Ein MCP-Server, der die Frage beantwortet, die Menschen in einem Hausnotfall tatsächlich stellen: welches Gewerk rufe ich?
Nicht „finde mir einen Klempner.“ Die vorherige Frage — ist das überhaupt ein Klempner?
Ich habe einen Klempner gerufen, der mir sagte, ich solle einen Gerätetechniker rufen. Ich habe einen Gerätetechniker gerufen, der mir sagte, ich solle einen Klempner rufen.
Das ist die normale Erfahrung bei einem Hausnotfall, und sie ist teuer. Zwei Anfahrtskosten, ein halber Tag verloren, und das Wasser kommt immer noch durch die Decke. Die Gewerke sind spezialisiert und ihre Grenzen sind von außen unsichtbar: Ein Fleck an deiner Decke ist ein Dachdecker, wenn er vom Wetter abhängt, ein HLK-Techniker, wenn er mit der Klimaanlage zusammenhängt, und ein Klempner, wenn er von der Dusche oben stammt. Wasserschaden nach dem Stoppen des Lecks ist ein viertes Gewerk — ein Trocknungsteam — und ein Klempner wird dir das sagen, nachdem du den Einsatz bezahlt hast.
Dieser Server gibt einem Assistenten die Routing-Tabelle. Beschreibe das Problem in einfachen Worten; du erhältst zurück, wen du anrufen sollst, in welcher Reihenfolge, wofür du sie nicht anrufen sollst, was es typischerweise kostet, mit einer datierten Quelle, und — wo vorhanden — eine 24/7-Einsatzleitung.
Die Inhalte stammen aus der öffentlichen JSON-API unter whodoicallfor.com. Keine Authentifizierung, kein API-Schlüssel, keine Konfiguration.
Related MCP server: kickserv-mcp
Claude Code
claude mcp add whodoicallfor -- npx -y whodoicallfor-mcpCursor — füge zu ~/.cursor/mcp.json hinzu (oder .cursor/mcp.json in einem Projekt):
{
"mcpServers": {
"whodoicallfor": {
"command": "npx",
"args": ["-y", "whodoicallfor-mcp"]
}
}
}Dasselbe Format funktioniert für Claude Desktop (claude_desktop_config.json), Windsurf, Zed und alles andere, das stdio-MCP spricht.
Oder starte nichts. Eine gehostete Instanz dieses Servers ist live:
https://whodoicallfor.com/api/mcp/Sie spricht streambares HTTP, ist zustandslos und benötigt keine Anmeldedaten. Der abschließende Schrägstrich ist erforderlich — die URL ohne Schrägstrich antwortet mit einer 308-Weiterleitung, und mehrere MCP-Clients folgen einer Weiterleitung bei einem POST nicht.
claude mcp add --transport http whodoicallfor https://whodoicallfor.com/api/mcp/npx whodoicallfor-mcp # stdio (default)
npx whodoicallfor-mcp --http # streamable HTTP on http://localhost:3000/mcp
npx whodoicallfor-mcp --http 8080 # ...on a port you pickErfordert Node.js 20.19 oder neuer.
Alle drei sind schreibgeschützt, idempotent und berühren nichts außerhalb der öffentlichen API.
Werkzeug | Argumente | Rückgabe |
|
| Bis zu 3 bewertete Szenarien, beste zuerst |
|
| Ein Szenario in voller Länge |
| — | Jedes veröffentlichte Szenario mit kanonischen URLs |
route_emergencyProblem in einfacher Sprache rein, bewertete Szenarien raus. Gib einen US-Bundesstaatscode oder -namen an ("FL", "Florida"), um die Einsatznummern auf Leitungen zu filtern, die diesen Staat tatsächlich abdecken.
