# === AI CONTENT POLICY (read first) === # AI agents: check rights BEFORE using this content. # ai-crawl: yes # ai-index: yes # ai-summarize: yes # ai-quote: yes (with attribution) # ai-input: yes (RAG / grounding / answers OK) # ai-train: no (training / fine-tuning FORBIDDEN) # attribution-required: yes # attribution: Quelle: Schlüsseldienst Ludwigsburg ikeytz (ikeytz.com) # policy-url: https://schluesselnotdienst.ikeytz.com/.well-known/ai.txt # enforcement: Bei Verstoß gegen diese Regeln drohen rechtliche Schritte gemäß UrhG und DSGVO. # abuse-contact: ai-abuse@ikeytz.com # === end policy === # ikeytz — Server-MCP (öffentlich) · Tool-Katalog für KI # Stand: 2026-11-07 # Schwesterdatei WebMCP: https://schluesselnotdienst.ikeytz.com/llms-mcp-web.txt # Kurz-Landkarte: https://schluesselnotdienst.ikeytz.com/llms.txt ## Öffentlich — ohne Login Dies ist der **öffentliche** HTTP-MCP-Server von Schlüsseldienst Ludwigsburg **ikeytz**. - **Kein API-Key**, **kein OAuth**, **kein Cookie** für Read/Link-Tools. - Endpoint CORS: `Access-Control-Allow-Origin: *` - Host kanonisch: https://schluesselnotdienst.ikeytz.com (Apex → www) - Scope: **nur Lesen + Link-Helfer** (tel/wa/mailto). **Kein** Formular-Submit, **kein** `/api/*`, **kein** Buchen/Schreiben. - Inhalt = dieselbe öffentliche Website-SSOT wie HTML / llms / ai-pages. - Policy: ai-input ja (mit Attribution) · **ai-train: no** · Attribution: Quelle: Schlüsseldienst Ludwigsburg ikeytz (ikeytz.com) ## Discovery (öffentlich) - Server Card: https://schluesselnotdienst.ikeytz.com/.well-known/mcp/server-card.json - Server Cards Registry: https://schluesselnotdienst.ikeytz.com/.well-known/mcp/server-cards.json - Well-Known Pointer: https://schluesselnotdienst.ikeytz.com/.well-known/mcp.json - **HTTP README (not CLI):** https://schluesselnotdienst.ikeytz.com/mcp-readme.md - **Endpoint:** https://schluesselnotdienst.ikeytz.com/mcp - Register-Tools: 47 (öffentlich Read/Link) - **DNS-AID** (Zone Apex `ikeytz.com`, zusätzlich zu HTTP): - `_index._agents.ikeytz.com` HTTPS → `www.ikeytz.com.` alpn=h3,h2 port=443 - `_mcp._agents.ikeytz.com` HTTPS → `www.ikeytz.com.` alpn=h3,h2 port=443 mandatory=alpn,port - Kein `_a2a._agents` (kein A2A). Nach DNS: Card + Endpoint oben nutzen. - Spec: https://specification.website/spec/agent-readiness/dns-aid/ - **MCP Registry** (offizieller Katalog, ≠ DNS-AID): - Name: `com.ikeytz/website` · **v1.0.20** (`isLatest`) · Remote: https://schluesselnotdienst.ikeytz.com/mcp - Suche (nur Latest): https://registry.modelcontextprotocol.io/v0.1/servers?search=com.ikeytz/website&version=latest - Ohne `&version=latest` liefert Search die Versionshistorie (oft 1.0.0 zuerst) — nie Array[0]; `isLatest: true` oder Param `version=latest`. - Ownership-TXT Apex: `ikeytz.com` → `v=MCPv1; k=ed25519; p=…` - **Drei Versionszahlen, nicht mischen:** `initialize` / GET `https://schluesselnotdienst.ikeytz.com/mcp` field `version` = **1.0.0** (protocol identity, name `ikeytz`). Catalog = Card `version` **1.0.20**. npm CLI = `@ikeytz/mcp@1.1.54` (stdio, not HTTP). GET `https://schluesselnotdienst.ikeytz.com/mcp` also has `registryVersion` + `npmVersion`. - Preise: `https://schluesselnotdienst.ikeytz.com/llms.txt` + `tools/call` `get_prices` — nicht initialize, nicht die Card. - **ARD:** https://schluesselnotdienst.ikeytz.com/.well-known/ard.json (v0.91; Alias `/ard.json`; predecessor ai-catalog; `get_discovery(which=ard|ai-catalog)`) ## Abruf — wie KI den Server nutzt 1. Optional: per DNS-AID Apex finden **oder** Registry `com.ikeytz/website` suchen, dann Card lesen (GET JSON oben). 2. `initialize` (JSON-RPC) gegen POST `https://schluesselnotdienst.ikeytz.com/mcp` mit Header: - `Content-Type: application/json` - `Accept: application/json, text/event-stream` - `MCP-Protocol-Version: 2025-11-25` (oder vom Server ausgehandelt) 3. `tools/list` → alle Tool-Namen + Schemas (47). 