envato.com | Envato | media,saas | AU
creativemarket.com | Creative Market | media,saas
issuu.com | Issuu | media,saas
readwise.io | Readwise | consumer-tech,saas
pocket-placeholder.invalid | skip | skip
feedly.com | Feedly | consumer-tech,saas
inoreader.com | Inoreader | consumer-tech,saas | BG
flipboard.com | Flipboard | media,consumer-tech
goodreads-placeholder.invalid | skip | skip
storygraph-placeholder.invalid | skip | skip
webtoons.com | WEBTOON | media | KR
tapas.io | Tapas | media
wattpad-placeholder.invalid | skip | skip
royalroad.com | Royal Road | media
archiveofourown.org | Archive of Our Own | media,ngo
# ── telecom, ISPs, connectivity, MVNO ──
xfinity.com | Xfinity (Comcast) | telecom
spectrum.com | Spectrum (Charter) | telecom
cox.com | Cox Communications | telecom
frontier.com | Frontier Communications | telecom
lumen.com | Lumen | telecom
windstream.com | Windstream | telecom
rogers.com | Rogers | telecom | CA
bell.ca | Bell Canada | telecom | CA
videotron.com | Vidéotron | telecom | CA
freedommobile.ca | Freedom Mobile | telecom | CA
fizz.ca | Fizz | telecom | CA
koodomobile.com | Koodo | telecom | CA
virginplus.ca | Virgin Plus | telecom | CA
sasktel.com | SaskTel | telecom | CA
eastlink.ca | Eastlink | telecom | CA
cogeco.ca | Cogeco | telecom | CA
teksavvy.com | TekSavvy | telecom | CA
ee.co.uk | EE | telecom | GB
o2.co.uk | O2 UK | telecom | GB
three.co.uk | Three UK | telecom | GB
sky.com | Sky | telecom,media | GB
virginmedia.com | Virgin Media | telecom | GB
talktalk.co.uk | TalkTalk | telecom | GB
giffgaff.com | giffgaff | telecom | GB
orange.com | Orange | telecom | FR
sfr.fr | SFR | telecom | FR
bouyguestelecom.fr | Bouygues Telecom | telecom | FR
free.fr | Free | telecom | FR
telekom.de | Deutsche Telekom | telecom | DE
o2online.de | O2 Germany | telecom | DE
vodafone.de | Vodafone Germany | telecom | DE
telefonica.com | Telefónica | telecom | ES
movistar.es | Movistar | telecom | ES
tim.it | TIM | telecom | IT
kpn.com | KPN | telecom | NL
proximus.be | Proximus | telecom | BE
swisscom.ch | Swisscom | telecom | CH
telia.se | Telia | telecom | SE
telenor.com | Telenor | telecom | "NO"
elisa.fi | Elisa | telecom | FI
telstra.com.au | Telstra | telecom | AU
optus.com.au | Optus | telecom | AU
spark.co.nz | Spark NZ | telecom | NZ
one.nz | One NZ | telecom | NZ
singtel.com | Singtel | telecom | SG
fi.google.com | Google Fi | telecom
mintmobile.com | Mint Mobile | telecom
visible.com | Visible | telecom
usmobile.com | US Mobile | telecom
boostmobile.com | Boost Mobile | telecom
cricketwireless.com | Cricket Wireless | telecom
viasat.com | Viasat | telecom,space
hughesnet.com | HughesNet | telecom,space
twilio-placeholder.invalid | skip | skip
grasshopper.com | Grasshopper | telecom,saas
ooma.com | Ooma | telecom
nextiva.com | Nextiva | telecom,saas
vonage-placeholder.invalid | skip | skip
justcall.io | JustCall | telecom,saas | IN
cloudtalk.io | CloudTalk | telecom,saas | SK
talkdesk.com | Talkdesk | telecom,saas
five9.com | Five9 | telecom,saas
nice.com | NICE | telecom,saas | IL
genesys.com | Genesys | telecom,saas
# ── travel, hospitality, mobility, events, ticketing ──
travelport.com | Travelport | travel,saas | GB
avis.com | Avis Budget Group | travel,transport
enterprise.com | Enterprise Mobility | travel,transport
sixt.com | Sixt | travel,transport | DE
cloudbeds.com | Cloudbeds | travel,saas
mews.com | Mews | travel,saas | NL
guesty.com | Guesty | travel,saas | IL
hostaway.com | Hostaway | travel,saas | FI
lodgify.com | Lodgify | travel,saas | ES
siteminder.com | SiteMinder | travel,saas | AU
eventbrite.com | Eventbrite | entertainment,saas
cvent.com | Cvent | enterprise,saas
bizzabo.com | Bizzabo | enterprise,saas | IL
splashthat.com | Splash | enterprise,saas
lu.ma | Luma | consumer-tech,saas
ticketmaster.com | Ticketmaster | entertainment
axs.com | AXS | entertainment
seatgeek.com | SeatGeek | entertainment
stubhub.com | StubHub | entertainment
dice.fm | DICE | entertainment | GB
ticketek.com.au | Ticketek | entertainment | AU
ticketleap-placeholder.invalid | skip | skip
universe-placeholder.invalid | skip | skip
tixr.com | Tixr | entertainment
showpass.com | Showpass | entertainment | CA
opentable.com | OpenTable | food,consumer-tech
resy.com | Resy | food,consumer-tech
sevenrooms.com | SevenRooms | food,saas
yelp.com | Yelp | consumer-tech,internet
tock.com | Tock | food,saas
toast-placeholder.invalid | skip | skip
touchbistro.com | TouchBistro | food,saas | CA
olo.com | Olo | food,saas
chownow.com | ChowNow | food,saas
mindbodyonline.com | Mindbody | health,saas
classpass.com | ClassPass | health,consumer-tech
glofox.com | Glofox | health,saas | IE
zenoti.com | Zenoti | health,saas
booksy.com | Booksy | consumer-tech,saas | PL
fresha.com | Fresha | consumer-tech,saas | GB
vagaro.com | Vagaro | consumer-tech,saas
# ── marketing, CRM, support, CX, analytics, ad tech ──
moengage.com | MoEngage | enterprise,saas | IN
clevertap.com | CleverTap | enterprise,saas | IN
airship.com | Airship | enterprise,saas
batch.com | Batch | enterprise,saas | FR
mparticle.com | mParticle | developer,saas
tealium.com | Tealium | enterprise,saas
rudderstack.com | RudderStack | developer,open-source,saas
snowplow.io | Snowplow | developer,open-source,saas | GB
criteo.com | Criteo | enterprise,media | FR
taboola.com | Taboola | enterprise,media | IL
outbrain.com | Outbrain | enterprise,media
pubmatic.com | PubMatic | enterprise,media
magnite.com | Magnite | enterprise,media
indexexchange.com | Index Exchange | enterprise,media | CA
openx.com | OpenX | enterprise,media
adroll.com | AdRoll | enterprise,saas
stackadapt.com | StackAdapt | enterprise,saas | CA
branch.io | Branch | developer,saas
appsflyer.com | AppsFlyer | developer,saas | IL
adjust.com | Adjust | developer,saas | DE
singular.net | Singular | developer,saas
kochava.com | Kochava | developer,saas
close.com | Close | enterprise,saas
copper.com | Copper | enterprise,saas
insightly.com | Insightly | enterprise,saas
nutshell.com | Nutshell | enterprise,saas
keap.com | Keap | enterprise,saas
ontraport.com | Ontraport | enterprise,saas
qualified.com | Qualified | enterprise,saas
kayako.com | Kayako | enterprise,saas
happyfox.com | HappyFox | enterprise,saas
gainsight.com | Gainsight | enterprise,saas
churnzero.com | ChurnZero | enterprise,saas
totango.com | Totango | enterprise,saas
vitally.io | Vitally | enterprise,saas
planhat.com | Planhat | enterprise,saas | SE
baremetrics.com | Baremetrics | finance,saas
chartmogul.com | ChartMogul | finance,saas | DE
alchemer.com | Alchemer | enterprise,saas
delighted.com | Delighted | enterprise,saas
medallia.com | Medallia | enterprise,saas
qualaroo.com | Qualaroo | enterprise,saas
bazaarvoice.com | Bazaarvoice | commerce,saas
trustpilot.com | Trustpilot | consumer-tech,internet | DK