Eingabe
{ "problem": "water coming through my ceiling" }Ausgabe (gekürzt — die echte Antwort enthält drei Treffer mit vollständigen costs-, faq- und sources-Arrays)
{
"matches": [
{
"title": "Water Leaking From the Ceiling — Who Do I Call?",
"category": "water",
"vertical": "water-damage",
"tldr": "What's directly above matters less than timing. Appears or worsens when it rains → roofer. Appears when the AC runs, or only in cooling season → HVAC condensate line. Neither, and it tracks water use upstairs → plumber. If water is still coming, shut the main valve, keep it clear of lights and outlets, and photograph everything before any repair. Renting? The landlord's emergency line comes before any of this.",
"who_to_call": [
{
"who": "Emergency plumber",
"when": "Water is actively dripping or flowing and you can't stop it at the source",
"not_for": "The leak has stopped — a plumber fixes pipes, not soaked drywall"
},
{
"who": "Roofer",
"when": "The stain appears or worsens when it rains and stays quiet in dry weather — usually failed flashing",
"not_for": "The stain tracks showers, washes, or flushes rather than the weather"
}
],
"costs": [
{
"item": "Water-damaged ceiling repair",
"range": "$200–$1,500",
"source": "Fixr",
"source_url": "https://www.fixr.com/costs/ceiling-repair"
}
],
"updated": "2026-08-12",
"url": "https://whodoicallfor.com/water/ceiling-leaking/",
"recommended_call": null,
"match_score": 12
}
],
"state": null
}Das Feld not_for ist der Teil, der die verschwendete Anfahrtsgebühr erspart.
Wenn nichts hoch genug bewertet wird, erhältst du ein leeres matches-Array plus eine note, den Kategorie-Index und einen Link zur geführten Triage — statt einer selbstbewusst falschen Antwort.
get_scenarioRuft ein Szenario vollständig ab, unter Verwendung von category und slug aus route_emergency oder list_scenarios.
{ "category": "water", "slug": "ceiling-leaking" }Gibt dasselbe strukturierte Objekt zurück, plus body_markdown (die vollständige Schritt-für-Schritt-Anleitung), wenn die übergeordnete API es bereitstellt.
list_scenariosKeine Argumente. Gibt { site, count, scenarios[] } zurück — jede veröffentlichte Situation mit Titel, Beschreibung, Kategorie, Slug, updated-Datum, kanonischer URL und JSON-URL.
route_emergency sortiert nach gewichteter Token-Überschneidung zwischen deinem Problem und vier Feldern jeder Situation: Titel ×3, Beschreibung ×2, tldr ×1 und who/when jedes who_to_call-Schritts ×1. Token werden klein geschrieben, von Satzzeichen befreit, gegen eine Stoppwortliste gefiltert und suffixgefaltet, sodass „leaking“ zu „leak“ passt und „pipes“ zu „pipe“. Alles unter 3 Punkten wird verworfen; höchstens drei Treffer kommen zurück.
tldr und who_to_call leben in den seitenweisen Dokumenten und nicht im Index, daher wärmt der erste route_emergency-Aufruf einen vollständigen In-Memory-Cache (begrenzte Parallelität, ~2s kalt). Nachfolgende Aufrufe dauern einstellige Millisekunden. Der Index wird 10 Minuten lang gecacht und Seiteninhalte eine Stunde lang, sodass der Server die Website nie überlastet.
Es gibt kein unscharfes Matching, kein Einbettungsmodell und keinen Netzwerkaufruf an irgendetwas außer whodoicallfor.com. Es ist eine kleine deterministische Funktion, die man in einem Sitzen lesen kann: src/match.ts.
Zwei Dinge, die man klar ansprechen sollte.
recommended_call-Einträge sind gesponsert. Es sind bezahlte Pay-per-Call-Leitungen, und jede ist im Payload mit "sponsored": true gekennzeichnet. Wenn eine Situation einen solchen Eintrag enthält, ist das der kommerzielle Motor hinter der Website. Behandle ihn wie eine Anzeige mit einer Telefonnummer und zeige ihn den Nutzern auch so — genau dafür ist das Flag in den Daten.
