datagokr — Claude Code MCP server: search data.go.kr datasets
Finds datasets on the data.go.kr portal from a plain-language description and shows columns and the first row without an API key.
claude mcp add --transport http datagokr https://datagokr.dev/mcpWhat datagokr — 공공데이터포털 검색을 더 쉽게 (find data.go.kr datasets easily) does
The server indexes the data.go.kr catalogue down to column names, so a table turns up even when its title does not match your wording. Related datasets are grouped by topic, and before you download anything you can see which fields exist and what the first row looks like. Search, structure lookup and previews run through the public server at https://datagokr.dev/mcp, with no key and no account.
Full file downloads, usage applications and open API calls with your own key happen locally and need the datagokr-mcp package plus Python 3.11 or newer. Applications can be batched, up to 50 datasets at a time, and downloaded files land in ~/datagokr/<dataset_id>/. Your key and portal session stay on your machine and go only to the official data.go.kr services.
Settings resolve in this order: function or CLI arguments, environment variables, a .env in the working directory, ~/.config/datagokr/config.toml, then defaults. Sharing one TOML file across clients is simpler, and it deserves chmod 600. In the web versions of Claude and ChatGPT only the first stage works — no applications, no downloads there.
What it helps with
- find a dataset by meaning instead of an exact title
- inspect column names and the first row before downloading
- file usage applications for several datasets at once
- pull a full dataset along with its earlier versions
- query an open API with your own key without building the URL by hand
Example prompts for the agent
- Find nationwide parking lot data and show me its columns
- Which datasets have latitude, longitude and hospital name together?
- Apply for these datasets and download them
How to install datagokr — 공공데이터포털 검색을 더 쉽게 (find data.go.kr datasets easily)
Platform CLI: remote server
claude mcp add --transport http datagokr https://datagokr.dev/mcpPlatform CLI: remote server
codex mcp add datagokr-find-datagokr-datasets-easily --url https://datagokr.dev/mcpConfig file
{
"mcpServers": {
"datagokr-find-datagokr-datasets-easily": {
"url": "https://datagokr.dev/mcp"
}
}
}Config file
{
"servers": {
"datagokr-find-datagokr-datasets-easily": {
"type": "http",
"url": "https://datagokr.dev/mcp"
}
}
}Platform CLI: remote server
gemini mcp add -t http datagokr-find-datagokr-datasets-easily https://datagokr.dev/mcpStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about datagokr — 공공데이터포털 검색을 더 쉽게 (find data.go.kr datasets easily)
How do I install datagokr in Claude Code?
Add the remote server with claude mcp add --transport http datagokr https://datagokr.dev/mcp. For the second stage, applications and downloads, install the local package: pip install datagokr-mcp.
What does datagokr do?
It searches data.go.kr datasets from a plain-language description and shows their structure and first row. With the local package it also files usage applications and downloads files.
Do I need an API key to search?
No. Search, column lookup and first-row previews work without a key or a login. The key is only needed for applications, downloads and open API calls.