reviews.io | REVIEWS.io | commerce,saas | GB
okendo.io | Okendo | commerce,saas | AU
judge.me | Judge.me | commerce,saas | GB
g2.com | G2 | enterprise,internet
agorapulse.com | Agorapulse | enterprise,saas | FR
sendible.com | Sendible | enterprise,saas | GB
socialbee.com | SocialBee | enterprise,saas | RO
planoly.com | Planoly | enterprise,saas
sketch.com | Sketch | developer,software | NL
invisionapp.com | InVision | developer,saas
zeplin.io | Zeplin | developer,saas
marvelapp.com | Marvel | developer,saas | GB
imagekit.io | ImageKit | developer,saas | IN
sirv.com | Sirv | developer,saas | GB
bynder.com | Bynder | enterprise,saas | NL
brandfolder.com | Brandfolder | enterprise,saas
canto.com | Canto | enterprise,saas
frontify.com | Frontify | enterprise,saas | CH
brightcove.com | Brightcove | media,saas
jwplayer.com | JW Player | media,saas
kaltura.com | Kaltura | media,saas
acquia.com | Acquia | developer,saas
higherlogic.com | Higher Logic | enterprise,saas
khoros.com | Khoros | enterprise,saas
on24.com | ON24 | enterprise,saas
livestorm.co | Livestorm | enterprise,saas | FR
demio.com | Demio | enterprise,saas
bigmarker.com | BigMarker | enterprise,saas
airmeet.com | Airmeet | enterprise,saas | IN
seismic.com | Seismic | enterprise,saas
highspot.com | Highspot | enterprise,saas
showpad.com | Showpad | enterprise,saas | BE
6sense.com | 6sense | enterprise,saas
demandbase.com | Demandbase | enterprise,saas
bombora.com | Bombora | enterprise,saas
cognism.com | Cognism | enterprise,saas | GB
chilipiper.com | Chili Piper | enterprise,saas
celonis.com | Celonis | enterprise,saas | DE
planful.com | Planful | finance,saas
venasolutions.com | Vena | finance,saas | CA
mosaic.tech | Mosaic | finance,saas
pigment.com | Pigment | finance,saas | FR
blackline.com | BlackLine | finance,saas
floqast.com | FloQast | finance,saas
sovos.com | Sovos | finance,saas
jaggaer.com | JAGGAER | enterprise,saas
ivalua.com | Ivalua | enterprise,saas | FR
ziphq.com | Zip | enterprise,saas
egnyte.com | Egnyte | enterprise,saas
citrix.com | Citrix | enterprise,software
uipath.com | UiPath | enterprise,ai,saas
automationanywhere.com | Automation Anywhere | enterprise,ai,saas
c3.ai | C3 AI | enterprise,ai,saas
palantir.com | Palantir | enterprise,ai
sap-placeholder.invalid | skip | skip
freshworks-placeholder.invalid | skip | skip
# ── IoT, hardware, consumer devices, smart home ──
particle.io | Particle | developer,infrastructure
balena.io | balena | developer,infrastructure | GB
losant.com | Losant | developer,infrastructure
blynk.io | Blynk | developer,infrastructure
thingsboard.io | ThingsBoard | developer,open-source
ubidots.com | Ubidots | developer,infrastructure
golioth.io | Golioth | developer,infrastructure
memfault.com | Memfault | developer,infrastructure
ring.com | Ring | consumer-tech
arlo.com | Arlo | consumer-tech
wyze.com | Wyze | consumer-tech
ecobee.com | ecobee | consumer-tech | CA
smartthings.com | SmartThings | consumer-tech
philips-hue.com | Philips Hue | consumer-tech | NL
tile.com | Tile | consumer-tech
eufy.com | eufy | consumer-tech
govee.com | Govee | consumer-tech | CN
tp-link.com | TP-Link | consumer-tech,telecom | CN
netgear.com | Netgear | consumer-tech,telecom
ubnt.com | Ubiquiti | telecom,infrastructure | | status.ui.com
eero.com | eero | consumer-tech,telecom
meraki.com | Cisco Meraki | telecom,infrastructure
tesla-placeholder.invalid | skip | skip
rivian-placeholder.invalid | skip | skip
chargepoint.com | ChargePoint | automotive,energy
evgo.com | EVgo | automotive,energy
electrifyamerica.com | Electrify America | automotive,energy
flo.com | FLO | automotive,energy | CA
enphase.com | Enphase | energy,consumer-tech
solaredge.com | SolarEdge | energy | IL
tado.com | tado° | energy,consumer-tech | DE
octopus.energy | Octopus Energy | energy | GB
ovoenergy.com | OVO Energy | energy | GB
bulb-placeholder.invalid | skip | skip
# ── weather, maps, geodata APIs ──
openweathermap.org | OpenWeather | weather,developer | GB
tomorrow.io | Tomorrow.io | weather,developer
weatherapi.com | WeatherAPI | weather,developer
here.com | HERE Technologies | transport,developer | NL
tomtom.com | TomTom | transport,developer | NL
maptiler.com | MapTiler | developer,saas | CH
radar.com | Radar | developer,saas
foursquare.com | Foursquare | developer,consumer-tech
what3words.com | what3words | developer | GB
smartystreets.com | Smarty | developer,saas
loqate.com | Loqate | developer,saas | GB
# ── government digital services, universities IT ──
cloud.gov | cloud.gov (GSA) | government,cloud | | cloudgov.statuspage.io
login.gov | Login.gov (GSA) | government,cyber
payments.service.gov.uk | GOV.UK Pay | government,payments | GB | status.payments.service.gov.uk
notifications.service.gov.uk | GOV.UK Notify | government | GB | status.notifications.service.gov.uk
canada.ca | Government of Canada digital services | government | CA | status.canada.ca
digital.canada.ca | Canadian Digital Service | government | CA | status.notification.canada.ca
service.nsw.gov.au | Service NSW | government | AU
digital.gov.au | Digital Transformation Agency | government | AU | status.digital.gov.au
mygov.au-placeholder.invalid | skip | skip
usds-placeholder.invalid | skip | skip
grants.gov | Grants.gov | government
sam.gov | SAM.gov | government
nasa-placeholder.invalid | skip | skip
huit.harvard.edu | Harvard University IT | education | | status.huit.harvard.edu
its.umich.edu | University of Michigan ITS | education | | status.its.umich.edu
it.cornell.edu | Cornell IT | education | | it.cornell.edu
princeton.edu | Princeton University OIT | education | | status.princeton.edu
yale.edu | Yale ITS | education | | status.yale.edu
columbia.edu | Columbia University IT | education | | status.cuit.columbia.edu
nyu.edu | NYU IT | education | | status.nyu.edu
upenn.edu | Penn ISC | education | | status.upenn.edu
cmu.edu | Carnegie Mellon Computing Services | education | | status.cmu.edu
ucla.edu | UCLA IT Services | education | | status.it.ucla.edu
ucsd.edu | UC San Diego ITS | education | | status.ucsd.edu
ucdavis.edu | UC Davis IET | education | | status.ucdavis.edu
uw.edu | University of Washington IT | education | | status.uw.edu
utoronto.ca | University of Toronto ITS | education | CA | status.utoronto.ca
ubc.ca | UBC IT | education | CA | status.it.ubc.ca
mcgill.ca | McGill IT | education | CA | status.mcgill.ca
uwaterloo.ca | University of Waterloo IST | education | CA | status.uwaterloo.ca
ualberta.ca | University of Alberta IST | education | CA | status.ualberta.ca
umontreal.ca | Université de Montréal TI | education | CA | statut.umontreal.ca
uqam.ca | UQAM | education | CA | statut.uqam.ca
ulaval.ca | Université Laval | education | CA | etat.ulaval.ca
concordia.ca | Concordia University | education | CA | status.concordia.ca
yorku.ca | York University UIT | education | CA | status.yorku.ca