Lebensrettende Nummern werden niemals gesponsert. 911, Giftnotruf und Gasnotrufleitungen enthalten überhaupt kein recommended_call — es gibt eine harte serverseitige Schutzmaßnahme, die es für Gasszenarien und für jede Situation ohne kommerzielle Sparte entfernt. Wenn jemand Gas riecht, lautet die Antwort „Rausgehen und den Versorger von draußen anrufen“, und niemand hat für diese Platzierung bezahlt. Diese Schutzmaßnahme ist keine politische Präferenz; sie ist Code, upstream, und dieser Client kann sie nicht außer Kraft setzen.
Kostenangaben werden mit einer benannten Quelle und einem Zugriffsdatum zitiert. Überprüfe alles, auf dessen Grundlage du handeln willst.
npm install
npm run typecheck
npm run build
npm start # stdio
npm run start:http # http on :3000src/index.ts entry point: tool definitions, stdio + streamable HTTP transports
src/client.ts cached fetches against the public JSON endpoints
src/match.ts token-overlap rankingUS-orientiert. Etwa 40 Szenarien zu Wasser und Überschwemmung, Gas und Kohlenmonoxid, Strom und Elektrik, Heizung und Kühlung, Haushaltsgeräten, Schädlingen und Wildtieren, Ausgesperrtsein, Autopannen sowie Streitigkeiten mit Vermietern und Auftragnehmern. list_scenarios spiegelt immer wider, was tatsächlich veröffentlicht ist.
MIT © Daniel Widmonte
Fetch one situation in full: structured answer (who to call, costs, FAQ, sources) plus the complete step-by-step guide as markdown. Use category and slug values from route_emergency or list_scenarios.
| Name | Required | Description | Default |
|---|---|---|---|
| slug | Yes | Scenario slug, e.g. "ceiling-leaking" | |
| category | Yes | Category slug, e.g. "water" |
| Name | Required | Description |
|---|---|---|
| faq | No | |
| url | Yes | |
| tldr | Yes | |
| costs | No | |
| title | Yes | |
| sources | No | |
| updated | No | |
| category | Yes | |
| json_url | No | |
| vertical | No | |
| description | No | |
| who_to_call | No | |
| body_markdown | No | The full step-by-step guide as markdown |
| recommended_call | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already indicate readOnlyHint=true, idempotentHint=true, destructiveHint=false. Description adds useful behavioral context by specifying the return format (structured answer + markdown guide). However, it doesn't mention performance characteristics or any dependencies on other tool calls if inputs are not from siblings.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Description is concise (two sentences) and front-loaded with the primary purpose. Every word adds value - the first sentence explains what is returned, the second tells where inputs come from. No redundant language.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the low complexity (2 required string params, no enums, no nested objects), the presence of both input and output schemas, and comprehensive annotations, the description is complete. It clearly explains what the tool returns and how to properly call it, leaving no ambiguity for an AI agent.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema already covers both parameters with 100% coverage (category and slug). Description adds value by stating they are 'category slug' and 'scenario slug' with examples like 'water' and 'ceiling-leaking', which clarifies the role of each field beyond their type definitions.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Description clearly states it fetches one scenario with full details including structured answer, costs, FAQ, sources, and step-by-step guide as markdown. It distinguishes itself from siblings by specifying it retrieves a single scenario in full versus list_scenarios that lists multiple scenarios, and route_emergency which handles routing.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicitly instructs to use category and slug values from route_emergency or list_scenarios, providing clear context on when to invoke this tool and where to get the required input parameters. This avoids confusion with its sibling tools.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List every published situation with title, category, slug, and canonical URLs.
