Data sources

EDINET: Japan's official corporate disclosure system

All structured filing data comes from EDINET, the Financial Services Agency's electronic disclosure system. EDINET is the regulatory source of truth for every listed company in Japan, covering securities reports, large shareholding filings, tender offers, extraordinary reports, treasury stock filings, and the rest of the document family.

Japan Finsight syncs EDINET daily. Every filing since January 2018 is ingested. Every tool response includes the filing identifier (edinet_document_id) so you can audit any number back to its source on EDINET's portal.

Parser layer

The XBRL parsing that turns EDINET's CSV dumps into typed structured data is handled by edinet-tools, an MIT-licensed Python library, developed and maintained as a separate open-source project. pip install edinet-tools gives you the same parsers Japan Finsight uses internally, by design. The substrate is verifiable, not a black box.

Enrichment layer

Several Japan Finsight tools wrap LLM-extracted enrichments on top of the deterministic XBRL parsing:

• get_capital_policy: target payout ratios and dividend skip reasons extracted from annual report text blocks via Gemini 2.5 Flash Lite, with source quotes embedded in the response.
• get_related_party_transactions: counterparties, transaction types, and amounts structured from text via LLM, with substring-match guards on every quote.
• get_holding_actions: activist intent annotations on large-shareholding filings (Doc 350). Curated registry of known activist filers, plus LLM-extracted change proposals and desired outcomes.
• get_material_events: extraordinary report (Doc 180) classification via FSA ordinance-item-code regex.

Enrichment policy: we extract facts the company stated in the filing with source quotes. We do not derive judgments about the company. Empty results mean "the company didn't state this," not "the company doesn't do this." Every enriched field is auditable back to the source filing text.

Companion projects

• edinet-tools: the parser library (PyPI).
• kabu-agent: a companion AI agent for working with EDINET data (open source).

What we don't source from

• Bloomberg / FactSet / Refinitiv: none. All structured data is sourced directly from FSA EDINET.
• Live market data: not redistributed through public tools; commercial licensing required.
• TDNet: earnings short-reports (決算短信) are not yet ingested; planned for a future release.