# config/sources.d/22-housing-labour-opendata.yaml — housing & real estate, labour & workforce, open-data portals
# (CKAN / Socrata / Opendatasoft / data.gouv.fr / data.europa.eu catalog APIs via jsonlist, llm: false) and national
# statistics offices not yet covered by the founding file. Written 2026-09-08; every sensor validated with
# apps/engine/src/validate.ts (OK only — no empty feeds, no thin pages).
#
# Not listed because blocked by design (bot management / WAF / no server-rendered content) — 2026-09-08:
#   housing: MBA (Cloudflare), Zoopla (Cloudflare), Domain.com.au (Akamai), SeLoger (Datadome), Notaires de France
#     (interstitial), S&P Dow Jones Indices RSS (403), ICE Mortgage Technology / ir.theice.com (Cloudflare),
#     Blackstone, RioCan, Cadillac Fairview, Colliers, CBRE (Cloudflare), Minto (403), ACQ (Cloudflare),
#     SHQ habitation.gouv.qc.ca (403), OREA (every path answers a soft 404), REBGV/gvrealtors.ca (Next.js shell, 403 on
#     reports), Realtor.ca (client-rendered), Halifax HPI (error page), Nationwide HPI microsite (redirects to
#     nationwide.co.uk — covered there), Brookfield Properties / Boardwalk / Simon / Vornado / Public Storage /
#     Ivanhoé Cambridge / Cushman & Wakefield / Sotheby's Canada (no feed, no server-rendered news index),
#     Allied / Canderel / Dream (WordPress feeds present but empty), Offerpad (articles index is a 79-char shell).
#   labour: OSHA (Akamai 403 on every path), Glassdoor Economic Research (403), AFL-CIO (rss.xml has no items,
#     /press 403), SEIU (feeds empty), LinkedIn Economic Graph (research/blog indexes are thin JS shells), Lightcast
#     (client-rendered blog), Ontario Ministry of Labour (news.ontario.ca RSS URLs return a JS shell), Commission des
#     normes du travail (merged into CNESST), ILO (international agents file).
#   open data: catalog.data.gov CKAN API answers 404 "Not Found" on every action (API gateway) — nothing to poll;
#     Open Data Charter feed is empty.
#   statistics: IBGE (Cloudflare), BPS Indonesia (Cloudflare), INEGI (press room 403, noticias.xml empty), INDEC
#     (novedades page is a thin shell), Statistics Norway RSS feeds (statistikk/nyheter) carry zero items,
#     Statistics Finland (client-rendered), Swiss FSO (client-rendered, no feed), DANE Colombia (Joomla shell),
#     e-Stat Japan (no feed; Statistics Bureau stat.go.jp covered instead), MOSPI India (rss.xml is an HTML page,
#     press-release index is a 1 KB shell), Statistics Netherlands press page 404 (news sitemap used instead).
sources:
  # ───────────────────────── Housing & real estate — Canada ─────────────────────────
  - id: cmhc
    name: Canada Mortgage and Housing Corporation
    domain: cmhc-schl.gc.ca
    categories: [real-estate, government]
    tier: B
    weight: 1.3
    aliases: [cmhc, schl, société canadienne d'hypothèques et de logement]
    products:
      - { name: Housing Market Outlook, type: product }
      - { name: Rental Market Report, type: product }
    discover: { rss: true, sitemap: true, pages: true }
    notes: "No RSS; the Housing Observer index is server-rendered and lists new reports and press releases."
    sensors:
      - { name: housing observer, url: "https://www.cmhc-schl.gc.ca/observer", type: HTML, connector: http, tier: B }
  - id: crea
    name: Canadian Real Estate Association
    domain: crea.ca
    categories: [real-estate]
    tier: B
    weight: 1.2
    aliases: [crea, aci, mls hpi]
    products:
      - { name: MLS Home Price Index, type: index, aliases: [mls hpi] }
      - { name: REALTOR.ca, type: service, aliases: [realtor.ca] }
    discover: { rss: true, sitemap: true }
    notes: "realtor.ca listings are client-rendered (React) — not monitorable without a browser; CREA's WordPress feed carries stats releases and news."
    sensors:
      - { name: news feed, url: "https://www.crea.ca/feed/realtime/", type: RSS, connector: rss, tier: B }
  - id: apciq
    name: Association professionnelle des courtiers immobiliers du Québec
    domain: apciq.ca
    categories: [real-estate]
    tier: B
    aliases: [apciq, qpareb, centris, quebec professional association of real estate brokers]
    products:
      - { name: Centris, type: service }
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: actualités (fr), url: "https://apciq.ca/feed/", type: RSS, connector: rss, tier: B }
      - { name: news (en), url: "https://apciq.ca/en/feed/", type: RSS, connector: rss, tier: B }
  - id: trreb
    name: Toronto Regional Real Estate Board
    domain: trreb.ca
    categories: [real-estate]
    tier: B
    aliases: [trreb, treb, toronto real estate board]
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: news feed, url: "https://trreb.ca/feed/", type: RSS, connector: rss, tier: B }
  - id: tal-quebec
    name: Tribunal administratif du logement
    domain: tal.gouv.qc.ca
    categories: [real-estate, government, legal]
    tier: C
    aliases: [tal, tribunal administratif du logement, régie du logement]
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: actualités, url: "https://www.tal.gouv.qc.ca/fr/actualites/", type: HTML, connector: http, tier: C }
  - id: rbq
    name: Régie du bâtiment du Québec
    domain: rbq.gouv.qc.ca
    categories: [real-estate, government]
    tier: C
    aliases: [rbq, régie du bâtiment]
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: salle de presse, url: "https://www.rbq.gouv.qc.ca/salle-de-presse/", type: HTML, connector: http, tier: C }
  - id: chba
    name: Canadian Home Builders' Association
    domain: chba.ca
    categories: [real-estate]
    tier: C
    aliases: [chba]
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: news feed, url: "https://www.chba.ca/feed/", type: RSS, connector: rss, tier: C }
  - id: apchq
    name: Association des professionnels de la construction et de l'habitation du Québec
    domain: apchq.com
    categories: [real-estate]
    tier: C
    aliases: [apchq]
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: sitemap (fr), url: "https://www.apchq.com/__sitemap__/fr.xml", type: SITEMAP, connector: sitemap, tier: C, config: { maxUrls: 2000 } }
  - id: remax
    name: RE/MAX
    domain: remax.com
    categories: [real-estate]
    tier: C
    aliases: [re/max, remax canada]
    discover: { rss: true }
    sensors:
      - { name: remax canada blog feed, url: "https://blog.remax.ca/feed/", type: RSS, connector: rss, tier: C }
      - { name: remax news feed, url: "https://news.remax.com/feed", type: RSS, connector: rss, tier: C }
  - id: royal-lepage
    name: Royal LePage
    domain: royallepage.ca
    categories: [real-estate]
    tier: C
    aliases: [royal lepage]
    products:
      - { name: Royal LePage House Price Survey, type: product }
    discover: { rss: true, sitemap: false, pages: true }
    notes: "Sitemap index only points to gzipped listing/agent maps (404); the news index is server-rendered."