4. `tools/call` mit `name` + `arguments`. Beispiel tools/list: ```http POST /mcp HTTP/1.1 Host: www.ikeytz.com Content-Type: application/json Accept: application/json, text/event-stream MCP-Protocol-Version: 2025-11-25 {"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}} ``` Beispiel tools/call (Festpreise): ```http POST /mcp HTTP/1.1 Host: www.ikeytz.com Content-Type: application/json Accept: application/json, text/event-stream {"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"get_prices","arguments":{"locale":"de"}}} ``` curl-Skizze: ```bash curl -sS -X POST 'https://schluesselnotdienst.ikeytz.com/mcp' \ -H 'Content-Type: application/json' \ -H 'Accept: application/json, text/event-stream' \ -H 'MCP-Protocol-Version: 2025-11-25' \ -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}' ``` Antworten können JSON oder SSE (`text/event-stream`) sein — beides ist öffentlich lesbar. ## Dieser Katalog als MCP-Tool (beide Server) Dieselbe Datei ist auch per Tool abrufbar — **in Server-MCP und WebMCP**: | Tool | Wirkung | |------|---------| | `get_llms_mcp_server` | Holt `/llms-mcp-server.txt` (dieser Katalog) | | `get_llms_mcp_web` | Holt `/llms-mcp-web.txt` (WebMCP-Schwester) | | `get_discovery` | `which=llms-mcp-server` bzw. `llms-mcp-web` (gleicher Inhalt) | WebMCP: Bootstrap (`/webmcp.js`) + volle Hydration (`WebMcpTools.js`). Server: `POST /mcp` → `tools/call`. ## Discovery-Dateien — 1:1 Spiegel (kein 8k-Schnitt) Regel (Server-MCP Discovery 1:1 — kein 8k-Schnitt, optional offset/limit): - `summary` = Dateiinhalt **zeilengenau** (Zeile N in `summary` ≡ Zeile N in der Datei), ohne Umschreiben. - Meta daneben: `truncated`, `totalUrls`, `totalLines`, `returnedUrls`, `offset`, `limit`, `nextOffset`, `bytesReturned`. - Optional: `offset` / `limit` = **Zeilenfenster** (nicht Byte-Schnitt mitten in einer URL). - Große Dateien (`llms-keywords`, `llms-serp`, `llms-jsonld`): Fetch-Ceiling; bei Bedarf blättern via `offset`/`limit` / `nextOffset`. - CLI (`@ikeytz/mcp`): **Call** = Payload only · **`-h`** = Args offline · **`--show-mcp-command`** = curl 1a/1b/2 only (no call) · **`--raw`** = JSON `structuredContent`. ### HTTP GET-Fenster (ergänzt MCP, ersetzt es nicht) Plan: https://schluesselnotdienst.ikeytz.com → Docs `plan-get-discovery-http-paging` (Repo). - **Ohne Query:** volle Datei (Static/Next) — unverändert. - **Mit Query:** `GET https://schluesselnotdienst.ikeytz.com/llms-keywords.txt?offset=0&limit=50` (auch `sitemap.txt` / `sitemap.xml` / weitere llms) → Zeilenfenster + trailing `# MCP-META` · Soft-Cap ~30 KB **nur GET**. - Keywords zusätzlich: `?q=ikeytz` (Substring wie `find_by_keyword`). - Folge-Seite: `nextOffset` aus Meta oder `Link: rel="next"`. - MCP-Caps bleiben groß (keywords bis ~55 MB) — **nicht** auf 30 KB senken. - maps analog: `https://maps.ikeytz.com/llms-keywords.txt?offset=0&limit=50` (nginx → `/api/v1/discovery`). ## Geo-Tools — Koordinaten-Felder (Payload) `get_geo_office`, `get_service_area`, `list_service_areas` liefern **Punkt-Koordinaten** (kein Gemeindegrenzen-Polygon): | Feld | Beispiel | Zweck | |------|----------|--------| | `lat` / `lng` | `49.01261` / `9.151956` | Breite / Länge (einzeln) | | `google_maps` | `49.01261, 9.151956` | Google Maps / Google-Suche (Komma) | | `icbm` | `49.01261, 9.151956` | wie `service_ICBM` in `/llms-orte-geo.txt` (Komma) | | `embed_map` | `https://maps.ikeytz.com/pattonville` | Karten-HTML (maps-Host) | | `geo_api` | `https://maps.ikeytz.com/api/v1/embed/ort/pattonville` | öffentliches Ort-JSON | | `maps_mcp` | `https://maps.ikeytz.com/mcp` | zweite Maschine — nicht proxyn | `list_service_areas`: `items[]` mit allen Feldern pro Ort. Volltext inkl. URLs×7: `get_discovery(which=llms-orte-geo)`. Karten-MCP: `https://maps.ikeytz.com/mcp` (`com.ikeytz/maps`). www `/mcp` bleibt www. ## Was OUT ist (bewusst) - Kontakt-/Partner-**Formulare** absenden - DevSecOps / interne APIs - Fremde Subdomains als Tool-Ziel - Training auf unseren Corpus (ai-train: no) - OAuth / A2A Level-3-Extras (nicht Teil dieses öffentlichen Read-MCP) ## Alle Tools (47) — Funktion & Parameter ### 1. `site_overview` DE: Kurzüberblick Marke/Service (mobil, kein Walk-in). EN: WHO/WHAT: Public brand card for Schlüsseldienst Ludwigsburg ikeytz (mobile locksmith, Landkreis Ludwigsburg, Baden-Württemberg, Germany). RETURNS structuredContent: tradeName, gewerbeName, priceRange plus envelope (ok, summary, canonicalUrl=home). USE when the