estv-tax
Enables AI assistants to calculate Swiss income, wealth, inheritance, capital payment, and corporate taxes for any municipality, compare tax burdens, and find cheapest locations without an API key.
README
Swiss Tax Calculator — MCP Server & Python API
Licensed under MIT by Aiyatullah Saiyed.
A dual-purpose Swiss tax toolkit: use it as a Python API client in your applications, or run it as an MCP server for AI assistants like Claude. Wraps the official ESTV Swiss Federal Tax Calculator — no API key or authentication required.
Covers income, wealth, lump-sum capital, inheritance/gift, and corporate taxes for every Swiss municipality across tax years 2010–2026.
Features
| Category | Tools |
|---|---|
| Household tax | find_location, calculate_tax, list_deductions, calculate_tax_from_taxable_amounts |
| Compare places | compare_locations, find_cheapest_nearby |
| Tax planning | plan_capital_withdrawals, deduction_value |
| Other taxes | calculate_capital_payment_tax, calculate_inheritance_tax, calculate_company_tax |
| Reference | explain_tax_brackets, get_tax_years |
Quick Start
Install
# Clone the repository
git clone https://github.com/aiyatullah/swiss-tax-calculator-mcp.git
cd swiss-tax-calculator-mcp
# Install dependencies
uv sync
Usage as Python API
Import EstvClient directly in your Python code to query the ESTV tax calculator programmatically.
Basic Example — Find a Location
from estv_mcp import EstvClient
client = EstvClient()
# Search for a municipality by name or postal code
locations = client.search_location("Zug", tax_year=2026)
for loc in locations[:3]:
print(f"{loc['City']} ({loc['Canton']}) — ID: {loc['TaxLocationID']}")
Calculate Income Tax
from estv_mcp.api import EstvClient, RELATIONSHIP, CONFESSION, INCOME_TYPE
client = EstvClient()
# Step 1: Find the tax location
locations = client.search_location("8001", tax_year=2026)
location_id = locations[0]["TaxLocationID"]
# Step 2: Get the tax budget (deduction sheet)
budget = client.tax_budget({
"SimKey": None,
"TaxYear": 2026,
"TaxLocationID": location_id,
"Relationship": RELATIONSHIP["single"],
"Confession1": CONFESSION["none"],
"Confession2": 0,
"Children": [],
"Age1": 35,
"RevenueType1": INCOME_TYPE["employed"],
"Revenue1": 120_000,
"Age2": 0,
"RevenueType2": 0,
"Revenue2": 0,
"Fortune": 50_000,
})
# Step 3: Calculate detailed taxes
result = client.detailed_taxes({
"SimKey": None,
"TaxYear": 2026,
"TaxLocationID": location_id,
"Relationship": RELATIONSHIP["single"],
"Confession1": CONFESSION["none"],
"Confession2": 0,
"Children": [],
"Age1": 35,
"RevenueType1": INCOME_TYPE["employed"],
"Revenue1": 120_000,
"Age2": 0,
"RevenueType2": 0,
"Revenue2": 0,
"Fortune": 0,
"Budget": budget,
})
print(f"Total tax: CHF {result['TotalTax']:,.0f}")
print(f"Federal: CHF {result.get('IncomeTaxFed', 0):,.0f}")
print(f"Cantonal: CHF {result.get('IncomeTaxCanton', 0):,.0f}")
print(f"Municipal: CHF {result.get('IncomeTaxCity', 0):,.0f}")
Compare Municipalities
from estv_mcp.api import EstvClient, RELATIONSHIP, CONFESSION, INCOME_TYPE, GROUP_CAPITALS
client = EstvClient()
# Compare tax burden across all 26 cantonal capitals
rows = client.many_simple_taxes({
"SimKey": None,
"TaxYear": 2026,
"TaxGroupID": GROUP_CAPITALS,
"Relationship": RELATIONSHIP["single"],
"Confession1": CONFESSION["none"],
"Confession2": 0,
"Children": [],
"Age1": 35,
"RevenueType1": INCOME_TYPE["employed"],
"Revenue1": 100_000,
"Age2": 0,
"RevenueType2": 0,
"Revenue2": 0,
"Fortune": 0,
})
ranked = sorted(rows, key=lambda r: r["TotalTax"])
print("Top 5 cheapest cantonal capitals:")
for r in ranked[:5]:
loc = r["Location"]
print(f" {loc['City']} ({loc['Canton']}): CHF {r['TotalTax']:,.0f}")
Inheritance Tax
from estv_mcp.api import EstvClient, BENEFICIARY, GROUP_CAPITALS
client = EstvClient()
# Inheritance tax for a sibling across cantonal capitals
rows = client.many_inheritance_taxes({
"SimKey": None,
"TaxYear": 2026,
"TaxGroupID": GROUP_CAPITALS,
"OnlyGroupID": BENEFICIARY["sibling"][0],
"OnlyPersonID": BENEFICIARY["sibling"][1],
"Donation": False,
"Amount": 500_000,
})
ranked = sorted(rows, key=lambda r: r["TaxTotal"])
print(f"Cheapest: {ranked[0]['Location']['City']} — CHF {ranked[0]['TaxTotal']:,.0f}")
