SuperAPI AI Directory

japan-company-info-mcp-bridge

Offline Japanese company due diligence: EDINET XBRL financials, NTA corporate numbers, major shareholders, gBizINFO certifications and industry benchmarks.

Open website

About this tool

Purpose

japan-company-info-mcp-bridge is an MCP (Model Context Protocol) server for Japanese corporate due diligence, featuring 100% offline operation. It integrates four official Japanese data sources into a single retrieval layer, allowing AI assistants to locally query Japanese companies' registration information, statutory financial statements, shareholder structures, and industry benchmarks. The official description emphasizes "Zero cloud leakage. Zero Docker.", meaning queries are completed on the user's machine without going through third-party APIs.

Main Capabilities

The server exposes five domain tools, all normalized into precise queries, plus one general fallback tool:

  • edinet_financials_usgaap(company_name, metric?, fiscal_year?): Queries EDINET statutory financial statements (income statement, balance sheet, cash flow statement) and executive compensation, with accounting concepts mapped between J-GAAP / US GAAP / IFRS.
  • japan_corporate_registry(company_name, info_type?): Queries the 13-digit National Tax Agency corporate number, registered address, legal status, and gBizINFO certifications / subsidies.
  • japan_shareholders(company_name, top_n?): Extracts major shareholders and equity structure from the "Major Shareholders" section of annual securities reports.
  • japan_industry_benchmarks(industry, metric?): Obtains industry benchmarks (operating profit margin, ordinary profit margin, equity ratio, ROE) from the Ministry of Finance's Corporate Enterprise Statistical Survey via e-Stat.
  • japan_company_search(query, method?, top_k?): Keyword / fuzzy / semantic search, used to discover companies when the target company is unknown or the name is unclear.
  • search_knowledge(query, method?, top_k?): Raw knowledge base search (fallback). method can be auto (default), bm25 (exact keyword), trigram (fuzzy / identifier), vector (semantic), hybrid (RRF fusion).

All five domain tools internally resolve to search_knowledge, so the retrieval layer remains generic with the .orb data capsule.

A built-in J-GAAP / US GAAP / IFRS mapping dictionary is included, and the official description specifically notes that it can correctly distinguish 営業利益 (Operating Income) from 経常利益 (Ordinary Income, a concept unique to J-GAAP), avoiding mistranslation when reading Japanese disclosure documents in English.

Usage

The server bundles the mcporb-runtime binary and a pre-indexed Orb (BM25 + trigram + optional dense vector retrieval). Retrieval runs on the user's machine. On first startup, the runtime downloads the query embedding model (about 220MB) in the background to enable the semantic vector method; before the download completes and thereafter in a permanently offline state, the bm25, trigram, and auto methods work without any network access.

To connect to Claude Desktop: add this server in claude_desktop_config.json . Before the package is published to npm, use the GitHub form:"args": ["-y", "github:dqj1998/japan-company-info-mcp-bridge"]. You can also add the same server in Settings → MCP → Add Server (with command as npx, args same as above). Supports running locally from a cloned repository.

Platform support: bundled runtime binaries in bin/ (no external system libraries required, runs on minimal / headless images): macOS Apple Silicon (arm64) — mcporb-runtime-darwin-arm64 Linux x86-64 — mcporb-runtime-linux-x64 Linux arm64 — mcporb-runtime-linux-arm64 Windows x64 — mcporb-runtime-win32-x64.exe Not bundled: macOS Intel (x86-64). index.js resolves mcporb-runtime-<platform>-<arch>, and if no matching binary is found, gives a clear message and exits.

Use cases

The officially listed use cases include: cross-border equity analysis, KYB / due diligence entity verification, M&A target screening, and reading Japanese statutory disclosure documents in English without mistranslating 営業利益 and 経常利益.

Limitations and data notes

The free version indexes only 192 blue-chip companies (EDINET-listed, Nikkei 225 constituents). Queries for companies outside this set return the closest available match; the full dataset (3,193+ listed companies, including complete XBRL financial statements, major shareholders, gBizINFO details, and Ministry of Finance Corporate Enterprise Statistical Survey industry benchmarks) requires a one-time purchase at mcporb.store.

Regarding retrieval reliability: searching by corporate number or securities code (exact identifiers) is most reliable, followed by Japanese company names; English company name search only covers companies with official English names (the rest is best-effort).

Data sources and attribution: 法人番号公表サイト (National Tax Agency) — corporate registration; EDINET (Financial Services Agency) — annual securities report XBRL financial statements; gBizINFO (Ministry of Economy, Trade and Industry) — certifications, subsidies, awards; 財務省 法人企業統計調査 — industry benchmarks (full version). Data is redistributed under the terms of each source; see NOTICE for details. The bridge code is MIT licensed (see LICENSE); the bundled Orb data and mcporb-runtime binaries are not MIT, are licensed separately, and see NOTICE for details.