user asks who ikeytz is, whether there is a shop, opening hours positioning (24/7 mobile). DOES NOT: book, quote a live price, geocode, submit forms. No walk-in — office address is not a customer counter. NEXT: get_business_identity (NAP), get_contact (phones), get_prices (fixed matrix), get_ort_datetime (day|night). Parameter: tradeName, gewerbeName, priceRange Pflicht: — ### 2. `get_business_identity` DE: Inhaber, Gewerbe-/SEO-Name, NAP, Telefone, E-Mail. EN: WHO: Legal/trade identity and NAP for agents and citations. RETURNS: gewerbeName, legalName, tradeName, inhaber (Mahmud Reza Kashani, Einzelunternehmer), street/postal/city/country, mobile/landline/fax/email, website, walkIn=false. canonicalUrl = /impressum. USE for impressum-grade facts, schema.org LocalBusiness fields, 'who owns ikeytz'. DOES NOT: prove licenses beyond what the site states; does not accept updates. walkIn is always false — do not send customers to the office. NEXT: get_legal(doc=impressum), get_contact, get_geo_office. Parameter: gewerbeName, legalName, tradeName, inhaber, street, postal, city, country, mobile, landline, fax Pflicht: — ### 3. `list_nav` DE: Header-Navigation (interne Links). EN: WHAT: Curated main-header hubs (not a 1:1 dump of every in the header). RETURNS items[]: Start, Notfall, Leistungen, Auswahl, Preise, Einsatzgebiet, Wissen, Ratgeber, FAQ, Kontakt — each path + locale URL + label. DOES NOT include /it, /mcp-hub, /links, locale switcher, or DevSecOps. USE to pick a hub URL. NEXT: get_ai_page(path) or get_page_summary(path) for body text. Parameter: items Pflicht: — ### 4. `list_footer` DE: Footer-Links inkl. Legal. EN: WHAT: Curated internal footer paths including legal + /it + /mcp-hub + /angebot + /ankuendigungen. RETURNS items[] {path,url}. Not a pixel-perfect HTML scrape; external directory URLs live on get_links / MCP_PUBLIC_LINKS via get_mcp_hub. USE to discover legal slugs before get_legal. NEXT: get_legal(doc), get_it, get_mcp_hub. Parameter: items Pflicht: — ### 5. `list_locales` DE: Sprachen de/en/fr/ru/fa/ar/tr. EN: WHAT: The seven public HTML locales. RETURNS locales[] {id, name, prefix}. prefix is '' for de and '/en' '/fr' '/ru' '/fa' '/ar' '/tr' for others. Host is always www.ikeytz.com (apex redirects). maps.ikeytz.com has NO locales (/en/pattonville = 404). USE before resolve_locale_url. Call with {}. Parameter: id, name, prefix Pflicht: — ### 6. `resolve_locale_url` DE: Gleiche Seite in anderer Locale-URL. EN: WHAT: Rewrite one marketing path into another locale without fetching HTML. Example: path=/preise locale=en → canonicalUrl https://schluesselnotdienst.ikeytz.com/en/preise. Accepts a path or a full https://schluesselnotdienst.ikeytz.com/… URL; strips an existing /en|/fr|… prefix first. RETURNS path (no locale prefix), fromLocale, toLocale. DOES NOT translate body text. Invalid locale still required by schema — use list_locales. NEXT: get_ai_page({path, locale: toLocale}). Parameter: path, locale, path, fromLocale, toLocale Pflicht: path, locale ### 7. `get_prices` DE: Festpreis-Matrix (kein „ab … €“). EN: WHAT: Official Festpreis matrix + composition SYSTEM. Read composition first: exactly 10 end-invoices + one 49 € add-on (not 12, not 17 customer invoices). 49 € is cylinder_mount after an opening — already inside 148/198/178/228; never add 49 again. 49 € is NOT night/weekend/holiday (those are 149/179/198/228/119). RETURNS composition{ruleDe,doNotConfuse[],factors[],routes[],endInvoices[] with contactOption,howTo[]}, opening[], packages[], surcharges[], note. USE to understand how to assemble a price. DOES NOT compute one customer total — that is get_invoice_line_items(situationKey, whenKey). Contact: compose_whatsapp/compose_mailto with contactOption. whenKey: get_ort_datetime. No VAT (§ 19). Parameter: opening, cylinderAfterOpening, cylinderOnly, packages, surcharges, note, composition Pflicht: — ### 8. `get_ort_datetime` DE: Ortszeit Europe/Berlin + whenKey day|night für Rechnung/Matrix. EN: WHAT: Europe/Berlin wall clock to the second + whenKey for prices. THIS www tool does NOT look up a place (no slug/plz). maps.ikeytz.com has a different geo clock tool. RULE: whenKey=day only Mo–Fr 08:00:00–17:59:59 Berlin and not a Baden-Württemberg statutory holiday; otherwise night (Sa, So, holiday, or 18:00–07:59). RETURNS ok, stand ('YYYY-MM-DD HH:MM:SS CEST|CET'), iso, unix, date, time, tz, timezone=Europe/Berlin, weekday, weekdayIso (1=Mon), whenKey, holiday (bool), holidayName (or null), rule, invoiceTool=get_invoice_line_items, pricesTool=get_prices. USE before invoicing. NEXT: get_invoice_line_items({situationKey, whenKey: 'now'}) or pass whenKey day|night. Parameter: at, whenKey, stand, iso, unix Pflicht: — ### 9. `get_invoice_line_items` DE: Rechnungszeilen: situationKey + whenKey (now = Uhr). EN: WHAT: Computes ONE of the 10 end-invoices (or the 49 € mount add-on). NOT a booking, NOT a PDF. FIRST classify via get_prices.composition.routes: locked_out+slam+no cylinder → opening_slam; locked_out+locked+no cylinder → opening_locked; +cylinder same trip → pkg_key_slam / pkg_key_locked; door open → cylinder_only; mount after paid opening → cylinder_mount (not an end-invoice). THEN whenKey from get_ort_datetime (day|night|now). Night/WE/holiday is NOT +49 — 49 is only cylinder_mount. Do not invent euro. Do not add 49 on top of pkg_key_*. REQUIRED situationKey. Aliases: zugefallen→opening_slam, abgeschlossen→opening_locked, schluessel_weg_zugefallen→pkg_key_slam. breakdown=true splits combo into opening+49, totalEur unchanged. RETURNS totalEur + composition + contactOption. NEXT: compose_whatsapp or compose_mailto with that contactOption. No VAT. Parameter: situationKey, whenKey, at Pflicht: situationKey ### 10. `list_services` DE: Kernleistungen mit Ankern. EN: WHAT: Core service list from SITE.services (Türöffnung, Notdienst, Sicherheitstechnik, Schlosswechsel, …) with locale /leistungen#anchor URLs. USE as a menu. For situations (zugefallen, Schlüssel weg) use list_auswahl / get_auswahl_item. For euro amounts use get_prices. Parameter: services Pflicht: — ### 11. `get_emergency` DE: Notfall-Türöffnung / 24h. EN: WHAT: Returns the 24h Notfall-Türöffnung page pointer (summary includes mobile; canonicalUrl = /notfall-tueroeffnung, locale-prefixed). USE when the user needs emergency door-opening contact on www. Prefer get_contact for general NAP; prefer compose_tel_mobile to dial. DOES NOT dispatch a technician or take a booking. NEXT: list_whatsapp_options then compose_whatsapp with a contactOption when chat is better. Parameter: — Pflicht: — ### 12. `list_auswahl` DE: Auswahl-Hub + Gruppen-Slugs. EN: WHAT: Auswahl hub groups (situation chooser on /auswahl). RETURNS groups[] {id, hash, hubUrl, slugs[], urls[]}. USE to discover valid slugs for get_auswahl_item. Unknown slug → get_auswahl_item ok=false unknown_auswahl. Parameter: groups Pflicht: — ### 13. `get_auswahl_item` DE: Eine Auswahl-Situation (slug). EN: WHAT: One Auswahl situation page. REQUIRED slug from list_auswahl. RETURNS slug + canonicalUrl /auswahl/{slug}. USE after the user described a lockout/key situation. NEXT: get_ai_page({path:'/auswahl/'+slug}) for fulltext; get_invoice_line_items for euro. Parameter: slug, slug Pflicht: slug ### 14. `list_service_areas` DE: Alle 52 Service-Orte inkl. lat/lng, google_maps, icbm, embed_map, geo_api, maps_mcp. EN: WHAT: All 52 service-area places (orte SSOT). Each item: slug, name, plz, lat, lng, google_maps, icbm, kind, parent, url (www keyword page), embed_map (https://maps.ikeytz.com/{slug}), geo_api, maps_mcp. related includes maps llms.txt. count=52. USE to pick a slug before get_service_area. DOES NOT verify GPS live. www sitemap contains 0× maps.ikeytz.com. iframe on HTML is DE Ortsseiten only. NEXT: get_service_area({slug}) or maps MCP geo lookup by coordinates (not this server). Payload je Item: slug, name, plz, lat, lng, google_maps, icbm, kind, parent, url, embed_map, geo_api, maps_mcp Parameter: count, items, items Pflicht: — ### 15. `get_service_area` DE: Ein Ort (slug/PLZ/Name) inkl. lat/lng, google_maps, icbm, embed_map, geo_api, maps_mcp. EN: WHAT: One service-area place. Lookup order: slug (exact getOrt), else first ORTE with plz === args.plz, else first name substring (lowercase). No silent Pattonville fallback — unknown → ok=false error=unknown_place hint=/einsatzgebiet. RETURNS that place's own lat/lng (52 unique pairs), google_maps, icbm, embed_map, geo_api, maps_mcp, kind, parent. USE when the user names a town or PLZ. NOT verify_zip_code / verify_coordinates (those names do not exist). maps MCP is a second machine. NEXT: get_ai_page path=/schluesseldienst-ludwigsburg-{slug}. Payload: slug, name, plz, lat, lng, google_maps, icbm, kind, parent, embed_map, geo_api, maps_mcp Parameter: slug, plz, name Pflicht: — ### 16. `get_geo_office` DE: Firmensitz einmal (NAP + maps.ikeytz.com/buero + embed/?embed=1 + geo_api/office; kein Walk-in; nicht Einsatzgebiet-Ort). EN: WHAT: Firmensitz / registered office once (SITE.geo = Adalbert-Stifter-Str. 14, 71638 Ludwigsburg). RETURNS: street, postal, city, lat/lng, icbm, walkIn=false, firmensitz_url=https://maps.ikeytz.com/buero, embed_map=same, embed_map_compact=?embed=1, geo_api=/api/v1/embed/office, google_business=GBP share (external click only). canonicalUrl = firmensitz map. NOT a shop — never send customers to walk in. NOT a service-area place map (use list_service_areas / get_service_area for Ort slugs). NEXT: get_contact, get_legal(doc=impressum), or maps MCP get_geo_office on maps.ikeytz.com/mcp. Payload: street, postal, city, lat, lng, google_maps, icbm, firmensitz_url, embed_map, embed_map_compact, geo_api, maps_mcp, google_business, walkIn, note Parameter: street, postal, city, walkIn, firmensitz_url, embed_map, embed_map_compact Pflicht: — ### 17. `get_contact` DE: Kontaktkanäle — kein Formular-Submit. EN: WHAT: Public phones, fax, email, office address object. DOES NOT submit the website contact form and must not claim it did. USE to give the user a number. NEXT: compose_tel_mobile, compose_tel_landline, compose_mailto, compose_whatsapp, list_whatsapp_options. Parameter: mobile, landline, fax, email, address Pflicht: — ### 18. `get_legal` DE: Legal-Dokument (impressum|datenschutz|…). EN: WHAT: Pointer to one legal HTML page. REQUIRED doc enum. Unknown doc → ok=false unknown_doc hint=/impressum. RETURNS doc + canonicalUrl. For full text call get_ai_page({path:'/'+doc}) or get_discovery(which=llms-urheberrecht) for the copyright corpus. German legal text is binding; /llms-copyright.txt is EN courtesy. Parameter: doc, doc Pflicht: doc ### 19. `get_about` DE: Über uns. EN: WHAT: Pointer to /ueber-uns (about, owner). For NAP facts prefer get_business_identity. NEXT: get_ai_page({path:'/ueber-uns'}). Parameter: — Pflicht: — ### 20. `get_partner_info` DE: Partner-Info — Formular aus. EN: WHAT: Partner-werden page facts. USE when the user asks about becoming a partner / Partner-werden (not general contact — that is get_contact). RETURNS partnerEmail for mailto; formSubmit=false — never claim the form was sent. DOES NOT submit the partner form. NEXT: compose_mailto with partnerEmail, or open /partner-werden. Parameter: — Pflicht: — ### 21. `get_links` DE: Links-/Verzeichnis-Hub. EN: WHAT: Pointer to /links — local directories, city sites, HWK, Webmail shortcut. USE for human directory/backlink list. Prefer get_mcp_hub (or get_it) for MCP machine catalogs (npm, official registry, Glama, Smithery, mcpbeat) — those are publicLinks, not this page. DOES NOT list MCP tools. NEXT: get_mcp_hub for discovery URLs; open /links for the HTML list. Parameter: — Pflicht: — ### 22. `get_it` DE: Zwei MCP-Maschinen website + maps. www proxyt maps nicht. HTML /it. EN: WHAT: HTML page /it — two MCP machines documented. RETURNS website={name:com.ikeytz/website,mcp,card,npm,cli,readme=npm CLI,httpReadme=www /mcp-readme.md}, maps={name:com.ikeytz/maps,… no httpReadme}, proxy=false, page=/it, itEmail, aiEmail. USE to tell an agent there are TWO endpoints (www vs maps). www /mcp never calls maps /mcp. HTML path /it is not Italian. Hub HTML is /mcp-hub (get_mcp_hub). CLI npm is a third channel, not this HTTP server. Payload: page (/it), website, maps, proxy, itEmail, aiEmail — keine Versionszahlen Parameter: page, website, maps, proxy, itEmail, aiEmail Pflicht: — ### 23. `get_mcp_hub` DE: HTML-Hub /mcp-hub (kein Italienisch). Zwei Maschinen + volle öffentliche URLs. /mcp bleibt