    sensors:
      - { name: news, url: "https://www.royallepage.ca/en/realestate/news/", type: HTML, connector: http, tier: C }
  - id: capreit
    name: Canadian Apartment Properties REIT
    domain: capreit.ca
    categories: [real-estate, finance]
    tier: C
    aliases: [capreit, cap reit]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://www.capreit.ca/feed/", type: RSS, connector: rss, tier: C }
  - id: first-capital
    name: First Capital REIT
    domain: fcr.ca
    categories: [real-estate, finance]
    tier: C
    aliases: [first capital, fcr]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://fcr.ca/feed/", type: RSS, connector: rss, tier: C }
  - id: smartcentres
    name: SmartCentres REIT
    domain: smartcentres.com
    categories: [real-estate, finance]
    tier: C
    aliases: [smartcentres, smart centres]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://smartcentres.com/feed/", type: RSS, connector: rss, tier: C }
  - id: choice-properties
    name: Choice Properties REIT
    domain: choicereit.ca
    categories: [real-estate, finance]
    tier: C
    aliases: [choice properties, choice reit]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://www.choicereit.ca/feed/", type: RSS, connector: rss, tier: C }
  - id: cominar
    name: Cominar
    domain: cominar.com
    categories: [real-estate, finance]
    tier: C
    aliases: [cominar]
    discover: { rss: false, sitemap: true, pages: true }
    sensors:
      - { name: news, url: "https://www.cominar.com/en/news/", type: HTML, connector: http, tier: C }
  - id: oxford-properties
    name: Oxford Properties
    domain: oxfordproperties.com
    categories: [real-estate, finance]
    tier: C
    aliases: [oxford properties, omers]
    discover: { rss: false, sitemap: true }
    sensors:
      - { name: sitemap, url: "https://www.oxfordproperties.com/sitemap.xml", type: SITEMAP, connector: sitemap, tier: C, config: { maxUrls: 2000 } }

  # ───────────────────────── Housing & real estate — United States ─────────────────────────
  - id: zillow
    name: Zillow
    domain: zillow.com
    categories: [real-estate, technology]
    tier: B
    weight: 1.2
    aliases: [zillow group, zillow research]
    products:
      - { name: Zillow Home Value Index, type: index, aliases: [zhvi] }
    discover: { rss: false, sitemap: false }
    notes: "zillow.com (incl. /research/feed/) is behind bot management (403); the press room on mediaroom.com is open."
    sensors:
      - { name: press releases feed, url: "https://zillow.mediaroom.com/press-releases?pagetemplate=rss", type: RSS, connector: rss, tier: B }
  - id: redfin
    name: Redfin
    domain: redfin.com
    categories: [real-estate, technology]
    tier: B
    aliases: [redfin news]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://www.redfin.com/news/feed/", type: RSS, connector: rss, tier: B }
  - id: realtor-com
    name: Realtor.com
    domain: realtor.com
    categories: [real-estate, technology]
    tier: B
    aliases: [realtor.com research, move inc]
    discover: { rss: true }
    sensors:
      - { name: research feed, url: "https://www.realtor.com/research/feed/", type: RSS, connector: rss, tier: B }
  - id: nar
    name: National Association of REALTORS
    domain: nar.realtor
    categories: [real-estate]
    tier: B
    weight: 1.2
    aliases: [nar, national association of realtors, existing-home sales]
    products:
      - { name: Existing-Home Sales, type: index }
      - { name: Pending Home Sales Index, type: index }
    discover: { rss: false, sitemap: true, pages: true }
    notes: "The /rss-feeds page is a Next.js shell with no feed links; the press-release and Economists' Outlook indexes are server-rendered."
    sensors:
      - { name: press releases, url: "https://www.nar.realtor/press-releases", type: HTML, connector: http, tier: B }
      - { name: economists outlook, url: "https://www.nar.realtor/news/economists-outlook", type: HTML, connector: http, tier: C }
  - id: nahb
    name: National Association of Home Builders
    domain: nahb.org
    categories: [real-estate]
    tier: B
    aliases: [nahb, housing market index, hmi]
    products:
      - { name: Housing Market Index, type: index, aliases: [nahb hmi] }
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: blog feed, url: "https://www.nahb.org/blog/feed", type: RSS, connector: rss, tier: B }
      - { name: news sitemap, url: "https://www.nahb.org/news-sitemap.xml", type: SITEMAP, connector: sitemap, tier: B, config: { maxUrls: 5000 } }
  - id: attom
    name: ATTOM Data Solutions
    domain: attomdata.com
    categories: [real-estate]
    tier: C
    aliases: [attom, attom data]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://www.attomdata.com/feed/", type: RSS, connector: rss, tier: C }
  - id: freddie-mac
    name: Freddie Mac
    domain: freddiemac.com
    categories: [real-estate, finance]
    tier: B
    weight: 1.2
    aliases: [freddie mac, fhlmc, pmms]
    products:
      - { name: Primary Mortgage Market Survey, type: index, aliases: [pmms, mortgage rates survey] }
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: pmms mortgage rates, url: "https://www.freddiemac.com/pmms", type: HTML, connector: http, tier: B }
      - { name: my home blog feed, url: "https://myhome.freddiemac.com/blog/feed", type: RSS, connector: rss, tier: C }
  - id: fannie-mae
    name: Fannie Mae
    domain: fanniemae.com
    categories: [real-estate, finance]
    tier: B
    weight: 1.2
    aliases: [fannie mae, fnma]
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: newsroom, url: "https://www.fanniemae.com/newsroom", type: HTML, connector: http, tier: B }
  - id: hud
    name: U.S. Department of Housing and Urban Development
    domain: hud.gov
    categories: [real-estate, government]
    tier: B
    weight: 1.2
    aliases: [hud]
    discover: { rss: true, sitemap: true, pages: true }
    notes: "huduser.gov RSS (pub.xml) answers 202 with an empty WAF challenge and /press is a thin shell; the /news index is server-rendered."
