Enables ChatGPT Desktop to compare cities for cost-of-living, quality-of-life, and safety-net values using the cityparity MCP server.
Cost-of-living and quality-of-life comparison tools that AI agents can call. Point an agent at two cities and it returns take-home pay, a full cost breakdown, the equivalent salary you'd need in the target city, and the non-cash deltas people actually move for: vacation, parental leave, universal healthcare.
Hosted at https://mcp.cityparity.com/mcp. Free, no API key, rate-limited at the edge. Built on the Model Context Protocol (spec 2025-06-18, Streamable HTTP).
Site: https://cityparity.com
Install docs (every client): https://cityparity.com/mcp
Machine-readable docs: https://mcp.cityparity.com/llms.txt
OpenAPI (REST mirror): https://mcp.cityparity.com/openapi.json
About 165 cities across 69 countries. Most calculators stop at salary and rent. cityparity prices the social safety net (childcare subsidies, parental leave, statutory vacation, universal healthcare), so "the lower-salary city actually pays more" stops being a hand-wave and becomes a number.
Related MCP server: CostAPI
Tool | What it does |
| Full scenario comparison between two cities: take-home, cost breakdown, equivalent target salary, lifestyle deltas, quality score. |
| Discover supported city slugs, grouped by country. |
| One-city profile: tax shape, headline costs, safety-net values. |
| Top N cities by composite quality score, with custom weights and region/country filters. |
| Parental leave, universal-healthcare flag, vacation, public holidays for 1 to 20 cities. |
| Inbound-worker regimes: Italy impatriati, Portugal IFICI, Belgium expat, Poland B2B ryczałt, Greece inbound. |
City slugs are kebab-case (san-francisco, hong-kong). Call list_cities first if you're unsure.
Most modern clients speak Streamable HTTP and need no install. Just point them at the URL.
claude mcp add --transport http cityparity https://mcp.cityparity.com/mcp{
"mcpServers": {
"cityparity": {
"url": "https://mcp.cityparity.com/mcp"
}
}
}Claude Desktop config lives at ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows). ChatGPT Desktop uses …/OpenAI/ChatGPT/mcp.json. Restart the app after editing.
{
"name": "cityparity",
"url": "https://mcp.cityparity.com/mcp"
}mcpServers:
cityparity:
url: https://mcp.cityparity.com/mcpIf your client can't do HTTP MCP yet, the npm package in this repo bridges stdio → HTTP:
{
"mcpServers": {
"cityparity": {
"command": "npx",
"args": ["-y", "cityparity-mcp"]
}
}
}cityparity-mcp is a ~150-line stdio→HTTP forwarder. It reads line-delimited JSON-RPC from stdin, POSTs each line to the MCP endpoint, and writes the responses back to stdout (streaming SSE line by line). That's the whole job: no data, no calc engine, no secrets. Read bin/cityparity-mcp.mjs if you want to verify.
npm install -g cityparity-mcp # or use npx, as aboveEnv var | Default | Purpose |
|
| Override the upstream endpoint (local dev) |
| unset | Log request/response framing to stderr |
Run the tests (zero dependencies, Node ≥ 20):
npm testThe full methodology (how taxes, costs, childcare subsidies, and safety-net scoring work) is on the homepage. One thing worth flagging: RSU income is not an input. Grants are treated as source-only because they usually don't follow you across employers. If you're keeping a US employer remotely, treat the result as directional in that direction.
Queries reach mcp.cityparity.com and nothing else. No accounts, no client-side telemetry, nothing personally identifiable logged beyond standard HTTP server logs.
MIT. cityparity's underlying data and calculation engine are not part of this package. This repo is the public MCP surface plus the open-source stdio bridge.