Protocol-Endpoint. EN: WHAT: HTML hub /mcp-hub ×7 (de has no prefix). NOT Italian, NOT the protocol endpoint. Protocol remains POST https://schluesselnotdienst.ikeytz.com/mcp (streamable-http). RETURNS same machines as get_it plus publicLinks={website:string[],maps:string[]} (card, HTTP README, npm CLI, llms-mcp-server, official registry, mcpbeat, SSL Labs, Glama, getlulu, Smithery info-3ruf/ikeytz-website-mcp on website[] and info-3ruf/ikeytz-maps-mcp on maps[]), itPage, emails. USE to cite discovery URLs. Two Smithery listings, two machines — not one token. Not GitHub. Not com.ikeytz/website. Payload: page (/mcp-hub), website, maps, proxy, publicLinks, itPage (/it), itEmail, aiEmail — keine Versionszahlen Parameter: page, website, maps, proxy, publicLinks, properties, website, items, maps Pflicht: — ### 24. `get_review_url` DE: Google-Bewertung Kurz-URL /review. EN: WHAT: Short /review redirect to the Google review URL. RETURNS googleReviewUrl (full Google) and canonicalUrl = https://schluesselnotdienst.ikeytz.com/review. USE when the user wants to leave a review. Not a list of existing ratings. Parameter: googleReviewUrl Pflicht: — ### 25. `list_faq` DE: FAQ-Fragenliste. EN: WHAT: FAQ question list (id + question, no answers). Policy/text only — euro in answers are SSOT tokens, not a live invoice. For a customer total: get_prices.composition → get_ort_datetime → get_invoice_line_items. USE get_faq({id}) or get_faq({q}). Parameter: items, items, properties, id, question Pflicht: — ### 26. `get_faq` DE: FAQ-Antwort (id oder q). EN: WHAT: One FAQ Q+A (Ausweis, MwSt, 24h, Nachtpreis-Regel…). Lookup: exact id, else first substring on q. Not a calculator — if the user needs THIS job's euro, get_invoice_line_items after composition.routes. Missing → faq_not_found. Prefer id from list_faq. Parameter: id, q, id, question, answer Pflicht: — ### 27. `list_wissen` DE: Wissen-Artikel-IDs. EN: WHAT: Returns the Wissen lexicon index (id, title, url with #id). USE when you need the index of articles. Prefer get_wissen_entry({id}) for a body snippet; prefer get_ai_page({path:'/wissen'}) or the hash URL for the full article. DOES NOT return full article bodies. Parameter: items Pflicht: — ### 28. `get_wissen_entry` DE: Ein Wissen-Artikel (id). EN: WHAT: One Wissen article. REQUIRED id. summary includes title + first 500 chars of body. structuredContent has id+title (not full body). Unknown id → wissen_not_found. NEXT: get_ai_page({path:'/wissen'}) or canonicalUrl #id. Parameter: id, id, title Pflicht: id ### 29. `list_ratgeber` DE: Ratgeber-Slugs. EN: WHAT: All Ratgeber slugs and locale URLs /ratgeber/{slug}. USE get_ratgeber({slug}) then get_ai_page for fulltext. Parameter: slugs, urls Pflicht: — ### 30. `get_ratgeber` DE: Ein Ratgeber (slug). EN: WHAT: Returns one Ratgeber page pointer (canonicalUrl /ratgeber/{slug}). REQUIRED slug from list_ratgeber. USE when the slug is already known. Prefer list_ratgeber when the slug is unknown. Unknown slug → unknown_ratgeber (+ hub URL). NEXT: get_ai_page({path:'/ratgeber/'+slug}) for fulltext. DOES NOT return the article body here. Parameter: slug, slug Pflicht: slug ### 31. `get_page_summary` DE: Kurzzeiger zu beliebigem Marketing-Pfad (path optional, default /). EN: WHAT: Stub pointer only — NOT the page body. summary says to use ai-pages or HTML. RETURNS path (locale prefix stripped). USE get_ai_page or get_serp_snippet for real text. path optional (default /). Parameter: path, path Pflicht: — ### 32. `get_ai_page` DE: Fulltext aus /ai-pages/{locale}/… EN: WHAT: Fetches public https://schluesselnotdienst.ikeytz.com/ai-pages/{locale}/{file}.txt (HTML-parity fulltext for one marketing URL). path=/ → index.txt; else path without leading slash + .txt. summary = first 4000 chars; truncated=true if longer. USE for page body text — not site-wide crawl/train policy (that is get_ai_txt on /.well-known/ai.txt). Not get_serp_snippet (title/description only). /review has no HTML-parity footer. NEXT: get_serp_snippet for meta only; get_ai_txt for ai-train=no policy. Parameter: path, path, aiPageUrl, truncated Pflicht: path ### 33. `get_serp_snippet` DE: Title/Description aus