    sensors:
      - { name: news, url: "https://www.hud.gov/news", type: HTML, connector: http, tier: B }
  - id: fhfa
    name: Federal Housing Finance Agency
    domain: fhfa.gov
    categories: [real-estate, finance, government]
    tier: B
    aliases: [fhfa, house price index]
    products:
      - { name: FHFA House Price Index, type: index, aliases: [fhfa hpi] }
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: news releases, url: "https://www.fhfa.gov/news/news-release", type: HTML, connector: http, tier: B }
  - id: cotality
    name: Cotality (formerly CoreLogic)
    domain: cotality.com
    categories: [real-estate]
    tier: C
    aliases: [cotality, corelogic, core logic]
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: newsroom, url: "https://www.cotality.com/newsroom", type: HTML, connector: http, tier: C }
      - { name: sitemap, url: "https://www.cotality.com/sitemap.xml", type: SITEMAP, connector: sitemap, tier: C, config: { maxUrls: 4000 } }
  - id: prologis
    name: Prologis
    domain: prologis.com
    categories: [real-estate, logistics, finance]
    tier: C
    aliases: [prologis, pld]
    discover: { rss: false }
    notes: "prologis.com is behind Cloudflare (403); the investor-relations site publishes an open RSS."
    sensors:
      - { name: press releases feed, url: "https://ir.prologis.com/news-events/press-releases/rss", type: RSS, connector: rss, tier: C }
  - id: habitat-for-humanity
    name: Habitat for Humanity
    domain: habitat.org
    categories: [real-estate, ngo]
    tier: C
    aliases: [habitat for humanity, habitat]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://www.habitat.org/rss.xml", type: RSS, connector: rss, tier: C }
  - id: opendoor
    name: Opendoor
    domain: opendoor.com
    categories: [real-estate, technology]
    tier: C
    aliases: [opendoor]
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: newsroom, url: "https://www.opendoor.com/newsroom", type: HTML, connector: http, tier: C }
  - id: compass
    name: Compass
    domain: compass.com
    categories: [real-estate, technology]
    tier: C
    aliases: [compass real estate]
    discover: { rss: false, sitemap: false, pages: true }
    sensors:
      - { name: newsroom, url: "https://www.compass.com/newsroom/", type: HTML, connector: http, tier: C }
  - id: jll
    name: JLL
    domain: jll.com
    categories: [real-estate]
    tier: C
    aliases: [jll, jones lang lasalle]
    discover: { rss: false, sitemap: true, pages: true }
    sensors:
      - { name: newsroom (us), url: "https://www.jll.com/en-us/newsroom", type: HTML, connector: http, tier: C }
  - id: altus-group
    name: Altus Group
    domain: altusgroup.com
    categories: [real-estate]
    tier: C
    aliases: [altus group, altus]
    discover: { rss: false, sitemap: true }
    sensors:
      - { name: sitemap, url: "https://www.altusgroup.com/sitemap.xml", type: SITEMAP, connector: sitemap, tier: C, config: { maxUrls: 3000 } }

  # ───────────────────────── Housing & real estate — UK / Europe / Australia ─────────────────────────
  - id: rightmove
    name: Rightmove
    domain: rightmove.co.uk
    categories: [real-estate]
    tier: B
    aliases: [rightmove, rightmove house price index]
    products:
      - { name: Rightmove House Price Index, type: index }
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://www.rightmove.co.uk/news/feed/", type: RSS, connector: rss, tier: B }
  - id: nationwide-hpi
    name: Nationwide House Price Index
    domain: nationwide.co.uk
    homepage: https://www.nationwide.co.uk/media/hpi
    categories: [real-estate, finance]
    tier: B
    aliases: [nationwide house price index, nationwide hpi, nationwide building society]
    discover: { rss: false, pages: true }
    notes: "nationwidehousepriceindex.co.uk redirects to the Onclusive-hosted media centre on nationwide.co.uk; the reports index is server-rendered."
    sensors:
      - { name: hpi reports, url: "https://www.nationwide.co.uk/media/hpi/reports", type: HTML, connector: http, tier: B }
  - id: hm-land-registry
    name: HM Land Registry
    domain: gov.uk
    homepage: https://www.gov.uk/government/organisations/land-registry
    categories: [real-estate, government]
    tier: B
    aliases: [hm land registry, land registry, uk house price index]
    products:
      - { name: UK House Price Index, type: index, aliases: [uk hpi] }
    discover: { rss: false }
    sensors:
      - { name: publications feed, url: "https://www.gov.uk/government/organisations/land-registry.atom", type: ATOM, connector: rss, tier: B }
  - id: rea-group
    name: REA Group
    domain: rea-group.com
    categories: [real-estate, technology]
    tier: C
    aliases: [rea group, realestate.com.au]
    products:
      - { name: realestate.com.au, type: service }
    discover: { rss: false, sitemap: true, pages: true }
    notes: "Corporate WordPress feeds exist but carry no items; the blog archive is server-rendered."
