Estat Japan MCP server

@pipeworx/estat-japan

e-Stat (Japan) MCP — Japanese government statistics aggregator (~5000 tables across population, economy, labor, trade, prices, agriculture, environment). Free with an app ID.

Part of Pipeworx — an MCP gateway connecting AI agents to 1679+ live data sources.

Tools

• searchstats(query, limit?, startposition?, lang?) — find stats tables
• getmetadata(statsdata_id, lang?) — dimensions + code lists for a table
• getdata(statsdataid, limit?, startposition?, lang?, filters?) — observations
• listdatacatalog(query?, limit?, startposition?, lang?, datatype?, surveyyears?, statscode?, stats_field?) — browse catalog

Auth

• Platform key: gateway env PLATFORMESTATKEY (an appId issued by e-stat.go.jp)
• BYO: ?_apiKey= after registering at https://www.e-stat.go.jp/api/

Data source

https://api.e-stat.go.jp/rest/3.0/app/json/ — appId as query param.

Default language is Japanese (J); pass lang: "E" for English where available.

Timeouts

e-Stat is slow and sets no timeout of its own. Measured 2026-08-27: /getStatsList takes ~26s uncached, and /getDataCatalog ran past the gateway's 75s request deadline at page sizes 1, 5 and 20 alike — page size is not what makes it slow.

Each request therefore carries its own budget — 55s (about 20s under the gateway's own 75s deadline), and 35s for listdatacatalog, which is the endpoint that does not finish at all rather than one that finishes late. Exceeding the budget returns a soft failure rather than hanging:

{ "found": false, "reason": "upstream_timeout", "endpoint": "...", "timeout_ms": 35000,
  "hint": "... use search_stats ..." }

limit defaults to 20 on listdatacatalog (100 on getdata, 20 on searchstats), so a bare call is already bounded — page size was never what made the catalog slow. The catalog accepts e-Stat's own narrowing params (query, datatype, statscode, surveyyears); during the 2026-08-27 stall none of them shortened it, so treat them as filters rather than as a fix. searchstats searches the same tables and answered in seconds throughout.

Quick Start

Add to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):

{
  "mcpServers": {
    "estat-japan": {
      "url": "https://gateway.pipeworx.io/estat-japan/mcp"
    }
  }
}

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/estat-japan/mcp returns the tools in the table above plus the shared Pipeworx meta-tools — askpipeworx, discovertools, search_within, remember/recall and the rest of the gateway-wide set. So the tool count you see is larger than this table: a single-pack endpoint currently lists roughly 30 shared tools alongside the pack's own. The connection's initialize response states its exact scope, and is the authoritative answer for a given day.

This is deliberate, not multiplexing by accident. The meta-tools are what let a scoped connection answer a question this pack does not cover — via ask_pipeworx, which routes across the whole catalog — without you adding a second MCP server. There is currently no way to mount a pack endpoint without them; if the extra schemas cost you more context than the routing is worth, connect to the full gateway once rather than to several pack endpoints.

Or connect to the full Pipeworx gateway to get every pack's tools listed directly, instead of just this one's:

{
  "mcpServers": {
    "pipeworx": {
      "url": "https://gateway.pipeworx.io/mcp"
    }
  }
}

Both URLs reach the same gateway and the same 1679+ data sources. The only difference is which pack's tools are listed directly; ask_pipeworx reaches all of them from either one.

No MCP client? Call it over HTTP

curl -X POST https://gateway.pipeworx.io/v1/tools/search_stats \
  -H 'Content-Type: application/json' \
  -d '{"query":"失業率"}'

Shortened. The full README is on GitHub.