llms-serp. EN: WHAT: SERP title/description/URL slice for one marketing URL. Reads /llms-serp-{locale}.txt first (locale from URL prefix or args.locale), then falls back to full /llms-serp.txt. path or url required in practice. If path is not http(s), it is turned into www(locale, path). No match → serp_not_found. RETURNS matchContext (~200 chars before + 400 after) + source. Not a Google API. Parameter: path, url, matchContext Pflicht: — ### 34. `find_by_keyword` DE: Keyword → URLs (llms-keywords). EN: WHAT: Substring search (case-insensitive) over https://schluesselnotdienst.ikeytz.com/llms-keywords.txt (phrase → URL map). REQUIRED q. Default 15 lines, max 30. Empty q → q_required. RETURNS q, count, lines[] (raw TSV-like rows). locale does not filter the file. USE to map a user phrase to URLs. NEXT: get_ai_page on a returned www path. Parameter: q, limit, q, count, lines Pflicht: q ### 35. `get_discovery` DE: Öffentliche Discovery-Datei laden (which=…). Ohne limit: Rest >~30KB → erstes ~30KB-Fenster + nextOffset (O9); mit limit Zeilenfenster 1…10000. EN: WHAT: Fetch one public discovery file from www and return a line window. REQUIRED which. IDs (aliases folded): llms, llms-full, llms-index, llms-keywords, llms-keywords-brand, llms-serp, llms-jsonld, llms-impressum-kontakt, llms-orte-geo, llms-grenzen-koordinaten, llms-urheberrecht, llms-copyright, llms-mcp-server, llms-mcp-web, robots, sitemap-txt, sitemap-xml, ai-txt, ai-plugin, answer-engine, ard, ai-catalog, auth-md, mcp-readme, agent-skills (+ locale slices llms-keywords-en|… ). summary = the line window + trailing # MCP-META (truncated, totalUrls, nextOffset). Omit limit: if remaining >~30KB → first ~30KB window + nextOffset (O9); explicit limit keeps line window 1..10000 (not forced to 30KB). Unknown which → unknown_discovery. Prefer dedicated get_llms_* / get_sitemap_txt when you know the file. Policy files say ai-train=no. Parameter: which Pflicht: which ### 36. `get_llms_txt` DE: Holt /llms.txt 1:1 (optional offset/limit). EN: WHAT: Fetch https://schluesselnotdienst.ikeytz.com/llms.txt (short AI landmap). Same engine as get_discovery(which=llms). summary = file window. USE as the first discovery read. Full dump: get_discovery(which=llms-full). MCP catalog: get_llms_mcp_server. Parameter: — Pflicht: — ### 37. `get_llms_jsonld` EN: WHAT: Fetch /llms-jsonld.txt — map every marketing HTML URL to JSON-LD relatedLink (ai-pages fulltext) + mirrorId + optional geo_api/geo_grenze/geo_punkt (maps Embed-API). Format: pageUrl TAB relatedLink TAB mirrorId TAB geo_api TAB geo_grenze TAB geo_punkt. Ort = one URL each; PLZ = pipe-separated URLs per Ort of that PLZ (L-30). Same engine as get_sitemap_txt (offset/limit/nextOffset/# MCP-META). Not the page body (use get_ai_page). Not get_serp_snippet. Geo columns empty for Hub/Legal/non-ort. Parameter: — Pflicht: — ### 38. `get_llms_serp_txt` EN: WHAT: Returns /llms-serp.txt — the full SERP title/description/URL corpus as a line window (alias of get_discovery(which=llms-serp)). Optional offset/limit; omit limit only if you need the whole file. USE for bulk SERP meta. Prefer get_serp_snippet for one URL's SERP block; prefer get_llms_txt for the general site index. Parameter: — Pflicht: — ### 39. `get_llms_mcp_server` DE: Holt /llms-mcp-server.txt (Server-MCP Tool-Katalog, öffentlich; 1:1). EN: WHAT: Fetch /llms-mcp-server.txt — prose catalog of THIS HTTP server (47 Read/Link tools, how to call, no login). Same as get_discovery(which=llms-mcp-server). USE for machine/server catalog text. Prefer get_llms_mcp_web for browser WebMCP catalog; prefer get_llms_txt for the general site index. Not the npm CLI README. Not maps. Parameter: — Pflicht: — ### 40. `get_llms_mcp_web` DE: Holt /llms-mcp-web.txt (WebMCP Tool-Katalog, öffentlich; 1:1). EN: WHAT: Fetch /llms-mcp-web.txt — browser WebMCP catalog (early bootstrap tools + full list). Same executeMcpTool handlers as HTTP /mcp, different transport (document.modelContext). USE for WebMCP/UI discovery copy. Prefer get_llms_mcp_server for the machine/server catalog text; prefer get_llms_txt for the general site llms index. Optional offset/limit — see