uottawa.ca | University of Ottawa | education | CA | status.uottawa.ca
sfu.ca | Simon Fraser University IT | education | CA | status.sfu.ca
ucalgary.ca | University of Calgary IT | education | CA | status.ucalgary.ca
mun.ca | Memorial University | education | CA | status.mun.ca
dal.ca | Dalhousie University | education | CA | status.dal.ca
queensu.ca | Queen's University ITS | education | CA | status.queensu.ca
uwo.ca | Western University | education | CA | status.uwo.ca
mcmaster.ca | McMaster University | education | CA | status.mcmaster.ca
ox.ac.uk | University of Oxford IT Services | education | GB | status.it.ox.ac.uk
cam.ac.uk | University of Cambridge UIS | education | GB | status.uis.cam.ac.uk
imperial.ac.uk | Imperial College London ICT | education | GB | status.imperial.ac.uk
ucl.ac.uk | UCL ISD | education | GB | status.ucl.ac.uk
ed.ac.uk | University of Edinburgh IS | education | GB | status.is.ed.ac.uk
manchester.ac.uk | University of Manchester IT | education | GB | status.itservices.manchester.ac.uk
kcl.ac.uk | King's College London IT | education | GB | status.kcl.ac.uk
bristol.ac.uk | University of Bristol IT | education | GB | status.bristol.ac.uk
gatech.edu | Georgia Tech OIT | education | | status.gatech.edu
purdue.edu | Purdue IT | education | | status.purdue.edu
wisc.edu | UW–Madison DoIT | education | | status.doit.wisc.edu
illinois.edu | University of Illinois Technology Services | education | | status.illinois.edu
umn.edu | University of Minnesota OIT | education | | status.umn.edu
osu.edu | Ohio State OTDI | education | | status.osu.edu
psu.edu | Penn State IT | education | | status.psu.edu
utexas.edu | UT Austin ITS | education | | status.utexas.edu
tamu.edu | Texas A&M Technology Services | education | | status.tamu.edu
ufl.edu | University of Florida IT | education | | status.it.ufl.edu
asu.edu | Arizona State University UTO | education | | status.asu.edu
arizona.edu | University of Arizona UITS | education | | status.arizona.edu
colorado.edu | CU Boulder OIT | education | | status.colorado.edu
duke.edu | Duke OIT | education | | status.oit.duke.edu
unc.edu | UNC ITS | education | | status.unc.edu
virginia.edu | UVA ITS | education | | status.its.virginia.edu
northwestern.edu | Northwestern IT | education | | status.northwestern.edu
uchicago.edu | UChicago IT Services | education | | status.uchicago.edu
jhu.edu | Johns Hopkins IT | education | | status.it.johnshopkins.edu
bu.edu | Boston University IS&T | education | | status.bu.edu
northeastern.edu | Northeastern ITS | education | | status.northeastern.edu
rutgers.edu | Rutgers OIT | education | | status.rutgers.edu
umd.edu | University of Maryland DIT | education | | status.umd.edu
usc.edu | USC ITS | education | | status.usc.edu
uci.edu | UC Irvine OIT | education | | status.uci.edu
berkeley.edu | UC Berkeley IT | education | | status.berkeley.edu
indiana.edu | Indiana University UITS | education | | status.iu.edu
msu.edu | Michigan State IT | education | | status.msu.edu
unimelb.edu.au | University of Melbourne | education | AU | status.unimelb.edu.au
sydney.edu.au | University of Sydney ICT | education | AU | status.sydney.edu.au
monash.edu | Monash University eSolutions | education | AU | status.monash.edu
unsw.edu.au | UNSW IT | education | AU | status.unsw.edu.au
uq.edu.au | University of Queensland ITS | education | AU | status.uq.edu.au
auckland.ac.nz | University of Auckland | education | NZ | status.auckland.ac.nz
ethz.ch | ETH Zurich IT Services | education | CH | status.ethz.ch
epfl.ch | EPFL | education | CH | status.epfl.ch
tum.de | TU Munich IT | education | DE | status.tum.de
kuleuven.be | KU Leuven ICTS | education | BE | status.kuleuven.be
tudelft.nl | TU Delft | education | NL | status.tudelft.nl
uva.nl | University of Amsterdam | education | NL | status.uva.nl
nus.edu.sg | National University of Singapore | education | SG | status.nus.edu.sg
# ── misc SaaS long tail (Cloud 100 / YC) ──
airtable-placeholder.invalid | skip | skip
gong-placeholder.invalid | skip | skip
verkada.com | Verkada | cyber,infrastructure
samsara.com | Samsara | logistics,infrastructure
motive.com | Motive | logistics,infrastructure
geotab.com | Geotab | logistics,infrastructure | CA
flock-placeholder.invalid | skip | skip
scale-placeholder.invalid | skip | skip
faire-placeholder.invalid | skip | skip
whatnot.com | Whatnot | commerce,media
flexport-placeholder.invalid | skip | skip
convoy-placeholder.invalid | skip | skip
loadsmart.com | Loadsmart | logistics,saas
uberfreight-placeholder.invalid | skip | skip
gem.com | Gem | enterprise,labour,saas
rippling-placeholder.invalid | skip | skip
lattice-placeholder.invalid | skip | skip
vercel-placeholder.invalid | skip | skip
webflow-placeholder.invalid | skip | skip
grafana-placeholder.invalid | skip | skip
redis-placeholder.invalid | skip | skip
kong-placeholder.invalid | skip | skip
apollo-placeholder.invalid | skip | skip
cribl.io | Cribl | developer,saas
teleport-placeholder.invalid | skip | skip
tines.com | Tines | cyber,saas | IE
torq.io | Torq | cyber,saas | IL
swimlane.com | Swimlane | cyber,saas
opal.dev | Opal | cyber,saas
conductorone.com | ConductorOne | cyber,saas
veza.com | Veza | cyber,saas
oso-placeholder.invalid | skip | skip
cerbos.dev | Cerbos | developer,cyber | GB
permit.io | Permit.io | developer,cyber | IL
warrant-placeholder.invalid | skip | skip
descript-placeholder.invalid | skip | skip
runway-placeholder.invalid | skip | skip
gamma-placeholder.invalid | skip | skip
tome.app | Tome | enterprise,ai,saas
beautiful.ai | Beautiful.ai | enterprise,ai,saas
notion-placeholder.invalid | skip | skip
coda-placeholder.invalid | skip | skip
clay.com | Clay | enterprise,ai,saas
instantly.ai | Instantly | enterprise,saas
lemlist.com | lemlist | enterprise,saas | FR
smartlead.ai | Smartlead | enterprise,saas
reply.io | Reply.io | enterprise,saas
mailshake.com | Mailshake | enterprise,saas
woodpecker.co | Woodpecker | enterprise,saas | PL
hunter.io | Hunter | enterprise,saas | FR
snov.io | Snov.io | enterprise,saas | UA
seamless.ai | Seamless.AI | enterprise,saas
rocketreach.co | RocketReach | enterprise,saas
peopledatalabs.com | People Data Labs | enterprise,developer
fullcontact-placeholder.invalid | skip | skip
brandwatch-placeholder.invalid | skip | skip
mention.com | Mention | enterprise,saas | FR
awario.com | Awario | enterprise,saas
talkwalker.com | Talkwalker | enterprise,saas | LU
cision.com | Cision | enterprise,media
prnewswire.com | PR Newswire | media,enterprise
businesswire.com | Business Wire | media,enterprise
globenewswire.com | GlobeNewswire | media,enterprise
prweb.com | PRWeb | media,enterprise
muckrack.com | Muck Rack | media,saas
prezly.com | Prezly | media,saas | BE
harmonic-placeholder.invalid | skip | skip
crunchbase.com | Crunchbase | finance,internet