Found a bug or want a city added? https://cityparity.com/contact/
Compare two cities for one household: take-home pay, full cost breakdown, the equivalent target salary needed to match source net cash, non-cash lifestyle deltas (vacation, parental leave, healthcare), and a 0-100 quality score on five weighted dimensions. Use this for a head-to-head between two named cities; for a single city call get_city_summary, and to rank many cities call rank_cities. Read-only, no side effects; returns a text summary plus structured JSON. RSU income is NOT a parameter; RSU is treated as source-only because grants typically do not follow you across employers.
| Name | Required | Description | Default |
|---|---|---|---|
| source_city | Yes | City slug for the user's current city. Use list_cities to discover valid slugs (e.g. 'nyc', 'seattle', 'tokyo'). | |
| target_city | Yes | City slug for the destination city. | |
| gross_salary | Yes | User's annual gross salary in the SOURCE city's local currency (e.g. USD for Seattle, NOK for Oslo). | |
| household | No | Family structure: working-age partner and dependent children. Omit for a single person with no kids. | |
| living | No | Housing and spending assumptions. Omit to use defaults: rent, 2 bedrooms, typical spending. | |
| advanced | No | Optional fine-tuning: retirement contribution, age bracket, trips home, transit mode, and inbound tax regime. Omit to use defaults. | |
| score_weights | No | Override the composite-score weights. Defaults: financial 30, healthcare 20, vacation 15, childcare 15, safety_net 20. |
| Name | Required | Description |
|---|---|---|
| source | Yes | |
| target | Yes | |
| equivalence | Yes | |
| non_cash_deltas | Yes | |
| quality_score | Yes | |
| meta | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true and destructiveHint=false. Description reinforces 'Read-only, no side effects' and adds RSU treatment detail, providing context beyond annotations without contradiction.
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?
Two paragraphs, front-loaded with key outputs, every sentence adds value. No wasted words; structure is logical and easy to scan.
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 complexity (7 parameters, nested objects) and that output schema exists, the description covers purpose, usage, limitations, and output format completely. No gaps.
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%, so schema already documents parameters. The description adds value by clarifying that RSU income is not a parameter, explaining the rationale. This goes beyond the baseline of 3.
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 it compares two cities for a household, listing specific outputs like take-home pay, cost breakdown, target salary, lifestyle deltas, and a quality score. It distinguishes from siblings by naming get_city_summary and rank_cities.
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 says 'Use this for a head-to-head between two named cities' and contrasts with alternatives for single city or ranking many cities. Also clarifies RSU is not a parameter because it's source-only.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Concise profile of one city: currency, tax shape (bracket count + top rate + payroll/national insurance), headline costs (rent / groceries / transit / childcare), safety-net values (parental leave, vacation, universal healthcare), and data freshness. Lighter than compare_cities; use when the user is asking about one place rather than a comparison. Read-only, no side effects; returns a text summary plus structured JSON.
| Name | Required | Description | Default |
|---|---|---|---|
| city | Yes | City slug. Use list_cities to discover valid slugs. |
| Name | Required | Description |
|---|---|---|
| slug | Yes | |
| name | Yes | |
| country | Yes | |
| currency | Yes | |
| compensation_index | Yes | |
| tax | Yes | |
| cost_snapshot | Yes | |
| safety_net | Yes | |
| govt_benefits | Yes | |
| inbound_tax_regime | Yes | |
| data_last_updated_field | Yes | |
| data_year_range | Yes | |
| methodology_url | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already provide readOnlyHint, destructiveHint false, idempotentHint true. The description reinforces 'Read-only, no side effects' and adds behavioral details like data freshness and the return type (text summary + structured JSON). This adds value 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 three sentences, each earning its place: first states purpose and content, second gives usage guidance, third declares side effects and output format. No wasted words, well-structured.
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 complexity (single parameter, output schema present), the description covers all necessary context: input, output format (text + JSON), read-only nature, and data freshness. Output schema handles return details, so description is fully adequate.
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 description does not add meaning beyond the input schema for the single 'city' parameter. Schema coverage is 100% and includes its own description. The description mentions 'slug' implicitly but does not elaborate further. Baseline of 3 applies.
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 provides a 'concise profile of one city' with specific categories (currency, tax, costs, safety-net), and distinguishes itself from the sibling 'compare_cities' by noting it is 'lighter' and for single-place queries.