    sensors:
      - { name: corporate blog, url: "https://www.rea-group.com/about-us/news-and-insights/blog/", type: HTML, connector: http, tier: C }
      - { name: realestate.com.au news feed, url: "https://www.realestate.com.au/news/feed/", type: RSS, connector: rss, tier: B }
  - id: immobilienscout24
    name: ImmoScout24
    domain: immobilienscout24.de
    categories: [real-estate, technology]
    tier: C
    aliases: [immoscout24, immobilienscout24, scout24]
    discover: { rss: false, sitemap: true }
    sensors:
      - { name: wissen sitemap, url: "https://www.immobilienscout24.de/wissen/sitemap.xml", type: SITEMAP, connector: sitemap, tier: C, config: { maxUrls: 3000 } }

  # ───────────────────────── Labour & workforce ─────────────────────────
  - id: us-dol
    name: U.S. Department of Labor
    domain: dol.gov
    categories: [labour, government]
    tier: A
    weight: 1.3
    aliases: [dol, department of labor, labor department]
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: news releases feed, url: "https://www.dol.gov/rss/releases.xml", type: RSS, connector: rss, tier: A }
  - id: eeoc
    name: U.S. Equal Employment Opportunity Commission
    domain: eeoc.gov
    categories: [labour, government, legal]
    tier: B
    aliases: [eeoc]
    discover: { rss: true }
    sensors:
      - { name: newsroom feed, url: "https://www.eeoc.gov/rss/newsroom", type: RSS, connector: rss, tier: B }
  - id: nlrb
    name: National Labor Relations Board
    domain: nlrb.gov
    categories: [labour, government, legal]
    tier: B
    aliases: [nlrb]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://www.nlrb.gov/rss.xml", type: RSS, connector: rss, tier: B }
  - id: esdc
    name: Employment and Social Development Canada
    domain: canada.ca
    homepage: https://www.canada.ca/en/employment-social-development.html
    categories: [labour, government]
    tier: B
    weight: 1.2
    aliases: [esdc, edsc, emploi et développement social canada, service canada, labour program]
    discover: { rss: false }
    sensors:
      - { name: news releases feed, url: "https://api.io.canada.ca/io-server/gc/news/en/v2?dept=departmentofemploymentandsocialdevelopment&type=newsreleases&sort=publishedDate&orderBy=desc&pick=50&format=atom", type: ATOM, connector: rss, tier: B }
  - id: cnesst
    name: Commission des normes, de l'équité, de la santé et de la sécurité du travail
    domain: cnesst.gouv.qc.ca
    categories: [labour, government]
    tier: B
    aliases: [cnesst, csst, commission des normes du travail, cnt]
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: communiqués, url: "https://www.cnesst.gouv.qc.ca/fr/salle-presse/communiques", type: HTML, connector: http, tier: B }
  - id: indeed-hiring-lab
    name: Indeed Hiring Lab
    domain: hiringlab.org
    categories: [labour, research]
    tier: B
    aliases: [hiring lab, indeed]
    discover: { rss: true }
    sensors:
      - { name: research feed, url: "https://hiringlab.indeed.com/feed/", type: RSS, connector: rss, tier: B }
  - id: adp-research
    name: ADP Research
    domain: adpresearch.com
    categories: [labour, research]
    tier: B
    aliases: [adp research institute, adp national employment report]
    products:
      - { name: ADP National Employment Report, type: product, aliases: [adp jobs report] }
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: research feed, url: "https://www.adpresearch.com/research/feed.xml", type: RSS, connector: rss, tier: B }
  - id: conference-board
    name: The Conference Board
    domain: conference-board.org
    categories: [labour, research, finance]
    tier: B
    aliases: [conference board, consumer confidence index, leading economic index]
    products:
      - { name: Consumer Confidence Index, type: index }
      - { name: Leading Economic Index, type: index, aliases: [lei] }
    discover: { rss: true }
    sensors:
      - { name: press releases feed, url: "https://www.conference-board.org/rss/rss.cfm?type=press", type: RSS, connector: rss, tier: B }
  - id: challenger-gray
    name: Challenger, Gray & Christmas
    domain: challengergray.com
    categories: [labour, research]
    tier: B
    aliases: [challenger gray, challenger job cuts report]
    products:
      - { name: Challenger Job Cuts Report, type: product }
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://www.challengergray.com/feed/", type: RSS, connector: rss, tier: B }
  - id: teamsters
    name: International Brotherhood of Teamsters
    domain: teamster.org
    categories: [labour]
    tier: C
    aliases: [teamsters, ibt]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://teamster.org/feed/", type: RSS, connector: rss, tier: C }
  - id: uaw
    name: United Auto Workers
    domain: uaw.org
    categories: [labour, automotive]
    tier: C
    aliases: [uaw]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://uaw.org/feed/", type: RSS, connector: rss, tier: C }
  - id: cupe
    name: Canadian Union of Public Employees
    domain: cupe.ca
    categories: [labour]
    tier: C
    aliases: [cupe, scfp]
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: news feed, url: "https://cupe.ca/rss/promoted", type: RSS, connector: rss, tier: C }
  - id: unifor
    name: Unifor
    domain: unifor.org
    categories: [labour]
    tier: C
    aliases: [unifor]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://www.unifor.org/rss.xml", type: RSS, connector: rss, tier: C }
  - id: ftq
    name: Fédération des travailleurs et travailleuses du Québec
    domain: ftq.qc.ca
    categories: [labour]
    tier: C
    aliases: [ftq]
    discover: { rss: true }
    sensors:
      - { name: actualités feed, url: "https://ftq.qc.ca/feed/", type: RSS, connector: rss, tier: C }
  - id: csn
    name: Confédération des syndicats nationaux
    domain: csn.qc.ca
    categories: [labour]
    tier: C
    aliases: [csn]
    discover: { rss: true }
    sensors:
      - { name: actualités feed, url: "https://www.csn.qc.ca/feed/", type: RSS, connector: rss, tier: C }
  - id: csq
    name: Centrale des syndicats du Québec
    domain: lacsq.org
    categories: [labour, education]
    tier: C
    aliases: [csq]
    discover: { rss: true }
    sensors:
      - { name: actualités feed, url: "https://www.lacsq.org/feed/", type: RSS, connector: rss, tier: C }
  - id: fiq
    name: Fédération interprofessionnelle de la santé du Québec