pitchbook.com | PitchBook | finance,saas
cbinsights.com | CB Insights | finance,saas
dealroom.co | Dealroom | finance,saas | NL
tegus-placeholder.invalid | skip | skip
alphasense.com | AlphaSense | finance,saas
kensho-placeholder.invalid | skip | skip
docsend-placeholder.invalid | skip | skip
carta-placeholder.invalid | skip | skip
vanta-placeholder.invalid | skip | skip
thoropass.com | Thoropass | cyber,saas
sprinto.com | Sprinto | cyber,saas | IN
scrut.io | Scrut Automation | cyber,saas | IN
hyperproof.com | Hyperproof | cyber,saas
auditboard.com | AuditBoard | enterprise,saas
workiva.com | Workiva | enterprise,saas
diligent.com | Diligent | enterprise,saas
boardable-placeholder.invalid | skip | skip
navex.com | NAVEX | enterprise,saas
logicgate.com | LogicGate | enterprise,saas
archer-placeholder.invalid | skip | skip
riskonnect-placeholder.invalid | skip | skip
resolver.com | Resolver | enterprise,saas | CA
ideagen.com | Ideagen | enterprise,saas | GB
intelex.com | Intelex | enterprise,saas | CA
cority.com | Cority | enterprise,saas | CA
benchling.com | Benchling | science,saas
dotmatics.com | Dotmatics | science,saas
labarchives.com | LabArchives | science,saas
scispot-placeholder.invalid | skip | skip
protocols.io | protocols.io | science,saas
biorender.com | BioRender | science,saas | CA
labguru-placeholder.invalid | skip | skip
quartzy.com | Quartzy | science,saas
opentrons.com | Opentrons | science
strateos-placeholder.invalid | skip | skip
emeraldcloudlab-placeholder.invalid | skip | skip
veeva.com | Veeva Systems | pharma,saas
medidata.com | Medidata | pharma,saas
iqvia.com | IQVIA | pharma,saas
castoredc.com | Castor | pharma,saas | NL
florence-placeholder.invalid | skip | skip
komodohealth.com | Komodo Health | health,saas
flatiron.com | Flatiron Health | health,saas
tempus.com | Tempus | health,ai
color.com | Color | health
23andme.com | 23andMe | health,consumer-tech
ancestry.com | Ancestry | consumer-tech
myheritage.com | MyHeritage | consumer-tech | IL
findagrave-placeholder.invalid | skip | skip
familysearch.org | FamilySearch | consumer-tech,ngo
# ── crypto exchanges & wallets (Canada / EU / Asia) & web3 infra ──
coinhako.com | Coinhako | crypto | SG
independentreserve.com | Independent Reserve | crypto | AU
btcmarkets.net | BTC Markets | crypto | AU
coinjar.com | CoinJar | crypto | AU
swyftx.com | Swyftx | crypto | AU
coinspot.com.au | CoinSpot | crypto | AU
bitvavo.com | Bitvavo | crypto | NL
bitstamp-placeholder.invalid | skip | skip
zonda-placeholder.invalid | skip | skip
upbit.com | Upbit | crypto | KR
bithumb.com | Bithumb | crypto | KR
bitflyer.com | bitFlyer | crypto | JP
coincheck.com | Coincheck | crypto | JP
wazirx.com | WazirX | crypto | IN
coindcx.com | CoinDCX | crypto | IN
coinswitch.co | CoinSwitch | crypto | IN
alchemypay-placeholder.invalid | skip | skip
moonpay.com | MoonPay | crypto,payments
ramp.network | Ramp Network | crypto,payments | PL
transak.com | Transak | crypto,payments
banxa.com | Banxa | crypto,payments | AU
simplex-placeholder.invalid | skip | skip
bitpay.com | BitPay | crypto,payments
coinpayments.net | CoinPayments | crypto,payments
nowpayments.io | NOWPayments | crypto,payments
blockdaemon.com | Blockdaemon | crypto,infrastructure
figment.io | Figment | crypto,infrastructure | CA
kiln.fi | Kiln | crypto,infrastructure | FR
p2p.org | P2P.org | crypto,infrastructure
stakefish-placeholder.invalid | skip | skip
allnodes.com | Allnodes | crypto,infrastructure
getblock.io | GetBlock | crypto,developer
nownodes.io | NOWNodes | crypto,developer
blockdaemon-placeholder.invalid | skip | skip
helius.dev | Helius | crypto,developer
triton-placeholder.invalid | skip | skip
alchemy-placeholder.invalid | skip | skip
tenderly.co | Tenderly | crypto,developer | RS
blocknative.com | Blocknative | crypto,developer
biconomy.io | Biconomy | crypto,developer
gelato.network | Gelato | crypto,developer | CH
pimlico.io | Pimlico | crypto,developer | GB
zerodev-placeholder.invalid | skip | skip
thirdweb.com | thirdweb | crypto,developer
crossmint.com | Crossmint | crypto,developer
reservoir.tools | Reservoir | crypto,developer
zora.co | Zora | crypto,internet
rarible.com | Rarible | crypto,internet
looksrare-placeholder.invalid | skip | skip
sound.xyz | Sound.xyz | crypto,media
mirror.xyz | Mirror | crypto,media
paragraph.xyz | Paragraph | crypto,media
farcaster.xyz | Farcaster | crypto,consumer-tech
lens.xyz | Lens Protocol | crypto,consumer-tech
dydx.exchange | dYdX | crypto
gmx.io | GMX | crypto
curve.fi | Curve | crypto
balancer.fi | Balancer | crypto
pancakeswap.finance | PancakeSwap | crypto
sushi.com | Sushi | crypto
1inch.io | 1inch | crypto
paraswap.io | ParaSwap | crypto
cowswap-placeholder.invalid | skip | skip
makerdao-placeholder.invalid | skip | skip
sky.money | Sky (MakerDAO) | crypto
frax.finance | Frax | crypto
rocketpool.net | Rocket Pool | crypto
eigenlayer.xyz | EigenLayer | crypto
across.to | Across | crypto
stargate.finance | Stargate | crypto
layerzero.network | LayerZero | crypto
wormhole.com | Wormhole | crypto
axelar.network | Axelar | crypto
hop.exchange | Hop Protocol | crypto
synapseprotocol.com | Synapse | crypto
safe.global | Safe | crypto,developer | DE
snapshot.org | Snapshot | crypto,developer
tally.xyz | Tally | crypto,developer
gitcoin.co | Gitcoin | crypto,open-source
optimism.io | Optimism | crypto
arbitrum.io | Arbitrum | crypto
base.org | Base | crypto
zksync.io | ZKsync | crypto
starknet.io | Starknet | crypto
scroll.io | Scroll | crypto
linea.build | Linea | crypto
mantle.xyz | Mantle | crypto
blast.io | Blast | crypto
polygon.technology | Polygon | crypto
avax.network | Avalanche | crypto
solana.com | Solana | crypto
near.org | NEAR | crypto
sui.io | Sui | crypto
aptosfoundation.org | Aptos | crypto
sei.io | Sei | crypto
celestia.org | Celestia | crypto
cosmos.network | Cosmos | crypto
polkadot.network | Polkadot | crypto | CH
cardano.org | Cardano | crypto
algorand.co | Algorand | crypto
tezos.com | Tezos | crypto | CH
hedera.com | Hedera | crypto
stellar.org | Stellar | crypto
ripple.com | Ripple | crypto
tron.network | TRON | crypto
tonkeeper-placeholder.invalid | skip | skip
ton.org | TON | crypto | CH
internetcomputer.org | Internet Computer (DFINITY) | crypto | CH
filecoin.io | Filecoin | crypto
arweave.org | Arweave | crypto
flow.com | Flow | crypto
immutable.com | Immutable | crypto,gaming | AU
ronin.network | Ronin | crypto,gaming
# ── consumer apps, social, dating, fitness, misc ──
bumble.com | Bumble | consumer-tech
hinge.co | Hinge | consumer-tech
match.com | Match | consumer-tech
grindr.com | Grindr | consumer-tech
nextdoor.com | Nextdoor | consumer-tech,internet
meetup.com | Meetup | consumer-tech,internet
strava-placeholder.invalid | skip | skip
fitbit-placeholder.invalid | skip | skip