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 explicitly guides usage: 'use when the user is asking about one place rather than a comparison.' It also contrasts with 'compare_cities' as a lighter alternative, giving clear context for selection.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Look up a city's inbound-worker tax regime, the special reduced-tax schemes for relocating workers that general cost-of-living tools do not surface. Returns null when the country has no modeled regime (most cities). Modeled regimes: Italy impatriati (Milan, Rome), Portugal IFICI (Lisbon, Porto), Belgium expat (Brussels), Poland B2B ryczałt (Warsaw, Kraków), Greece inbound (Athens). Each entry includes the mechanic (reduction percentage or flat rate), duration, income cap if any, and a conservative eligibility hint. Read-only, no side effects; returns a text summary plus structured JSON.
| Name | Required | Description | Default |
|---|---|---|---|
| city | Yes | City slug. Use list_cities to discover valid slugs. Most cities return regime=null because their country has no special inbound-worker tax regime. |
| Name | Required | Description |
|---|---|---|
| slug | Yes | |
| city | Yes | |
| country | Yes | |
| regime | Yes | |
| disclaimer | No | |
| methodology_url | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint, destructiveHint, idempotentHint; the description adds that it is read-only with no side effects and returns both text and structured JSON. This fully discloses behavior without contradicting 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 concise but informative, front-loading the purpose and listing modeled regimes. Every sentence adds value, though a minor reduction in length could improve conciseness. Structurally clear.
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 output schema exists, the description need not detail return values beyond mentioning text summary and structured JSON. It adequately covers the tool's scope, null behavior, and examples, leaving no major gaps.
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 100% description coverage for the city parameter, including guidance to use list_cities and mention of null returns. The description does not add significant new parameter-level information beyond the schema, meeting the baseline for high coverage.
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 looks up a city's inbound-worker tax regime, a specific and distinct function from general cost-of-living tools. It lists modeled regimes, distinguishing it from sibling tools like get_city_summary or compare_cities.
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 indicates most cities return null and lists specific modeled regimes, providing clear context for when the tool is applicable. However, it does not explicitly contrast with sibling tools like get_city_summary, so guidance on alternatives is slightly lacking.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Batch lookup of safety-net values — parental leave weeks + paid percentage, universal healthcare flag, vacation days, public holidays — for 1 to 20 cities at once, plus each city's safety_net dimension score (0-100) for relative strength. This is the only tool that accepts many cities in one call: use it to line several places up on family / leave / healthcare benefits. For one city's full profile (tax shape + costs + safety net together) call get_city_summary instead; for a two-city head-to-head that includes cost-of-living use compare_cities. Read-only, no side effects; returns a text summary plus structured JSON.
| Name | Required | Description | Default |
|---|---|---|---|
| cities | Yes | Array of city slugs (1-20). Use list_cities to discover valid slugs. |
| Name | Required | Description |
|---|---|---|
| cities | Yes | |
| not_found | No | |
| methodology_url | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare read-only and idempotent. The description adds that there are no side effects and that the return includes a text summary plus structured JSON. This is consistent and adds value beyond the 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, well-structured paragraph. Every sentence earns its place, starting with the core purpose and then differentiating from siblings.
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 (1 param, good schema, annotations, output schema exists), the description covers all necessary aspects: what it does, how many items, what metrics, and how it relates to other tools.
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 only parameter 'cities' already has a good schema description. The description adds context that it's a batch operation and mentions the dimension score. It does not repeat the schema but complements it well.
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 it performs a batch lookup of safety-net values (parental leave, healthcare, vacation, holidays) plus a dimension score for 1-20 cities. It distinguishes itself from siblings by specifying alternatives for full profiles and two-city comparisons.
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 states when to use (batch lookup for multiple cities) and when not to use (for full profile or two-city comparison). Names specific alternatives: get_city_summary and compare_cities.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List all city slugs supported by cityparity, grouped by country. Call this first to discover valid slugs before any other tool; the optional country filter narrows to one country. Read-only, no side effects; returns a text summary plus structured JSON.
| Name | Required | Description | Default |
|---|---|---|---|
| country | No | Optional case-insensitive country filter. Use the canonical name as stored in city_data.country (e.g. "Norway", "USA", "United Kingdom"). If omitted, returns all countries. |
| Name | Required | Description |
|---|---|---|
| count | Yes | |
| country_count | Yes | |
| groups | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already provide readOnlyHint and destructiveHint. Description adds 'no side effects' and 'returns a text summary plus structured JSON', which adds value 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?