print(f"Most expensive: {ranked[-1]['Location']['City']} — CHF {ranked[-1]['TaxTotal']:,.0f}")
Available API Operations
| Method | Description |
|---|---|
search_location(query, tax_year) |
Find municipalities by name or postal code |
search_location_geo(lat, lon, radius_km, tax_year) |
Find municipalities within a radius |
tax_budget(payload) |
Get the deduction/budget sheet for a household |
detailed_taxes(payload) |
Full tax calculation with breakdown |
simple_taxes(payload) |
Tax from pre-computed taxable amounts |
many_simple_taxes(payload) |
Compare tax across many municipalities |
many_capital_taxes(payload) |
Lump-sum capital withdrawal tax across municipalities |
many_inheritance_taxes(payload) |
Inheritance/gift tax across municipalities |
many_legal_entity_taxes(payload) |
Corporate tax across municipalities |
export_tax_scales(tax_year, tax_group_id) |
Raw tax bracket tables |
tax_year_range(calculator) |
Supported year range per calculator |
tax_version() |
Current ESTV data version |
Enum Reference
from estv_mcp.api import (
RELATIONSHIP, # single=1, married=2, concubinage=3, registered_partnership=4
CONFESSION, # reformed=1, roman_catholic=2, christ_catholic=3, none=4, other=5
INCOME_TYPE, # employed=1, self_employed=2, pensioner=3, other=4
GENDER, # male=1, female=2
LANGUAGE, # de=1, fr=2, it=3, en=4
CANTON_GROUP, # AG=1 ... ZH=26
GROUP_CAPITALS, # 88 — all 26 cantonal capitals
GROUP_SWITZERLAND, # 99 — every municipality (~2100)
CALCULATOR, # income_wealth=1, capital_payment=2, legal_entity=3, inheritance=5
BENEFICIARY, # spouse, child, sibling, unrelated, etc. → (GroupID, PersonID)
)
Caching
Responses are cached under ~/.cache/estv-mcp for one week. Control via environment variables:
| Variable | Effect |
|---|---|
ESTV_MCP_NO_CACHE=1 |
Disable caching entirely |
ESTV_MCP_CACHE_DIR=/path |
Custom cache directory |
ESTV_MCP_CACHE_TTL=3600 |
Cache lifetime in seconds |
Usage as MCP Server
Run this project as an MCP server to give AI assistants (Claude, etc.) access to all Swiss tax tools.
Register with Claude Code
# Project-local (this repo only)
claude mcp add estv-tax -- uv run --directory "$PWD" estv-mcp
# User-wide (available in every project)
claude mcp add -s user estv-tax -- uv run --directory "$PWD" estv-mcp
MCP Client Config (JSON)
For any MCP-compatible client, add to your config:
{
"mcpServers": {
"estv-tax": {
"command": "uv",
"args": ["run", "--directory", "/path/to/swiss-tax-calculator-mcp", "estv-mcp"]
}
}
}
What the AI Can Do
Once connected, the AI assistant can:
- "How much tax would I pay in Zurich on CHF 150,000?" — calls
find_location+calculate_tax - "Compare Zug vs Schwyz for a married couple with 2 kids" — calls
compare_locations - "Where's the cheapest place within 30km of Zurich HB?" — calls
find_cheapest_nearby - "Should I split my pillar 2 withdrawal over multiple years?" — calls
plan_capital_withdrawals - "What's the inheritance tax for a sibling in Geneva?" — calls
calculate_inheritance_tax - "How much would my company pay in Zug vs Lucerne?" — calls
calculate_company_tax
Development
uv sync # Install dependencies (including dev)
uv run pytest -q # Run tests (hits live ESTV API)
uv run python scripts/stdio_smoke.py # MCP stdio handshake test
uv run ruff check . && uv run ruff format --check . # Lint & format check
Golden Values
tests/golden.json pins known tax figures for closed tax years. The weekly CI run re-checks them against the live API to detect upstream changes.
uv run python scripts/update_golden.py # Update golden values (prints diffs)
CI
Tests run on Python 3.11/3.12/3.13 on every push, plus a weekly schedule to catch ESTV data changes.
Important Notes
- All amounts are CHF per year
income_type='employed'means gross salary — social contributions (AHV/IV/EO, ALV, NBU, BVG) are derived automatically- Override deductions via
deductions={id: value}using IDs fromlist_deductions(e.g.PRAEMIEN3Afor pillar 3a,SCHULDZINSENfor mortgage interest) compare_locations(scope='switzerland')covers ~2,100 municipalities; results are summarized to stay readable- Church tax only applies when
confessionis notnone - Figures are from the official ESTV model — not a binding tax assessment
License
MIT License — see LICENSE.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。