trainerroad.com | TrainerRoad | health,consumer-tech
wahoofitness.com | Wahoo | health,consumer-tech
polar.com | Polar | health,consumer-tech | FI
suunto.com | Suunto | health,consumer-tech | FI
coros.com | COROS | health,consumer-tech
eightsleep.com | Eight Sleep | health,consumer-tech
tonal.com | Tonal | health,consumer-tech
hydrow.com | Hydrow | health,consumer-tech
ifit.com | iFIT | health,consumer-tech
lifesum.com | Lifesum | health,consumer-tech | SE
flo.health | Flo | health,consumer-tech | GB
clue-placeholder.invalid | skip | skip
betterhelp.com | BetterHelp | health,consumer-tech
talkspace.com | Talkspace | health,consumer-tech
cerebral.com | Cerebral | health,consumer-tech
lyrahealth.com | Lyra Health | health,saas
springhealth.com | Spring Health | health,saas
modernhealth.com | Modern Health | health,saas
ginger-placeholder.invalid | skip | skip
omadahealth.com | Omada Health | health
livongo-placeholder.invalid | skip | skip
goodrx.com | GoodRx | health,consumer-tech
capsule.com | Capsule | health,consumer-tech
altoids-placeholder.invalid | skip | skip
pillpack-placeholder.invalid | skip | skip
rover.com | Rover | consumer-tech
wag.co | Wag! | consumer-tech
thumbtack.com | Thumbtack | consumer-tech
angi.com | Angi | consumer-tech
taskrabbit.com | Taskrabbit | consumer-tech
care.com | Care.com | consumer-tech
handy-placeholder.invalid | skip | skip
houzz.com | Houzz | consumer-tech,real-estate
apartments.com | Apartments.com | real-estate
rentcafe-placeholder.invalid | skip | skip
zumper.com | Zumper | real-estate
opendoor.com | Opendoor | real-estate
offerpad.com | Offerpad | real-estate
compass.com | Compass | real-estate
realtor.com | Realtor.com | real-estate
rightmove.co.uk | Rightmove | real-estate | GB
zoopla.co.uk | Zoopla | real-estate | GB
realestate.com.au | realestate.com.au (REA) | real-estate | AU
domain.com.au | Domain | real-estate | AU
immobilienscout24.de | ImmoScout24 | real-estate | DE
seloger.com | SeLoger | real-estate | FR
idealista.com | idealista | real-estate | ES
funda.nl | Funda | real-estate | NL
carvana.com | Carvana | automotive,commerce
carmax.com | CarMax | automotive,commerce
vroom-placeholder.invalid | skip | skip
autotrader.com | Autotrader | automotive,commerce
cars.com | Cars.com | automotive,commerce
cargurus.com | CarGurus | automotive,commerce
copart.com | Copart | automotive,commerce
kbb.com | Kelley Blue Book | automotive,media
edmunds.com | Edmunds | automotive,media
indeed.com | Indeed | labour,internet
glassdoor.com | Glassdoor | labour,internet
ziprecruiter.com | ZipRecruiter | labour,internet
monster.com | Monster | labour,internet
dice.com | Dice | labour,internet
welcometothejungle.com | Welcome to the Jungle | labour,internet | FR
seek.com.au | SEEK | labour,internet | AU
reed.co.uk | Reed | labour,internet | GB
totaljobs.com | Totaljobs | labour,internet | GB
stepstone.de | StepStone | labour,internet | DE
naukri.com | Naukri | labour,internet | IN
upwork.com | Upwork | labour,internet
fiverr.com | Fiverr | labour,internet | IL
toptal.com | Toptal | labour,internet
freelancer.com | Freelancer.com | labour,internet | AU
peopleperhour.com | PeoplePerHour | labour,internet | GB
malt.com | Malt | labour,internet | FR
contra.com | Contra | labour,internet
wetransfer.com | WeTransfer | consumer-tech,saas | NL
fromsmash.com | Smash | consumer-tech,saas | FR
filemail.com | Filemail | consumer-tech,saas | "NO"
hightail.com | Hightail | enterprise,saas
send-anywhere.com | Send Anywhere | consumer-tech,saas | KR
visme.co | Visme | enterprise,saas
piktochart.com | Piktochart | enterprise,saas | MY
snappa.com | Snappa | enterprise,saas | CA
kapwing.com | Kapwing | media,saas
veed.io | VEED | media,saas | GB
invideo.io | InVideo | media,saas,ai | IN
animoto.com | Animoto | media,saas
lumen5.com | Lumen5 | media,saas | CA
pictory.ai | Pictory | media,saas,ai
opus.pro | OpusClip | media,saas,ai
capcut.com | CapCut | media,consumer-tech | SG
canva-placeholder.invalid | skip | skip
remove.bg | remove.bg | media,saas,ai | AT
photoroom.com | Photoroom | media,saas,ai | FR
cleanup-placeholder.invalid | skip | skip
picsart.com | Picsart | media,consumer-tech
befunky.com | BeFunky | media,consumer-tech
pixlr.com | Pixlr | media,consumer-tech | SG
fotor.com | Fotor | media,consumer-tech | CN
# ── enterprise infrastructure & IT (Statuspage-heavy) ──
solarwinds-placeholder.invalid | skip | skip
manageengine.com | ManageEngine | enterprise,software | IN
freshservice-placeholder.invalid | skip | skip
sysaid.com | SysAid | enterprise,saas | IL
ivanti.com | Ivanti | enterprise,cyber,software
bmc.com | BMC Software | enterprise,software
cherwell-placeholder.invalid | skip | skip
topdesk.com | TOPdesk | enterprise,saas | NL
halopsa.com | HaloPSA | enterprise,saas | GB
syncromsp.com | Syncro | enterprise,saas
pulseway.com | Pulseway | enterprise,saas | IE
n-able.com | N-able | enterprise,saas
auvik.com | Auvik | enterprise,saas | CA
domotz.com | Domotz | enterprise,saas | GB
liongard.com | Liongard | enterprise,saas
itglue.com | IT Glue | enterprise,saas | CA
hudu.com | Hudu | enterprise,saas
passportal-placeholder.invalid | skip | skip
scalepad.com | ScalePad | enterprise,saas | CA
cove-placeholder.invalid | skip | skip
axcient.com | Axcient | enterprise,saas
acronis.com | Acronis | enterprise,cyber,software | CH
arcserve.com | Arcserve | enterprise,software
commvault.com | Commvault | enterprise,software
rubrik.com | Rubrik | enterprise,cyber,software
cohesity.com | Cohesity | enterprise,software
zerto-placeholder.invalid | skip | skip
nakivo.com | NAKIVO | enterprise,software
msp360.com | MSP360 | enterprise,software
spanning.com | Spanning | enterprise,saas
afi.ai | Afi | enterprise,saas
keepit.com | Keepit | enterprise,saas | DK
avepoint.com | AvePoint | enterprise,saas
sharegate.com | ShareGate | enterprise,saas | CA
bittitan.com | BitTitan | enterprise,saas
cloudm.io | CloudM | enterprise,saas | GB
quest.com | Quest Software | enterprise,software
sailpoint.com | SailPoint | cyber,saas
saviynt.com | Saviynt | cyber,saas
oktane-placeholder.invalid | skip | skip
secret-double-octopus-placeholder.invalid | skip | skip
hypr.com | HYPR | cyber,saas
transmitsecurity.com | Transmit Security | cyber,saas | IL
silverfort.com | Silverfort | cyber,saas | IL
authomize-placeholder.invalid | skip | skip
appomni.com | AppOmni | cyber,saas
obsidiansecurity.com | Obsidian Security | cyber,saas
valence-placeholder.invalid | skip | skip
nudgesecurity.com | Nudge Security | cyber,saas
grip-placeholder.invalid | skip | skip
zluri.com | Zluri | enterprise,saas | IN
torii.ai | Torii | enterprise,saas | IL
productiv.com | Productiv | enterprise,saas
vendr.com | Vendr | enterprise,saas
tropic.app | Tropic | enterprise,saas
cledara.com | Cledara | enterprise,saas | GB
spendflo-placeholder.invalid | skip | skip
"""