    domain: fiqsante.qc.ca
    categories: [labour, health]
    tier: C
    aliases: [fiq]
    discover: { rss: true }
    sensors:
      - { name: actualités feed, url: "https://www.fiqsante.qc.ca/feed/", type: RSS, connector: rss, tier: C }
  - id: clc
    name: Canadian Labour Congress
    domain: canadianlabour.ca
    categories: [labour]
    tier: C
    aliases: [clc, congrès du travail du canada, ctc]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://canadianlabour.ca/feed/", type: RSS, connector: rss, tier: C }
  - id: psac
    name: Public Service Alliance of Canada
    domain: psacunion.ca
    categories: [labour]
    tier: C
    aliases: [psac, afpc]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://psacunion.ca/rss.xml", type: RSS, connector: rss, tier: C }
  - id: wsib
    name: Workplace Safety and Insurance Board (Ontario)
    domain: wsib.ca
    categories: [labour, government]
    tier: C
    aliases: [wsib]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://www.wsib.ca/en/rss.xml", type: RSS, connector: rss, tier: C }
  - id: worksafebc
    name: WorkSafeBC
    domain: worksafebc.com
    categories: [labour, government]
    tier: C
    aliases: [worksafebc, worksafe bc]
    discover: { rss: false, sitemap: true, pages: true }
    sensors:
      - { name: news & events, url: "https://www.worksafebc.com/en/about-us/news-events", type: HTML, connector: http, tier: C }
  - id: lmic
    name: Labour Market Information Council
    domain: lmic-cimt.ca
    categories: [labour, research, statistics]
    tier: C
    aliases: [lmic, cimt]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://lmic-cimt.ca/feed/", type: RSS, connector: rss, tier: C }

  # ───────────────────────── Open-data portals (catalog APIs, llm: false) ─────────────────────────
  - id: canada
    extend: true
    aliases: [open government canada, open.canada.ca]
    categories: [open-data]
    sensors:
      - { name: open data catalogue (ckan), url: "https://open.canada.ca/data/api/3/action/package_search?sort=metadata_modified+desc&rows=50", type: REST_API, connector: jsonlist, tier: B, config: { itemsPath: result.results, keyField: id, titleField: title, summaryField: notes, dateField: metadata_modified, compareFields: [metadata_modified], maxItems: 50, urlTemplate: "https://open.canada.ca/data/en/dataset/{key}" } }
  - id: quebec
    extend: true
    aliases: [données québec, donneesquebec]
    categories: [open-data]
    sensors:
      - { name: données québec catalogue (ckan), url: "https://www.donneesquebec.ca/recherche/api/3/action/package_search?sort=metadata_modified+desc&rows=50", type: REST_API, connector: jsonlist, tier: B, config: { itemsPath: result.results, keyField: id, titleField: title, summaryField: notes, dateField: metadata_modified, compareFields: [metadata_modified], maxItems: 50, urlTemplate: "https://www.donneesquebec.ca/recherche/dataset/{key}" } }
  - id: ontario
    extend: true
    aliases: [ontario data catalogue, data.ontario.ca]
    categories: [open-data]
    sensors:
      - { name: data catalogue (ckan), url: "https://data.ontario.ca/api/3/action/package_search?sort=metadata_modified+desc&rows=50", type: REST_API, connector: jsonlist, tier: B, config: { itemsPath: result.results, keyField: id, titleField: title, summaryField: notes, dateField: metadata_modified, compareFields: [metadata_modified], maxItems: 50, urlTemplate: "https://data.ontario.ca/dataset/{key}" } }
  - id: british-columbia
    extend: true
    aliases: [bc data catalogue]
    categories: [open-data]
    sensors:
      - { name: bc data catalogue (ckan), url: "https://catalogue.data.gov.bc.ca/api/3/action/package_search?sort=metadata_modified+desc&rows=50", type: REST_API, connector: jsonlist, tier: B, config: { itemsPath: result.results, keyField: id, titleField: title, summaryField: notes, dateField: metadata_modified, compareFields: [metadata_modified], maxItems: 50, urlTemplate: "https://catalogue.data.gov.bc.ca/dataset/{key}" } }
  - id: alberta
    extend: true
    aliases: [open alberta]
    categories: [open-data]
    sensors:
      - { name: open alberta catalogue (ckan), url: "https://open.alberta.ca/api/3/action/package_search?sort=metadata_modified+desc&rows=50", type: REST_API, connector: jsonlist, tier: B, config: { itemsPath: result.results, keyField: id, titleField: title, summaryField: notes, dateField: metadata_modified, compareFields: [metadata_modified], maxItems: 50, urlTemplate: "https://open.alberta.ca/dataset/{key}" } }
  - id: uk-government
    extend: true
    aliases: [data.gov.uk]
    categories: [open-data]
    sensors:
      - { name: data.gov.uk catalogue (ckan), url: "https://ckan.publishing.service.gov.uk/api/action/package_search?sort=metadata_modified+desc&rows=50", type: REST_API, connector: jsonlist, tier: B, config: { itemsPath: result.results, keyField: id, titleField: title, summaryField: notes, dateField: metadata_modified, compareFields: [metadata_modified], maxItems: 50, urlTemplate: "https://www.data.gov.uk/dataset/{key}" } }
  - id: nasa
    extend: true
    aliases: [data.nasa.gov]
    categories: [open-data]
    sensors:
      - { name: open data catalogue (ckan), url: "https://data.nasa.gov/api/3/action/package_search?sort=metadata_modified+desc&rows=50", type: REST_API, connector: jsonlist, tier: C, config: { itemsPath: result.results, keyField: id, titleField: title, summaryField: notes, dateField: metadata_modified, compareFields: [metadata_modified], maxItems: 50, urlTemplate: "https://data.nasa.gov/dataset/{key}" } }
  - id: montreal-open-data
    name: Ville de Montréal
    domain: montreal.ca
    homepage: https://donnees.montreal.ca
    categories: [open-data, government]
    tier: B
    aliases: [ville de montréal, city of montreal, données ouvertes montréal]
    discover: { rss: false }
    llm: false
    sensors:
      - { name: données ouvertes (ckan), url: "https://donnees.montreal.ca/api/3/action/package_search?sort=metadata_modified+desc&rows=50", type: REST_API, connector: jsonlist, tier: B, config: { itemsPath: result.results, keyField: id, titleField: title, summaryField: notes, dateField: metadata_modified, compareFields: [metadata_modified], maxItems: 50, urlTemplate: "https://donnees.montreal.ca/dataset/{key}" } }
  - id: toronto-open-data
    name: City of Toronto
    domain: toronto.ca
    homepage: https://open.toronto.ca