Two concise sentences with front-loaded purpose and efficient additional guidance. No unnecessary words.
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 low complexity, complete annotations, and existing output schema, the description covers purpose, usage order, safety, and return type sufficiently.
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 has 100% coverage with informative description for the country parameter. Description echoes the filter purpose but does not add new details beyond schema.
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 verb 'list', resource 'city slugs', and grouping 'by country'. It distinguishes itself from siblings by noting it should be called first to discover valid slugs.
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 says 'Call this first to discover valid slugs before any other tool' and explains the optional country filter. Provides clear context for when to use.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Rank supported cities by composite quality-of-life score across five weighted dimensions (financial, healthcare, vacation, childcare, safety_net). Returns top N (default 10), with filters for country, countries, region (europe/asia/north_america/south_america/oceania), has_universal_healthcare, include_cities, and exclude_cities. Use this to shortlist across many cities; for a head-to-head between two named cities use compare_cities. The same scenario is applied to every city so scores are directly comparable (default: single person, 2BR rent, $100k USD-equivalent gross). Read-only, no side effects; returns a text summary plus structured JSON.
| Name | Required | Description | Default |
|---|---|---|---|
| weights | No | Override the composite-score weights. Defaults: financial 30, healthcare 20, vacation 15, childcare 15, safety_net 20. | |
| filter | No | Restrict which cities are ranked, by country, region, universal-healthcare flag, or explicit include/exclude slug lists. Omit to rank all supported cities. | |
| scenario | No | Override the default ranking scenario. The same scenario is applied to every city so scores are comparable. | |
| limit | No | Max results to return. Default 10. |
| Name | Required | Description |
|---|---|---|
| ranked | Yes | |
| considered | Yes | |
| skipped | Yes | |
| weights | Yes | |
| scenario_used | Yes | |
| methodology_url | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true, destructiveHint=false, idempotentHint=true. The description adds that it is 'Read-only, no side effects; returns a text summary plus structured JSON.' This aligns with annotations and adds context about output format.
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, well-structured paragraph. Every sentence provides essential information without redundancy. It is front-loaded with the core action and then details filters and usage.
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 complexity (4 parameters, nested objects, no required fields), the description covers all aspects: purpose, usage, behavioral traits, parameter details, and output. The presence of an output schema further supports completeness.
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%, but the description elaborates on each parameter: it mentions default weights (30,20,15,15,20), explains that filter combinations use intersection, and specifies defaults for scenario fields (e.g., has_partner default false, gross_salary_usd default 100000). This adds significant value beyond the schema.
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 uses specific verbs ('rank') and clearly identifies the resource ('cities', 'composite quality-of-life score'). It distinguishes from the sibling `compare_cities` tool by explicitly stating its use for shortlisting across many cities vs. head-to-head comparison.
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 explicit guidance: 'Use this to shortlist across many cities; for a head-to-head between two named cities use compare_cities.' It also clarifies that the same scenario is applied to every city, ensuring comparability.
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.
v0.1.1compare_citiesget_city_summaryget_inbound_tax_regimeget_safety_netlist_citiesrank_citiesScored across 6 tools
Each tool has a clearly distinct purpose: compare_cities for head-to-head, get_city_summary for single city profile, get_inbound_tax_regime for tax schemes, get_safety_net for batch safety net, list_cities for discovering slugs, and rank_cities for ranking. No overlapping functionality.
All tool names follow a consistent verb_noun snake_case pattern (e.g., compare_cities, get_city_summary, list_cities). No mixing of conventions.
With 6 tools, the set is well-scoped for a city comparison service. Each tool covers a necessary operation (list, single summary, compare, rank, specific lookups) without being too few or excessive.
The tool surface covers the full workflow: discovering cities, getting single city profiles, comparing two cities, ranking many, and looking up specific tax regimes or safety nets. No obvious gaps for the stated purpose.
Cost of living, fair global pay, healthcare costs, nomad visas and tax residency. Free, no key.
140+ calculators and data tools — finance, health, science, global comparisons and more.
Live healthcare pay data: advertised ranges by role/state/city, offer checks, transparency scores.
Query 4,300+ ranked metros across sixteen dimensions: profiles, search, and head-to-head comparisons