STATUS_JSON_PATHS = ["/api/v2/summary.json", "/summary.json", "/api/v1/summary"]
CHALLENGE = re.compile(r"just a moment|cf-chl|attention required|access denied|captcha|enable javascript and cookies|verify you are human|ddos-guard|incapsula|_incapsula", re.I)


class Bucket:
    def __init__(self, rate: float) -> None:
        self.rate, self.tokens, self.t, self.lock = rate, 1.0, time.monotonic(), threading.Lock()

    def take(self) -> None:
        while True:
            with self.lock:
                now = time.monotonic()
                self.tokens = min(1.0 + self.rate, self.tokens + (now - self.t) * self.rate)
                self.t = now
                if self.tokens >= 1:
                    self.tokens -= 1
                    return
                wait = (1 - self.tokens) / self.rate
            time.sleep(wait)


BUCKET = Bucket(RATE)
_local = threading.local()
_dns_cache: dict[str, bool] = {}
_dns_lock = threading.Lock()


def session() -> requests.Session:
    s = getattr(_local, "s", None)
    if s is None:
        s = requests.Session()
        s.headers.update({"User-Agent": UA, "Accept": "application/json, text/html;q=0.9, */*;q=0.8", "Accept-Language": "en"})
        _local.s = s
    return s


def resolves(host: str) -> bool:
    with _dns_lock:
        if host in _dns_cache:
            return _dns_cache[host]
    try:
        socket.getaddrinfo(host, 443, proto=socket.IPPROTO_TCP)
        ok = True
    except OSError:
        ok = False
    with _dns_lock:
        _dns_cache[host] = ok
    return ok


def get(url: str) -> dict | None:
    host = urlparse(url).hostname or ""
    if not resolves(host):
        return None
    BUCKET.take()
    try:
        r = session().get(url, timeout=TIMEOUT, allow_redirects=True, stream=True)
        body = b""
        for chunk in r.iter_content(65536):
            body += chunk
            if len(body) > MAX_BODY:
                break
        r.close()
        text = body.decode(r.encoding or "utf-8", errors="replace") if body else ""
        return {"status": r.status_code, "ct": (r.headers.get("content-type") or "").lower(), "final": r.url, "text": text, "len": len(body)}
    except requests.RequestException:
        return None


def json_shape(text: str) -> tuple[str, dict] | None:
    t = text.lstrip(" \n\r\t")
    if not t.startswith("{"):
        return None
    try:
        j = json.loads(t)
    except ValueError:
        return None
    if not isinstance(j, dict):
        return None
    page = j.get("page")
    if isinstance(page, dict) and isinstance(j.get("status"), dict) and ("components" in j or "incidents" in j) and "id" in page:
        return "statuspage", j
    if isinstance(j.get("subpages"), list) and any(isinstance(p, dict) and "summary" in p for p in j["subpages"]):
        return "incidentio", j
    if isinstance(page, dict) and ("activeIncidents" in j or "activeMaintenances" in j or isinstance(page.get("status"), str)):
        return "instatus", j
    if "ongoing_incidents" in j or "page_title" in j:
        return "incidentio", j
    if isinstance(j.get("result"), dict) and "status_overall" in j["result"]:
        return "statusio", j
    return None


def parse_candidates() -> list[dict]:
    out, seen = [], set()
    for ln in CANDIDATES.strip().splitlines():
        ln = ln.strip()
        if not ln or ln.startswith("#"):
            continue
        parts = [p.strip() for p in ln.split("|")]
        domain = parts[0].lower()
        if domain.endswith(".invalid") or domain in seen or len(parts) < 3:
            continue
        seen.add(domain)
        country = (parts[3].strip('"') if len(parts) > 3 and parts[3] else "US")
        hosts = [h.strip() for h in parts[4].split(",") if h.strip()] if len(parts) > 4 else []
        out.append({"domain": domain, "name": parts[1], "categories": [c.strip() for c in parts[2].split(",")], "country": country, "hosts": hosts})
    return out


def load_registry() -> tuple[dict[str, str], set[str], set[str], set[str]]:
    """id→domain, all sensor URLs, hosts of existing status sensors, all ids."""
    id_domain: dict[str, str] = {}
    urls: set[str] = set()
    status_hosts: set[str] = set()
    ids: set[str] = set()
    for f in ["config/sources.yaml"] + sorted(glob.glob("config/sources.d/*.yaml")):
        if f.endswith(FRAGMENT):
            continue
        d = yaml.safe_load(open(f)) or {}
        for s in d.get("sources", []):
            ids.add(s["id"])
            if s.get("domain") and not s.get("extend"):
                id_domain[s["id"]] = s["domain"].lower()
            for sen in s.get("sensors", []) or []:
                u = sen.get("url", "")
                urls.add(u.rstrip("/"))
                if sen.get("type") == "STATUSPAGE" or sen.get("connector") in ("statuspage", "statusjson"):
                    h = urlparse(u).hostname
                    if h:
                        status_hosts.add(h)
    return id_domain, urls, status_hosts, ids