    categories: [open-data, government]
    tier: B
    aliases: [city of toronto, toronto open data]
    discover: { rss: false }
    llm: false
    sensors:
      - { name: open data catalogue (ckan), url: "https://ckan0.cf.opendata.inter.prod-toronto.ca/api/3/action/package_search?sort=metadata_modified+desc&rows=50", type: REST_API, connector: jsonlist, tier: B, config: { itemsPath: result.results, keyField: id, titleField: title, summaryField: notes, dateField: metadata_modified, compareFields: [metadata_modified], maxItems: 50, urlTemplate: "https://open.toronto.ca/dataset/{key}" } }
  - id: vancouver-open-data
    name: City of Vancouver
    domain: vancouver.ca
    homepage: https://opendata.vancouver.ca
    categories: [open-data, government]
    tier: C
    aliases: [city of vancouver, vancouver open data]
    discover: { rss: false }
    llm: false
    sensors:
      - { name: open data catalogue (opendatasoft), url: "https://opendata.vancouver.ca/api/explore/v2.1/catalog/datasets?order_by=modified%20desc&limit=50", type: REST_API, connector: jsonlist, tier: C, config: { itemsPath: results, keyField: dataset_id, titleField: metas.default.title, summaryField: metas.default.description, dateField: metas.default.modified, compareFields: [metas.default.modified], maxItems: 50, urlTemplate: "https://opendata.vancouver.ca/explore/dataset/{key}/" } }
  - id: calgary-open-data
    name: City of Calgary
    domain: calgary.ca
    homepage: https://data.calgary.ca
    categories: [open-data, government]
    tier: C
    aliases: [city of calgary, calgary open data]
    discover: { rss: false }
    llm: false
    sensors:
      - { name: open data catalogue (socrata), url: "https://api.us.socrata.com/api/catalog/v1?domains=data.calgary.ca&order=updatedAt&limit=50", type: REST_API, connector: jsonlist, tier: C, config: { itemsPath: results, keyField: resource.id, titleField: resource.name, summaryField: resource.description, dateField: resource.updatedAt, urlField: link, compareFields: [resource.updatedAt], maxItems: 50 } }
  - id: edmonton-open-data
    name: City of Edmonton
    domain: edmonton.ca
    homepage: https://data.edmonton.ca
    categories: [open-data, government]
    tier: C
    aliases: [city of edmonton, edmonton open data]
    discover: { rss: false }
    llm: false
    sensors:
      - { name: open data catalogue (socrata), url: "https://api.us.socrata.com/api/catalog/v1?domains=data.edmonton.ca&order=updatedAt&limit=50", type: REST_API, connector: jsonlist, tier: C, config: { itemsPath: results, keyField: resource.id, titleField: resource.name, summaryField: resource.description, dateField: resource.updatedAt, urlField: link, compareFields: [resource.updatedAt], maxItems: 50 } }
  - id: nyc-open-data
    name: NYC Open Data
    domain: cityofnewyork.us
    homepage: https://opendata.cityofnewyork.us
    categories: [open-data, government]
    tier: B
    aliases: [nyc open data, city of new york, new york city]
    discover: { rss: false }
    llm: false
    sensors:
      - { name: open data catalogue (socrata), url: "https://api.us.socrata.com/api/catalog/v1?domains=data.cityofnewyork.us&order=updatedAt&limit=50", type: REST_API, connector: jsonlist, tier: B, config: { itemsPath: results, keyField: resource.id, titleField: resource.name, summaryField: resource.description, dateField: resource.updatedAt, urlField: link, compareFields: [resource.updatedAt], maxItems: 50 } }
  - id: chicago-open-data
    name: Chicago Data Portal
    domain: cityofchicago.org
    homepage: https://data.cityofchicago.org
    categories: [open-data, government]
    tier: C
    aliases: [chicago data portal, city of chicago]
    discover: { rss: false }
    llm: false
    sensors:
      - { name: open data catalogue (socrata), url: "https://api.us.socrata.com/api/catalog/v1?domains=data.cityofchicago.org&order=updatedAt&limit=50", type: REST_API, connector: jsonlist, tier: C, config: { itemsPath: results, keyField: resource.id, titleField: resource.name, summaryField: resource.description, dateField: resource.updatedAt, urlField: link, compareFields: [resource.updatedAt], maxItems: 50 } }
  - id: paris-open-data
    name: Ville de Paris (Open Data)
    domain: paris.fr
    homepage: https://opendata.paris.fr
    categories: [open-data, government]
    tier: C
    aliases: [ville de paris, paris data, opendata paris]
    discover: { rss: false }
    llm: false
    sensors:
      - { name: open data catalogue (opendatasoft), url: "https://opendata.paris.fr/api/explore/v2.1/catalog/datasets?order_by=modified%20desc&limit=50", type: REST_API, connector: jsonlist, tier: C, config: { itemsPath: results, keyField: dataset_id, titleField: metas.default.title, summaryField: metas.default.description, dateField: metas.default.modified, compareFields: [metas.default.modified], maxItems: 50, urlTemplate: "https://opendata.paris.fr/explore/dataset/{key}/" } }
  - id: data-gouv-fr
    name: data.gouv.fr (Etalab)
    domain: data.gouv.fr
    categories: [open-data, government]
    tier: B
    weight: 1.1
    aliases: [data.gouv.fr, etalab, dinum]
    discover: { rss: false }
    llm: false
    notes: "Sort key is last_update (not last_modified); dataset page URL is in `page`."
    sensors:
      - { name: datasets api, url: "https://www.data.gouv.fr/api/1/datasets/?sort=-last_update&page_size=50", type: REST_API, connector: jsonlist, tier: B, config: { itemsPath: data, keyField: id, titleField: title, summaryField: description_short, dateField: last_modified, urlField: page, compareFields: [last_modified], maxItems: 50 } }
  - id: data-europa-eu
    name: data.europa.eu (EU Open Data Portal)
    domain: data.europa.eu
    categories: [open-data, government, international]
    tier: B
    weight: 1.1
    aliases: [eu open data portal, european data portal, publications office open data]
    discover: { rss: false }
    llm: false
    sensors:
      - { name: datasets search api, url: "https://data.europa.eu/api/hub/search/search?q=&sort=modified%2Bdesc&limit=50", type: REST_API, connector: jsonlist, tier: B, config: { itemsPath: result.results, keyField: id, titleField: title.en, summaryField: description.en, dateField: modified, compareFields: [modified], maxItems: 50, urlTemplate: "https://data.europa.eu/data/datasets/{key}" } }
  - id: data-gov-au
    name: data.gov.au
    domain: data.gov.au