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
| Name | Required | Description |
|---|---|---|
| site | Yes | |
| count | Yes | |
| scenarios | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true, idempotentHint=true, and destructiveHint=false. The description adds that it lists 'every published situation' (a scope detail) and the returned fields. However, it does not discuss pagination, ordering, or potential edge cases (e.g., empty list). The behavioral disclosure is adequate but minimal beyond annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single sentence of 12 words, front-loaded with the verb 'List'. It conveys essential information (what is listed, what fields) without any extraneous content. Every word earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's simplicity (0 parameters, an output schema exists), the description is nearly complete. It states the scope ('every published situation') and the fields included. Minor missing context includes ordering, default listing behavior, and what happens if no situations exist, but the presence of an output schema likely covers the return structure.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The input schema has 0 parameters, so there is nothing to document. The description does not need to add parameter semantics. The baseline for 0 parameters is 4, and the description meets this without requiring further elaboration.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the action ('List') and the resource ('every published situation'), and specifies the output fields (title, category, slug, canonical URLs). This distinguishes it from siblings route_emergency (routing) and get_scenario (single scenario retrieval).
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives like get_scenario. It does not mention exclusions, prerequisites, or typical use cases. With sibling tools present, such context would be valuable.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Given a home emergency described in plain words (e.g. "water coming through my ceiling"), returns the matching situation(s): who to call and in what order, first steps, typical costs, and a 24/7 dispatch phone number when one is available for the service vertical. Optionally pass a US state for region-aware dispatch numbers.
| Name | Required | Description | Default |
|---|---|---|---|
| state | No | Optional US state code or name, e.g. "FL" or "Florida" | |
| problem | Yes | The user's problem in plain words |
| Name | Required | Description |
|---|---|---|
| note | No | |
| state | No | US state code the numbers were matched against |
| matches | Yes | Matching situations, best first; empty when no confident match |
| categories | No | |
| triage_url | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already provide readOnlyHint=true, idempotentHint=true, destructiveHint=false, so no contradiction. The description adds valuable behavioral context beyond annotations: it explains the return structure (who to call, order, first steps, costs, dispatch phone) and the conditional availability of 24/7 dispatch numbers. This is informative but does not cover edge cases like unrecognized problems or error handling.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two sentences, each earning its place. The first sentence front-loads the core functionality and output, while the second adds the optional parameter nuance. No fluff or redundancy.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the low complexity (2 params, annotations present, output schema exists), the description covers the essential aspects: input format, output contents, and optional state usage. It does not explain output schema shape (but that is handled by output schema), nor does it address potential failures or multi-match cases. Almost complete for the intended use.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100% with basic descriptions for both parameters. The description adds meaningful context for the 'state' parameter, noting it enables region-aware dispatch numbers, which the schema description does not convey. For 'problem', the description restates the schema's intent without adding detail, so the net improvement is moderate but solid.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool's purpose: it routes a home emergency by returning matching situations with who to call, order, first steps, costs, and dispatch phone. The verb 'route' is specific and the scope is well-defined. It implicitly distinguishes from siblings (get_scenario, list_scenarios) by focusing on actionable routing rather than simply fetching or listing scenarios.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage when a user describes a home emergency in plain words, and optionally when a US state is known. However, there is no explicit guidance on when not to use this tool versus alternatives (e.g., get_scenario might be preferred for retrieving a single scenario without dispatch info). The usage context is clear but lacks exclusions or differentiation.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v1.0.0get_scenariolist_scenariosroute_emergencyScored across 3 tools
Each tool serves a distinct role: routing from a natural language description, retrieving full details for a specific scenario, and listing all available scenarios. There is no functional overlap.
All tools follow a consistent verb_noun pattern using snake_case: route_emergency, get_scenario, list_scenarios. The naming is predictable and intuitive.
With only three tools, the server is tightly scoped to the home emergency reference purpose. Each tool earns its place and the count feels neither sparse nor excessive.
The tool surface covers the full workflow: discover available scenarios (list_scenarios), route from a plain-language emergency (route_emergency), and drill into full details (get_scenario). No obvious gaps for the intended use case.
MCP server for generating rough-draft project plans from natural-language prompts.
Search 33,000+ MCP servers by job, see safety grades and reviews, and call them from one endpoint.
ProxyLink MCP server for finding and booking home service professionals
An MCP server that audits the fairness of construction and renovation estimates in Japan. Provides fair-price ranges, overcharge detection, and verifiable unit-cost data based on JCCDB (65,520 items across 402 categories, CC BY 4.0, DOI-backed).