def apex(host: str) -> str:
    parts = host.split(".")
    if len(parts) >= 3 and parts[-2] in ("co", "com", "org", "gov", "ac", "net", "edu") and len(parts[-1]) == 2:
        return ".".join(parts[-3:])
    return ".".join(parts[-2:])


def brand(domain: str) -> str:
    return apex(domain).split(".")[0]


def probe_domain(c: dict, existing_urls: set[str], status_hosts: set[str]) -> dict:
    d = c["domain"]
    hits: list[dict] = []
    notes: list[str] = []
    probed_hosts: set[str] = set()

    def try_status_host(host: str, label: str) -> bool:
        if host in probed_hosts:
            return False
        if host in status_hosts:
            notes.append(f"{host}: already covered")
            return True
        probed_hosts.add(host)
        if not resolves(host):
            return False
        for p in STATUS_JSON_PATHS:
            url = f"https://{host}{p}"
            if url in existing_urls:
                return True
            r = get(url)
            if not r:
                continue
            if r["status"] == 200:
                sh = json_shape(r["text"])
                if sh:
                    flavor, j = sh
                    title = (j.get("page") or {}).get("name") if isinstance(j.get("page"), dict) else j.get("page_title", "")
                    hits.append({"kind": "status", "flavor": flavor, "url": url, "final": r["final"], "title": title or "", "via": label})
                    return True
        r = get(f"https://{host}/")
        if r and r["status"] == 200:
            body = r["text"]
            m = re.search(r"statusio_page_id['\"]?\s*[:=]\s*['\"]([0-9a-f]{20,})", body) or re.search(r"api\.status\.io/1\.0/status/([0-9a-f]{20,})", body)
            if m:
                url = f"https://api.status.io/1.0/status/{m.group(1)}"
                if url in existing_urls:
                    return True
                rr = get(url)
                if rr and rr["status"] == 200 and json_shape(rr["text"]):
                    hits.append({"kind": "status", "flavor": "statusio", "url": url, "final": rr["final"], "title": "", "via": label})
                    return True
            # Statuspage / Instatus embedded under another host (e.g. custom domain proxying)
            m = re.search(r"https://([a-z0-9.-]+)/api/v2/summary\.json", body)
            if m and m.group(1) not in probed_hosts:
                return try_status_host(m.group(1), f"embedded in {host}")
            fh = urlparse(r["final"]).hostname or host
            if fh != host and fh not in probed_hosts:
                return try_status_host(fh, f"redirect from {host}")
            prov = "hund" if "hund.io" in body else "sorry" if "sorryapp" in body else "firehydrant" if "firehydrant" in body else "betterstack" if ("betteruptime" in body or "betterstack" in body) else "freshstatus" if "freshstatus" in body else "statuspage-html" if "statuspage.io" in body else "custom"
            if CHALLENGE.search(body[:6000]):
                prov = "challenge"
            notes.append(f"{host}: {prov} status page, no machine-readable endpoint")
        elif r and r["status"] in (403, 429, 503):
            notes.append(f"{host}: http {r['status']}")
        return False

    found = False
    # 0. explicit hosts
    for h in c.get("hosts", []):
        if try_status_host(h, "explicit host"):
            found = True
            break
    # 1. status.<apex>
    if not found:
        sh = f"status.{apex(d)}"
        if sh in status_hosts:
            notes.append(f"{sh}: already covered")
            found = True
        elif resolves(sh):
            found = try_status_host(sh, "status host")
    # 2. <domain>/status
    if not found:
        r = get(f"https://{d}/status") or get(f"https://www.{d}/status")
        if r and r["status"] == 200:
            fh = urlparse(r["final"]).hostname or d
            if fh not in (d, f"www.{d}") and fh not in probed_hosts:
                found = try_status_host(fh, f"redirect from {d}/status")
            elif "text/html" in r["ct"]:
                m = re.search(r"https://([a-z0-9.-]+)/api/v2/", r["text"])
                if m and m.group(1) not in probed_hosts:
                    found = try_status_host(m.group(1), f"embedded in {d}/status")
    # 3. <brand>.statuspage.io / <brand>.instatus.com
    if not found:
        b = brand(d)
        for h in (f"{b}.statuspage.io", f"{b}.instatus.com"):
            if try_status_host(h, "brand host"):
                found = True
                break
    return {"domain": d, "hits": hits, "notes": notes}


def cmd_probe(args: argparse.Namespace) -> None:
    id_domain, urls, status_hosts, _ids = load_registry()
    cands = parse_candidates()
    if args.only:
        only = set(args.only.split(","))
        cands = [c for c in cands if c["domain"] in only]
    done: set[str] = set()
    if os.path.exists(args.out) and not args.fresh:
        for ln in open(args.out):
            try:
                done.add(json.loads(ln)["domain"])
            except ValueError:
                pass
    todo = [c for c in cands if c["domain"] not in done]
    print(f"{len(cands)} candidates, {len(done)} already probed, {len(todo)} to go · ≤ {RATE} req/s", file=sys.stderr)
    lock = threading.Lock()
    t0 = time.time()
    with open(args.out, "a") as out, ThreadPoolExecutor(max_workers=args.workers) as ex:
        futs = {ex.submit(probe_domain, c, urls, status_hosts): c for c in todo}
        for i, fut in enumerate(as_completed(futs), 1):
            c = futs[fut]
            try:
                res = fut.result()
            except Exception as e:  # noqa: BLE001
                res = {"domain": c["domain"], "hits": [], "notes": [f"probe error: {e}"]}
            res.update({"name": c["name"], "categories": c["categories"], "country": c["country"]})
            with lock:
                out.write(json.dumps(res) + "\n")
                out.flush()
            kinds = ",".join(sorted({h["kind"] + (":" + h["flavor"] if h.get("flavor") else "") for h in res["hits"]})) or "-"
            print(f"[{i}/{len(todo)} {time.time() - t0:5.0f}s] {c['domain']:28s} {kinds}", file=sys.stderr)


def slug(s: str) -> str:
    s = s.lower().replace("&", " and ")
    s = re.sub(r"\(.*?\)", "", s)
    s = re.sub(r"[^a-z0-9]+", "-", s).strip("-")
    return s


# Same organization declared earlier under another domain → extend instead of redeclaring.
EXTEND_OVERRIDE: dict[str, str] = {}
# Natural slug already taken by a different organization.
NEW_ID_OVERRIDE: dict[str, str] = {"notifications.service.gov.uk": "govuk-notify", "payments.service.gov.uk": "govuk-pay", "ubnt.com": "ubiquiti", "sky.money": "sky-money", "flow.com": "flow-blockchain", "base.org": "base-chain", "sei.io": "sei-network",
                                   "zip.co": "zip-au", "ziphq.com": "zip-procurement", "tally.so": "tally-forms", "tally.xyz": "tally-governance", "radar.com": "radar-geofencing",
                                   "flo.com": "flo-ev", "flo.health": "flo-health", "color.com": "color-health", "column.com": "column-bank", "increase.com": "increase-bank",
                                   "float.com": "float-scheduling", "hive.com": "hive-pm", "current.com": "current-bank", "dave.com": "dave-app", "route.com": "route-app",
                                   "copper.com": "copper-crm", "copper.co": "copper-custody", "close.com": "close-crm", "grain.com": "grain-ai", "mention.com": "mention-app",
                                   "clay.com": "clay-ai", "gem.com": "gem-recruiting", "one.nz": "one-nz", "free.fr": "free-mobile", "three.co.uk": "three-uk",
                                   "o2.co.uk": "o2-uk", "ee.co.uk": "ee-uk", "sky.com": "sky-uk", "orange.com": "orange-telecom", "tim.it": "tim-italia", "match.com": "match-group"}


