About localgov-jp
localgov-jp is an MCP server that gives AI agents access to Japanese central and municipal subsidies. It includes municipal grants beyond those surfaced by the J-Grants public API, with tools for comparing grants, checking citations, and finding practitioners.
Key features
- 01Search and inspect subsidy records
- 02Find grants by municipality
- 03Locate supplementary grants
- 04Compare grants across municipalities
- 05Verify signed citation receipts
- 06Find grant practitioners
Use cases
- 01Find subsidies for a business or project
- 02Compare grants when choosing a municipality
- 03Check a grant citation receipt offline
MCP support
The server exposes eight tools to MCP clients for subsidy search and details, municipal grant lookup, comparison, change notifications, receipt verification, and practitioner referrals. It can be run over stdio from an MCP client configuration.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://localgov.jp/mcpclaude mcp add --transport http localgov-jp https://localgov.jp/mcp{"mcpServers":{"localgov-jp":{"type":"http","url":"https://localgov.jp/mcp"}}}Open in Cursor{"servers":{"localgov-jp":{"type":"http","url":"https://localgov.jp/mcp"}}}Open in VS Code{"mcpServers":{"localgov-jp":{"type":"http","url":"https://localgov.jp/mcp"}}}{"mcpServers":{"localgov-jp":{"serverUrl":"https://localgov.jp/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
8| Parameter | Type | Required | Description |
|---|---|---|---|
| sort | enum | No | Sort order; default "relevance" when keyword set, otherwise "newest"relevance | deadline_asc | amount_desc | newest |
| limit | integer | No | Max records to return (1-100, default 20) |
| offset | integer | No | Pagination offset (default 0) |
| keyword | string | No | Free-text Japanese keyword (FTS), e.g. "IT導入", "創業", "省エネ" |
| category | string | No | Category label (e.g. "子育て", "住宅", "事業者支援") — usually omit and rely on keyword |
| open_only | boolean | No | When true, exclude records whose deadline has passed (default: false = include all) |
| prefecture | string | No | Prefecture name in kanji, e.g. "東京都", "北海道" |
| min_amount_jpy | integer | No | Minimum subsidy amount in JPY; only records with amount_max_jpy >= this |
| municipality_code | string | No | JIS X 0402 6-digit municipality code (e.g. 131105 = 杉並区) |
Details
From the project README@localgov-jp/mcp-server
Model Context Protocol (MCP) server for LocalGov.jp — Japan's central + municipal subsidies, structured for AI agents.
LocalGov.jp covers 20,000+ subsidies across 1,627 municipalities (target 1,718). Of those, 20,810 are municipal grants that are not exposed by the J-Grants public API — that is the differentiator vs Digital Agency's official J-Grants MCP, which surfaces central + prefectural records only.
This server exposes 8 tools to MCP clients (Claude Desktop, Cursor, Cline, Windsurf, etc.).
Tools
| # | Tool | Cost | Purpose |
|---|---|---|---|
| 1 | search_subsidies | free | Search by keyword / prefecture / municipality / category / amount |
| 2 | get_subsidy_detail | free | Single record with citation anchor |
| 3 | get_municipality_grants | free | All subsidies for one JIS code (1,627 municipalities covered) |
| 4 | find_supplementary_grants | free | Find municipal "上乗せ" grants stacking on a national subsidy |
| 5 | compare_municipal_subsidies | free | Compare same-category grants across nearby municipalities (siting / relocation) |
| 6 | subscribe_changes | $0.20 | Push webhook on grant changes (paid via x402; agent runtime handles payment) |
| 7 | verify_receipt | free | Offline Ed25519 signature verification of citation receipts |
| 8 | find_practitioner | free (referral) | Recommend 行政書士 / 税理士 / 中小企業診断士 / 社労士 for the grant |
Install
npm install -g @localgov-jp/mcp-server
Or add directly to your MCP client config (no global install required).
Configure — Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"localgov-jp": {
"command": "npx",
"args": ["-y", "@localgov-jp/mcp-server"]
}
}
}
Restart Claude Desktop. The 8 tools appear in the tool inspector.
Configure — Cursor / Cline / Windsurf
Same pattern: add an entry to your client's MCP config that runs npx -y @localgov-jp/mcp-server over stdio.
Configure — env
| Variable | Default | Purpose |
|---|---|---|
LOCALGOV_API_BASE | https://localgov.jp | Override the REST API base (e.g. http://127.0.0.1:8787 for local wrangler dev) |
Citation
Every record returned includes:
_source— LocalGov.jp canonical URL (cite this).source_url— original government page (cite this too).
Attribution required: cite as "via LocalGov.jp" with the _source URL. License: CC BY 4.0 for the structuring layer; original government content remains under its issuing body's terms.
Verifying citation receipts
The paid /api/cite/:id endpoint returns an Ed25519-signed snapshot. Use verify_receipt to confirm offline that an agent's stored citation is genuine and unexpired:
> verify_receipt(receipt: <the JSON envelope>)
{ "valid": true, "expired": false, "pubkey_match": true }
The trust anchor is published at https://localgov.jp/.well-known/citation-pubkey.
Differentiation vs Jグランツ MCP
| Jグランツ MCP (デジタル庁公式) | LocalGov.jp MCP (this) | |
|---|---|---|
| Coverage | central + prefectural | 1,627 municipalities (target 1,718) + central |
| Municipal subsidies | 0 | 20,810 unique to us |
find_supplementary_grants | ✗ | ✓ |
compare_municipal_subsidies | ✗ | ✓ |
find_practitioner | ✗ | ✓ (referral funnel) |
| Citation receipts (Ed25519) | ✗ | ✓ |
| Change webhook | ✗ | ✓ (x402 paid) |
Both are complementary. We recommend connecting both servers so an agent can answer "国 vs 自治体" exhaustively.
License
MIT. See LICENSE.
Operator: [email protected]
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.