    categories: [open-data, government]
    tier: C
    aliases: [data.gov.au, australian government open data]
    discover: { rss: false }
    llm: false
    sensors:
      - { name: catalogue (ckan), url: "https://data.gov.au/data/api/3/action/package_search?sort=metadata_modified+desc&rows=50", type: REST_API, connector: jsonlist, tier: C, config: { itemsPath: result.results, keyField: id, titleField: title, summaryField: notes, dateField: metadata_modified, compareFields: [metadata_modified], maxItems: 50, urlTemplate: "https://data.gov.au/data/dataset/{key}" } }
  - id: hdx
    name: Humanitarian Data Exchange (UN OCHA)
    domain: data.humdata.org
    categories: [open-data, international, ngo]
    tier: B
    aliases: [hdx, humdata, ocha data]
    discover: { rss: false }
    llm: false
    sensors:
      - { name: datasets (ckan), url: "https://data.humdata.org/api/3/action/package_search?sort=metadata_modified+desc&rows=50", type: REST_API, connector: jsonlist, tier: B, config: { itemsPath: result.results, keyField: id, titleField: title, summaryField: notes, dateField: metadata_modified, compareFields: [metadata_modified], maxItems: 50, urlTemplate: "https://data.humdata.org/dataset/{key}" } }
  - id: our-world-in-data
    name: Our World in Data
    domain: ourworldindata.org
    categories: [open-data, research, statistics]
    tier: B
    weight: 1.2
    aliases: [owid, our world in data]
    products:
      - { name: OWID ETL, type: software, aliases: [owid etl] }
    discover: { rss: true, sitemap: false }
    sensors:
      - { name: articles feed, url: "https://ourworldindata.org/atom.xml", type: ATOM, connector: rss, tier: B }
      - { name: datasets repo commits, url: "https://github.com/owid/owid-datasets/commits/master.atom", type: GITHUB_RELEASE, connector: github, tier: C, config: { repo: owid/owid-datasets, kind: commits, branch: master } }
  - id: data-commons
    name: Data Commons (Google)
    domain: datacommons.org
    categories: [open-data, technology]
    tier: C
    aliases: [data commons]
    discover: { rss: true }
    sensors:
      - { name: blog feed, url: "https://blog.datacommons.org/feed/", type: RSS, connector: rss, tier: C }
  - id: open-knowledge-foundation
    name: Open Knowledge Foundation
    domain: okfn.org
    categories: [open-data, ngo, open-source]
    tier: C
    aliases: [okfn, open knowledge, ckan]
    products:
      - { name: CKAN, type: software }
    discover: { rss: true }
    sensors:
      - { name: blog feed, url: "https://blog.okfn.org/feed/", type: RSS, connector: rss, tier: C }
      - { name: ckan releases, url: "https://github.com/ckan/ckan/releases.atom", type: GITHUB_RELEASE, connector: github, tier: C, config: { repo: ckan/ckan, kind: releases } }

  # ───────────────────────── National statistics offices ─────────────────────────
  - id: ons
    name: Office for National Statistics
    domain: ons.gov.uk
    categories: [statistics, government]
    tier: A
    weight: 1.3
    aliases: [ons, uk statistics]
    discover: { rss: true, sitemap: false, pages: true }
    sensors:
      - { name: release calendar feed, url: "https://www.ons.gov.uk/releasecalendar?rss", type: RSS, connector: rss, tier: A }
  - id: insee
    name: INSEE
    domain: insee.fr
    categories: [statistics, government]
    tier: A
    weight: 1.2
    aliases: [insee, institut national de la statistique et des études économiques]
    discover: { rss: true, sitemap: true, pages: true }
    notes: "No RSS any more; the statistics index (latest publications) is server-rendered."
    sensors:
      - { name: latest statistics (en), url: "https://www.insee.fr/en/statistiques", type: HTML, connector: http, tier: A }
  - id: destatis
    name: Destatis (Statistisches Bundesamt)
    domain: destatis.de
    categories: [statistics, government]
    tier: A
    weight: 1.2
    aliases: [destatis, statistisches bundesamt, federal statistical office of germany]
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: press releases (en), url: "https://www.destatis.de/EN/Press/_node.html", type: HTML, connector: http, tier: A }
      - { name: sitemap index, url: "https://www.destatis.de/Sitemap_Index.xml", type: SITEMAP, connector: sitemap, tier: C, config: { maxUrls: 3000, maxChildren: 2 } }
  - id: istat
    name: ISTAT
    domain: istat.it
    categories: [statistics, government]
    tier: B
    weight: 1.1
    aliases: [istat, istituto nazionale di statistica]
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: news feed (en), url: "https://www.istat.it/en/feed/", type: RSS, connector: rss, tier: B }
      - { name: notizie feed (it), url: "https://www.istat.it/feed/", type: RSS, connector: rss, tier: B }
  - id: ine-spain
    name: INE (Instituto Nacional de Estadística)
    domain: ine.es
    categories: [statistics, government]
    tier: B
    weight: 1.1
    aliases: [ine, ine españa, spanish statistical office]
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: sitemap, url: "https://ine.es/sitemap.xml", type: SITEMAP, connector: sitemap, tier: B, config: { maxUrls: 3000 } }
  - id: cbs-netherlands
    name: Statistics Netherlands (CBS)
    domain: cbs.nl
    categories: [statistics, government]
    tier: B
    weight: 1.1
    aliases: [cbs, centraal bureau voor de statistiek, statistics netherlands]
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: news sitemap, url: "https://www.cbs.nl/news-sitemap.xml", type: SITEMAP, connector: sitemap, tier: B, config: { maxUrls: 3000, maxChildren: 8 } }
  - id: statistics-sweden
    name: Statistics Sweden (SCB)
    domain: scb.se
    categories: [statistics, government]
    tier: B
    aliases: [scb, statistiska centralbyrån, statistics sweden]
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: statistiknyheter feed, url: "https://www.scb.se/feed/statistiknyheter", type: ATOM, connector: rss, tier: B }
      - { name: statistical news (en), url: "https://www.scb.se/en/finding-statistics/statistical-news/", type: HTML, connector: http, tier: B }
  - id: statistics-norway
    name: Statistics Norway (SSB)
    domain: ssb.no
    categories: [statistics, government]
    tier: B
    aliases: [ssb, statistisk sentralbyrå, statistics norway]
    discover: { rss: false, sitemap: true, pages: true }
    notes: "The RSS feeds (/rss/statistikk, /rss/nyheter) are served but empty; the upcoming-releases page is server-rendered."
