Skip to main content

MCP server for KOSIS (Statistics Korea) — Korean macro/social/economic statistics for AI agents

Project description

vivory-mcp-kosis

MCP server for KOSIS (Statistics Korea) — Korean macro/social/economic statistics for AI agents.

This package wraps the Vivory Korea Data Gateway and exposes 15 tools to any MCP-compatible client (Claude Desktop, Claude Code, etc.).

Built and maintained by Vivory. Data sourced from KOSIS · 통계청.


Why

Korean public statistics (GDP, CPI, unemployment, demographics, trade balance, etc.) are published by Statistics Korea via KOSIS — but the API is Korean-only, requires a key, returns JS-literal responses, and uses scattered taxonomies. This MCP server lets foreign analysts, AI agents, and developers query Korean statistics in English without any setup beyond a one-line install.

You → Claude → vivory-mcp-kosis → api.vivory.app → kosis.kr
                                  (caching, attribution, normalization)

No KOSIS account, no API key, no Korean phone — just install and ask.


Installation

Note: PyPI publication coming soon. Until then, install directly from this repo via Git.

Option A — Claude Desktop (recommended)

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "vivory-kosis": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/jayjodev/vivory-mcp.git#subdirectory=packages/mcp-server-kosis",
        "vivory-mcp-kosis"
      ]
    }
  }
}

Restart Claude Desktop. The 15 KOSIS tools appear in the tool palette.

Option B — Claude Code (CLI)

claude mcp add vivory-kosis -- uvx --from "git+https://github.com/jayjodev/vivory-mcp.git#subdirectory=packages/mcp-server-kosis" vivory-mcp-kosis

Option C — Manual pip install (from Git)

pip install "git+https://github.com/jayjodev/vivory-mcp.git#subdirectory=packages/mcp-server-kosis"
vivory-mcp-kosis  # runs the stdio MCP server

Option D — Once published to PyPI (planned)

pip install vivory-mcp-kosis           # or:
uvx vivory-mcp-kosis                    # zero-install run

Tools

Discovery (subscription-free)

Tool Purpose
kosis_categories List 16 ROOT categories (Population, Labor, CPI, GDP, etc.)
kosis_category_indicators List indicators inside a category (returns indicator_id for time-series)
kosis_indicators_search Search indicators by Korean keyword
kosis_indicator_timeseries Fetch time-series values by indicator_id or indicator_name
kosis_table_search Full-text search across the entire KOSIS catalog

Pre-built indicators

Tool Indicator
kosis_population Total population by region (annual)
kosis_gdp National accounts main indicators (quarterly)
kosis_employment Unemployment, employment, participation (monthly)
kosis_cpi Consumer Price Index — overall, food, core (monthly)
kosis_household_income Household Income & Expenditure by decile (quarterly)
kosis_trade_balance Exports, imports, net exports (monthly)

Meta

Tool Purpose
kosis_table_meta Table metadata (TBL/ORG/PRD/ITM/CMMT/UNIT/SOURCE/WGT/NCD)
kosis_table_explanation Survey methodology, legal basis, periodicity
kosis_statistic_tree Browse the hierarchical statistic tree

Aggregate

Tool Purpose
kosis_key_indicators Snapshot of 6 key indicators in one call

Example prompts to Claude

After installing the MCP server, just ask Claude:

"What's Korea's population trend over the last 10 years?" "How has the Korean Consumer Price Index moved over the last 24 months?" "Show me Korean quarterly GDP growth, last 12 quarters." "Search KOSIS for tables related to youth unemployment." "What's the indicator ID for 추계인구?"

Claude will pick the right tool automatically.


Data attribution

Every tool response includes an attribution block:

{
  "attribution": {
    "source": "KOSIS",
    "agency": "통계청 (Statistics Korea)",
    "agency_en": "Statistics Korea (KOSTAT)",
    "license": "Public data — commercial use allowed with attribution",
    "source_url": "https://kosis.kr",
    "attribution_required": "자료: KOSIS 국가통계포털, 통계청"
  }
}

When citing or redistributing data, include the attribution_required text. KOSIS data is licensed under 공공누리(KOGL) Type 1 — commercial use, modification, and redistribution permitted with source attribution.


Self-hosting

By default the server calls https://api.vivory.app/api. To point at your own Vivory deployment (or local development), set:

export VIVORY_API_BASE="http://localhost:8000/api"

You'll also need a working Vivory backend with KOSIS_API_KEY configured — see Vivory backend (private monorepo).


Rate limits

The Vivory gateway caches all KOSIS responses in Redis (TTL 1h~24h depending on endpoint), so individual MCP calls are typically sub-50ms. Upstream KOSIS limits are 1,000 calls/minute and 40,000 cells per response — Vivory caching hides these almost entirely from you.

For high-volume commercial use, contact contact@vivory.app about Pro tier ($29/mo) and Enterprise tier ($299/mo).


Project status


What's next

  • More public Korean datasets (BoK ECOS, NEIS schools, real estate, weather, air quality) under separate MCP servers
  • Anthropic MCP Directory submission
  • vivory-mcp-korea umbrella package combining all Korean data sources

🇰🇷 Built in Seoul · 🌐 vivory.app

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

vivory_mcp_kosis-0.1.0.tar.gz (11.4 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

vivory_mcp_kosis-0.1.0-py3-none-any.whl (11.6 kB view details)

Uploaded Python 3

File details

Details for the file vivory_mcp_kosis-0.1.0.tar.gz.

File metadata

  • Download URL: vivory_mcp_kosis-0.1.0.tar.gz
  • Upload date:
  • Size: 11.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for vivory_mcp_kosis-0.1.0.tar.gz
Algorithm Hash digest
SHA256 5b63c6c5d674e7870ccf5f4579770b5a9a295b5ec96c5330e78d93de20171eba
MD5 83e68448811cc9178a509f5b78a5dc21
BLAKE2b-256 54709956647600451b24c6281fdcb7fa374d7c076be87941b2aacb914d59a9b2

See more details on using hashes here.

Provenance

The following attestation bundles were made for vivory_mcp_kosis-0.1.0.tar.gz:

Publisher: publish-mcp-kosis.yml on jayjodev/vivory-mcp

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file vivory_mcp_kosis-0.1.0-py3-none-any.whl.

File metadata

File hashes

Hashes for vivory_mcp_kosis-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 a043f3cf750040b4dfd79d9635b228c92842738cc91b43c3c91dc7fd2cc27928
MD5 d5b746347c739a672b9dbdc913299e44
BLAKE2b-256 558862871c76c943f37afd5ed65584ef7e5652629b50c662f50f513e56273156

See more details on using hashes here.

Provenance

The following attestation bundles were made for vivory_mcp_kosis-0.1.0-py3-none-any.whl:

Publisher: publish-mcp-kosis.yml on jayjodev/vivory-mcp

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page