chrisryugj/korean-dart-mcp

npx -y korean-dart-mcp setup
npm install -g korean-dart-mcp
/plugin marketplace add chrisryugj/korean-dart-mcp

Usage

• •"삼성전자 최근 분기 매출·영업이익, 전년 동기 대비 변화"
• •"최근 30일 자기주식 취득 결정한 상장사 전부"
• •"삼성전자 임원 거래 최근 2년 매수/매도 클러스터 분석"
• •"카카오 최근 3년 회계 리스크 스코어"

Key features

• ✓83 DART APIs compressed into 15 tools (search_disclosures, get_xbrl, get_financials, download_document, get_attachments)
• ✓Analyst frames: insider_signal, disclosure_anomaly 0-100 score, buffett_quality_snapshot
• ✓Company name auto-resolution via preloaded SQLite FTS (korean/stock-code/corp_code)
• ✓HWP/HWPX/PDF/DOCX/XLSX attachments to markdown via kordoc engine
• ✓XBRL presentation/calculation linkbase parsing with calculation verification (markdown_full)
• ✓22 disclosure presets and 90-day auto-chunking for market-wide searches
• ✓ZIP slip/bomb defense and page parallelization

Best for

Retail investors and analysts who want to query published Korean corporate disclosures through an AI assistant in natural language

Caveats

• ⚠Requires a free OpenDART API key (DART_API_KEY, 20,000 requests/day)
• ⚠Endpoint 1:1 coverage is narrower than 83-tool Python MCP alternatives
• ⚠Research aid only, not investment advice; the ecosystem is still early
• ⚠First run downloads an ~116K-company dump (~5-10s) for company-name resolution