    sensors:
      - { name: upcoming releases (en), url: "https://www.ssb.no/en/kommende-publiseringer", type: HTML, connector: http, tier: B }
  - id: statistics-denmark
    name: Statistics Denmark
    domain: dst.dk
    categories: [statistics, government]
    tier: B
    aliases: [dst, danmarks statistik, statistics denmark]
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: scheduled releases, url: "https://www.dst.dk/en/Statistik/planlagte", type: HTML, connector: http, tier: B }
  - id: statistics-austria
    name: Statistics Austria
    domain: statistik.at
    categories: [statistics, government]
    tier: C
    aliases: [statistik austria, statistics austria]
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: sitemap index, url: "https://www.statistik.at/sitemap.xml", type: SITEMAP, connector: sitemap, tier: C, config: { maxUrls: 3000, maxChildren: 3 } }
  - id: statbel
    name: Statbel (Statistics Belgium)
    domain: statbel.fgov.be
    categories: [statistics, government]
    tier: B
    aliases: [statbel, statistics belgium]
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: news feed (en), url: "https://statbel.fgov.be/en/rss.xml", type: RSS, connector: rss, tier: B }
  - id: statistics-poland
    name: Statistics Poland (GUS)
    domain: stat.gov.pl
    categories: [statistics, government]
    tier: B
    aliases: [gus, główny urząd statystyczny, statistics poland]
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: news feed (en), url: "https://stat.gov.pl/rss/en/3/3.xml", type: RSS, connector: rss, tier: B }
  - id: czso
    name: Czech Statistical Office
    domain: csu.gov.cz
    categories: [statistics, government]
    tier: C
    aliases: [czso, csu, český statistický úřad]
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: news releases feed (en), url: "https://csu.gov.cz/rss/produkty?kodVlastnostiVystupu=RI&jazyk=EN", type: RSS, connector: rss, tier: C }
  - id: statistics-estonia
    name: Statistics Estonia
    domain: stat.ee
    categories: [statistics, government]
    tier: C
    aliases: [statistics estonia, statistikaamet]
    discover: { rss: true, sitemap: true }
    sensors:
      - { name: news feed (en), url: "https://www.stat.ee/en/rss", type: RSS, connector: rss, tier: C }
  - id: statistics-iceland
    name: Statistics Iceland
    domain: statice.is
    categories: [statistics, government]
    tier: C
    aliases: [statistics iceland, hagstofa íslands]
    discover: { rss: true, pages: true }
    sensors:
      - { name: news archive, url: "https://www.statice.is/publications/news-archive/", type: HTML, connector: http, tier: C }
  - id: ine-portugal
    name: Statistics Portugal (INE)
    domain: ine.pt
    categories: [statistics, government]
    tier: C
    aliases: [ine portugal, instituto nacional de estatística, statistics portugal]
    discover: { rss: true, pages: true }
    sensors:
      - { name: highlights (en), url: "https://www.ine.pt/xportal/xmain?xpid=INE&xpgid=ine_destaques", type: HTML, connector: http, tier: C }
  - id: cso-ireland
    name: Central Statistics Office (Ireland)
    domain: cso.ie
    categories: [statistics, government]
    tier: B
    aliases: [cso, cso ireland, central statistics office]
    discover: { rss: false, pages: true }
    notes: "/en/rss/ answers 403; the press-release index is server-rendered."
    sensors:
      - { name: press releases, url: "https://www.cso.ie/en/csolatestnews/pressreleases/", type: HTML, connector: http, tier: B }
  - id: statistics-bureau-japan
    name: Statistics Bureau of Japan
    domain: stat.go.jp
    categories: [statistics, government]
    tier: B
    aliases: [statistics bureau of japan, stat.go.jp, e-stat, sbj]
    discover: { rss: true, pages: true }
    notes: "e-Stat (e-stat.go.jp) publishes no feed; the Bureau's English news index is static HTML."
    sensors:
      - { name: news (en), url: "https://www.stat.go.jp/english/info/news/index.html", type: HTML, connector: http, tier: B }
  - id: kostat
    name: Statistics Korea (Ministry of Data and Statistics)
    domain: kostat.go.kr
    categories: [statistics, government]
    tier: B
    aliases: [kostat, statistics korea, ministry of data and statistics, mods]
    discover: { rss: true }
    notes: "kostat.go.kr now redirects to mods.go.kr (Ministry of Data and Statistics, 2025 reorganisation)."
    sensors:
      - { name: press releases feed, url: "https://mods.go.kr/board.es?mid=a20101000000&bid=11737&act=rss", type: RSS, connector: rss, tier: B }
  - id: nbs-china
    name: National Bureau of Statistics of China
    domain: stats.gov.cn
    categories: [statistics, government]
    tier: B
    weight: 1.1
    aliases: [nbs, national bureau of statistics, 国家统计局]
    discover: { rss: false, pages: true }
    notes: "No RSS (rss.xml 404); the English press-release index is server-rendered but the host is slow from North America."
    sensors:
      - { name: press releases (en), url: "https://www.stats.gov.cn/english/PressRelease/", type: HTML, connector: http, tier: B }
  - id: abs
    name: Australian Bureau of Statistics
    domain: abs.gov.au
    categories: [statistics, government]
    tier: A
    weight: 1.2
    aliases: [abs, australian bureau of statistics]
    discover: { rss: true, sitemap: true, pages: true }
    sensors:
      - { name: media releases, url: "https://www.abs.gov.au/media-centre/media-releases", type: HTML, connector: http, tier: A }
      - { name: release calendar, url: "https://www.abs.gov.au/release-calendar/future-releases", type: HTML, connector: http, tier: B }
  - id: stats-nz
    name: Stats NZ
    domain: stats.govt.nz
    categories: [statistics, government]
    tier: B
    aliases: [stats nz, statistics new zealand, tatauranga aotearoa]
    discover: { rss: true, sitemap: true, pages: true }
    notes: "News and release-calendar pages are client-rendered (thin); the sitemap index is the only server-side signal."
    sensors:
      - { name: sitemap index, url: "https://www.stats.govt.nz/sitemap.xml", type: SITEMAP, connector: sitemap, tier: B, config: { maxUrls: 3000, maxChildren: 3 } }
  - id: statssa
    name: Statistics South Africa
    domain: statssa.gov.za
    categories: [statistics, government]
    tier: B
    aliases: [stats sa, statistics south africa]
    discover: { rss: true }
    sensors:
      - { name: news feed, url: "https://www.statssa.gov.za/?feed=rss2", type: RSS, connector: rss, tier: B }
  - id: israel-cbs
    name: Israel Central Bureau of Statistics
    domain: cbs.gov.il
    categories: [statistics, government]
    tier: C
    aliases: [israel cbs, central bureau of statistics israel]
    discover: { rss: false, pages: true }
    sensors:
      - { name: media releases (en), url: "https://www.cbs.gov.il/en/mediarelease/Pages/default.aspx", type: HTML, connector: http, tier: C }
  - id: singstat
    name: Singapore Department of Statistics
    domain: singstat.gov.sg
    categories: [statistics, government]
    tier: C
    aliases: [singstat, department of statistics singapore, dos]
    discover: { rss: false, sitemap: true, pages: true }
    sensors:
      - { name: news, url: "https://www.singstat.gov.sg/news", type: HTML, connector: http, tier: C }
      - { name: sitemap, url: "https://www.singstat.gov.sg/sitemap.xml", type: SITEMAP, connector: sitemap, tier: C, config: { maxUrls: 3000 } }
  - id: isq
    name: Institut de la statistique du Québec
    domain: statistique.quebec.ca
    categories: [statistics, government]
    tier: B
    weight: 1.1
    aliases: [isq, institut de la statistique du québec]
    discover: { rss: true, sitemap: false, pages: true }
    sensors:
      - { name: communiqués, url: "https://statistique.quebec.ca/fr/institut/salle-de-presse/communiques", type: HTML, connector: http, tier: B }