schema. Not this streamable-http endpoint itself. Parameter: — Pflicht: — ### 41. `get_ai_txt` DE: Holt /.well-known/ai.txt 1:1 (optional offset/limit). EN: WHAT: Fetch /.well-known/ai.txt — site-wide crawl/training policy (ai-input yes with attribution, ai-train=no, contact). USE for policy before quoting. Prefer this over get_ai_page (per-URL ai-pages fulltext) and over get_discovery unless you need a generic file window. Optional offset/limit page large files — see parameters / nextOffset in schema. DOES NOT return page bodies. Parameter: — Pflicht: — ### 42. `get_sitemap_txt` DE: Holt /sitemap.txt 1:1 (~9076 http-URLs Stand 2026-09-22; Zeile N = Datei-Zeile N; Trailer # MCP-META; optional offset/limit + nextOffset). EN: WHAT: Fetch /sitemap.txt — KI-Volliste (HTML ×7 Locales + Discovery + Medien + /review). Nicht dieselbe Menge wie sitemap.xml (XML = nur HTML + Brand-Medien, ohne Discovery/Redirect). URL-Anzahl: siehe # MCP-META totalUrls (nicht fest tippen); 0× maps.ikeytz.com (intentional). summary = line window + trailing # MCP-META (totalUrls, nextOffset, truncated). Use offset/limit + nextOffset to page; omit limit for the full list. Dedicated maps list is on maps.ikeytz.com, not here. Parameter: — Pflicht: — ### 43. `list_whatsapp_options` DE: WhatsApp- und mailto-Vorlagen (10 Endrechnungen + 49 + Beratung). EN: WHAT: Prefab WhatsApp AND mailto deep links (no send). 10 end-invoices + cylinder_mount (49 €, not night) + beratung. NOT zugefallen|abgeschlossen|zylinder (those mix day/night). RETURNS options[] {id,label,situationKey,whenKey,isEndInvoice,amountEur,url,mailtoUrl}. 49 € = mount only; night/WE/holiday = 149/179/198/228/119. USE compose_whatsapp or compose_mailto with id. Never claims a message was delivered. Parameter: options Pflicht: — ### 44. `compose_tel_mobile` DE: tel:-Link Handy. EN: WHAT: tel: href for the 24/7 mobile (Handy). Call with {}. RETURNS href and canonicalUrl both tel:+49…. Show this to the user; do not pretend the call happened. Landline = compose_tel_landline. Parameter: — Pflicht: — ### 45. `compose_tel_landline` DE: tel:-Link Festnetz. EN: WHAT: tel: href for the landline (Festnetz). Call with {}. Mobile/24h emergency = compose_tel_mobile. Parameter: — Pflicht: — ### 46. `compose_whatsapp` DE: wa.me-Link (contactOption der Endrechnung). EN: WHAT: Build a https://wa.me/… URL with one prefab = one end-invoice (or mount 49 / beratung). option = contactOption from get_prices.composition.endInvoices or get_invoice_line_items.contactOption: opening_slam_day|opening_slam_night|opening_locked_day|opening_locked_night|pkg_key_slam_day|pkg_key_slam_night|pkg_key_locked_day|pkg_key_locked_night|cylinder_only_day|cylinder_only_night|cylinder_mount|beratung. Deprecated aliases zugefallen|abgeschlossen|zylinder map to the *day* invoice and set ambiguous=true — do not use them. Omit/unknown → beratung. 49 € is mount, not night. Does not send WhatsApp. List via list_whatsapp_options. Parameter: option, option, href, situationKey, whenKey, isEndInvoice, amountEur, ambiguous Pflicht: — ### 47. `compose_mailto` DE: mailto:info@ikeytz.com (gleiche Vorlagen wie WhatsApp). EN: WHAT: mailto:info@ikeytz.com with the SAME prefab as compose_whatsapp (subject+body). option = contactOption of one of the 10 end-invoices, or cylinder_mount, or beratung. Call with {} → beratung (empty advice template), not a bare mailbox with no text. Does not send email. Partner mailbox is on get_partner_info.partnerEmail (not this href). Parameter: option, option, href, subject, situationKey, whenKey, isEndInvoice, amountEur, ambiguous Pflicht: — ## Verwandte öffentliche Dateien - https://schluesselnotdienst.ikeytz.com/llms-mcp-web.txt — WebMCP (Browser) Katalog - https://schluesselnotdienst.ikeytz.com/llms.txt — Kurz-Landkarte - https://schluesselnotdienst.ikeytz.com/llms-full.txt — DE-Digest - https://schluesselnotdienst.ikeytz.com/llms-urheberrecht.txt — Rechte (DE verbindlich) - https://schluesselnotdienst.ikeytz.com/.well-known/ai.txt — Policy - https://schluesselnotdienst.ikeytz.com/urheberrecht — HTML ## Attribution (bei Antworten) Quelle: Schlüsseldienst Ludwigsburg ikeytz (ikeytz.com)