# `<brand>.statuspage.io` / `<brand>.instatus.com` hits whose page title is generic or belongs to another company
# (checked by hand on 2026-09-13): nx → "Nexo Systems", scylladb → "TestApp", phantom → "PhantomPay", workramp →
# "Confirm LearnUp", copper / play / you / ally / dave / stripe / rainbow / route / fabric / rocket / convert / continue /
# unified / ssl / launchpad / tally / hey / runn / sr / quillbot / leonardo / orca / jane → ambiguous one-word titles.
DROP_BRAND_HOSTS = {"nx", "scylladb", "phantom", "workramp", "copper", "play", "you", "ally", "dave", "stripe", "rainbow", "route", "fabric",
                    "rocket", "convert", "continue", "unified", "ssl", "launchpad", "tally", "hey", "runn", "sr", "quillbot", "leonardo", "orca", "jane"}


def cmd_gate(args: argparse.Namespace) -> None:
    rows = [json.loads(ln) for ln in open(args.inp) if ln.strip()]
    targets = [h["url"] for r in rows for h in r["hits"] if re.search(r"\.(statuspage\.io|instatus\.com)/", h["url"])]
    drop: list[str] = []
    for u in targets:
        r = get(u)
        if not r or r["status"] != 200:
            drop.append(u)
            continue
        try:
            j = json.loads(r["text"])
        except ValueError:
            drop.append(u)
            continue
        pg = j.get("page") or {}
        comps = len(j.get("components") or [])
        upd = str(pg.get("updated_at") or "")
        legacy = "old" in (pg.get("url") or "").split("//")[-1].split("/")[0].split(".")[0]
        if comps == 0 or (upd[:4].isdigit() and upd[:10] < "2024-09-01") or legacy:
            drop.append(u)
            print(f"drop {u}  components={comps} updated_at={upd[:10]} url={pg.get('url', '')}", file=sys.stderr)
    json.dump(drop, open(args.drop, "w"))
    print(f"{len(drop)} of {len(targets)} brand-host hits dropped → {args.drop}", file=sys.stderr)


def cmd_emit(args: argparse.Namespace) -> None:
    id_domain, urls, status_hosts, ids = load_registry()
    gated: set[str] = set(json.load(open(args.drop))) if args.drop and os.path.exists(args.drop) else set()
    domain_id = dict(EXTEND_OVERRIDE)
    for i, dom in id_domain.items():
        domain_id.setdefault(dom, i)
        domain_id.setdefault(apex(dom), i)
    rows = [json.loads(ln) for ln in open(args.inp) if ln.strip()]
    rows.sort(key=lambda r: r["domain"])
    used_urls: set[str] = set()
    new_ids: set[str] = set()
    entries: dict[str, tuple[list[str], list[str]]] = {}
    n_status = 0
    for r in rows:
        if not r["hits"]:
            continue
        d = r["domain"]
        existing = domain_id.get(d) or domain_id.get(f"www.{d}") or (domain_id.get(apex(d)) if apex(d) not in ("service.gov.uk", "gov.uk", "gc.ca", "canada.ca") else None)
        sensors: list[str] = []
        for h in r["hits"]:
            u = h["url"].rstrip("/")
            if u in urls or u in used_urls or h["final"].rstrip("/") in urls:
                continue
            if urlparse(u).hostname in status_hosts:
                continue
            if h.get("via") == "brand host" and (urlparse(u).hostname or "").split(".")[0] in DROP_BRAND_HOSTS:
                continue
            if h["url"] in gated:
                continue
            used_urls.add(u)
            if h["kind"] == "status":
                n_status += 1
                if h["flavor"] == "statuspage":
                    sensors.append(f'      - {{ name: status, url: "{h["url"]}", type: STATUSPAGE, connector: statuspage, tier: S }}')
                else:
                    label = {"instatus": "instatus", "incidentio": "incident.io", "statusio": "status.io"}[h["flavor"]]
                    sensors.append(f'      - {{ name: status ({label}), url: "{h["url"]}", type: STATUSPAGE, connector: statusjson, tier: S }}')
        if not sensors:
            continue
        seen_names: dict[str, int] = {}
        fixed = []
        for s in sensors:
            m = re.match(r"      - \{ name: ([^,]+),", s)
            nm = m.group(1)
            if nm in seen_names:
                seen_names[nm] += 1
                s = s.replace(f"name: {nm},", f"name: {nm} {seen_names[nm]},", 1)
            else:
                seen_names[nm] = 1
            fixed.append(s)
        if existing:
            head = [f"  - id: {existing}", "    extend: true"]
            if r.get("country"):
                cc = r["country"]
                head.append(f'    country: {json.dumps(cc) if cc in ("NO",) else cc}')
            head.append("    sensors:")
            key = existing
        else:
            sid = NEW_ID_OVERRIDE.get(d) or slug(r["name"]) or slug(d.split(".")[0])
            if sid in ids or sid in new_ids:
                sid = slug(d.replace(".", "-"))
            if sid in ids or sid in new_ids:
                sid = f"{sid}-status"
            new_ids.add(sid)
            cats = ", ".join(r["categories"])
            cc = r["country"]
            head = [f"  - id: {sid}", f"    name: {r['name']}", f"    domain: {d}", f"    categories: [{cats}]", "    tier: S",
                    f"    country: {json.dumps(cc) if cc in ('NO',) else cc}", "    discover: { rss: true, status: true }", "    sensors:"]
            key = sid
        if key in entries:
            entries[key][1].extend(fixed)
        else:
            entries[key] = (head, list(fixed))
    print(f"# config/sources.d/{FRAGMENT} — HEADER TO BE WRITTEN BY HAND ({n_status} status sensors, {len(entries)} organizations)")
    print("sources:")
    for head, sens in entries.values():
        for ln in head:
            print(ln)
        for s in sens:
            print(s)
    blocked = [n for r in rows for n in r["notes"] if "already covered" not in n]
    print("# ── probe notes (not sensors) ──", file=sys.stderr)
    for n in sorted(set(blocked)):
        print(f"#   {n}", file=sys.stderr)


def main() -> None:
    ap = argparse.ArgumentParser()
    sub = ap.add_subparsers(dest="cmd", required=True)
    p = sub.add_parser("probe")
    p.add_argument("--out", default="/tmp/probe58.jsonl")
    p.add_argument("--only")
    p.add_argument("--fresh", action="store_true")
    p.add_argument("--workers", type=int, default=12)
    p.set_defaults(fn=cmd_probe)
    g = sub.add_parser("gate")
    g.add_argument("--in", dest="inp", default="/tmp/probe58.jsonl")
    g.add_argument("--drop", default="/tmp/probe58-drop.json")
    g.set_defaults(fn=cmd_gate)
    e = sub.add_parser("emit")
    e.add_argument("--in", dest="inp", default="/tmp/probe58.jsonl")
    e.add_argument("--drop", default="/tmp/probe58-drop.json")
    e.set_defaults(fn=cmd_emit)
    a = ap.parse_args()
    a.fn(a)


if __name__ == "__main__":
    main()