corporate-events-mcp
Normalizes and serves corporate events (earnings, dividends, splits, M\&A) from US, UK, Canada, and Australia via a programmatic MCP API.
README
corporate-events-mcp
Cross-jurisdiction corporate events normalization via MCP. A single consistent schema covering earnings dates, dividend announcements, stock splits, and M&A announcements across US (SEC EDGAR), UK (GlobeNewswire/RNS), Canada (GlobeNewswire/CNW), and Australia (ASX Market Announcements).
Replaces Refinitiv Datastream ($10,000-22,000/yr), Bloomberg Terminal corporate actions module ($24,000/yr), and FactSet corporate events data ($8,000-15,000/yr) with a programmatic, agent-callable API.
What it is
| Signal | Source | Coverage |
|---|---|---|
| Earnings, dividends, splits, M&A | SEC EDGAR EFTS + submissions API | All public US companies, back to early 1990s |
| Regulatory announcements | GlobeNewswire UK RSS | UK/LSE-listed companies (sparse; FCA NSM requires auth) |
| Press releases, material changes | GlobeNewswire Canada RSS | TSX/TSX-V companies with ISIN and ticker |
| Market announcements | ASX MarkitDigital API | ASX200 + on-demand per code |
| Amendment resolution | In-process DB linkage | 8-K/A linked to original 8-K via amends_id |
No mobile device data. No scraping official registries. All sources are free and public.
Data Sources
| Source | Jurisdiction | Auth | Poll Interval |
|---|---|---|---|
| SEC EDGAR EFTS + submissions API | US | None (User-Agent header required) | Hourly |
| GlobeNewswire country/United-Kingdom RSS | UK | None | 2 hours |
| GlobeNewswire country/Canada RSS | CA | None | 12 hours |
| ASX MarkitDigital /companies/{code}/announcements | AU | None | 6 hours (rotating ASX200 batch) |
MCP Tools
get_company_events
Events for a specific company by ticker, ISIN, or company name substring. On-demand enrichment fires all four sources in parallel if the company is not cached.
Input: { "query": "AAPL", "jurisdiction": ["US"], "event_type": ["earnings"], "days_back": 90, "limit": 20 }
search_events
Filter events by type, jurisdiction, and date range. Served from SQLite — sub-10ms warm.
Input: { "event_type": ["dividend","earnings"], "jurisdiction": ["AU","CA"], "from_date": "2026-01-01", "limit": 50 }
get_recent_events
Latest events across all jurisdictions ordered by ingestion time.
Input: { "hours": 24, "event_type": ["merger_acquisition"] }
get_pipeline_health
Last poll time per source, event counts, status (ok/error/never). Verify freshness before time-sensitive queries.
Input: {} (no parameters)
fetch_historical_events
Historical 8-K filings for a US company from SEC EDGAR back to the early 1990s. Walks filings.files[] pagination. US only.
Input: { "query": "JPM", "from_date": "2015-01-01", "to_date": "2016-12-31", "limit": 200 }
Normalized Event Schema
{
"id": "a3f9b2c1d4e5f6a7b8c9d0e1",
"company": "Apple Inc.",
"ticker": "AAPL",
"isin": null,
"jurisdiction": "US",
"event_type": "earnings",
"event_date": null,
"announcement_date": "2026-04-30",
"source": "SEC_EDGAR",
"source_url": "https://www.sec.gov/Archives/edgar/data/320193/.../aapl-20260430.htm",
"raw_title": "8-K — 2.02,9.01",
"details": { "form_type": "8-K", "items": ["2.02", "9.01"], "accession": "0000320193-26-000011" },
"normalized_at": "2026-06-04T00:00:00.000Z",
"is_amendment": false,
"amends_id": null
}
Amendment Flagging
is_amendment: true is set when form type is 8-K/A. After upsert, resolveAmendments() queries the DB for the most recent prior 8-K from the same company with matching item types and sets amends_id to link them.
Environment Variables
| Variable | Required | Description |
|---|---|---|
PORT |
No (default 3000) | Server port |
DB_PATH |
No (default ./data/events.sqlite) | SQLite database path |
No external API keys required.
Running Locally
npm install
npm run dev
Ticker cache loads ~10 seconds after startup (10,365 US tickers). EDGAR polling runs immediately.
# Pipeline health
curl -s -X POST http://localhost:3000/mcp \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"get_pipeline_health","arguments":{}}}' \
| python3 -c "import sys,json; h=json.load(sys.stdin)['result']['structuredContent']; [print(f' {s[\"source\"]:12} {s[\"status\"]} events={s[\"events_indexed\"]}') for s in h['sources']]"
# AAPL events
curl -s -X POST http://localhost:3000/mcp \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"get_company_events","arguments":{"query":"AAPL","days_back":90}}}' \
| python3 -c "import sys,json; r=json.load(sys.stdin)['result']['structuredContent']; [print(e['announcement_date'], e['event_type'], e['raw_title'][:50]) for e in r['events'][:5]]"
# Historical depth
curl -s -X POST http://localhost:3000/mcp \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"fetch_historical_events","arguments":{"query":"JPM","from_date":"2015-01-01","to_date":"2016-12-31"}}}' \
| python3 -c "import sys,json; r=json.load(sys.stdin)['result']['structuredContent']; print('Total:', r['total_found'])"
Deployment (Railway)
Push to GitHub, connect repo in Railway. Set DB_PATH=/data/events.sqlite and add a volume mounted at /data to persist events across restarts.
Coverage Notes and Limitations
US (SEC EDGAR): Full coverage, hourly polling, historical depth to early 1990s.
Australia (ASX): ASX MarkitDigital API confirmed working at /companies/{code}/announcements. Background polling rotates 30 ASX200 codes in batches of 10 per cycle. On-demand fetch available for any ASX code.
Canada (GlobeNewswire): GlobeNewswire Canada RSS confirmed working with real TSX tickers and ISIN codes. SEDAR+ does not have a confirmed accessible public API endpoint.
UK (GlobeNewswire): GlobeNewswire UK country feed is sparse. FCA NSM returns HTTP 403. Investegate XML returns 404. UK coverage is best-effort — events_indexed may remain at 0 during quiet periods.
Not implemented: Exchange-validated settlement data (DTCC/CREST/CHESS are commercial), real-time intraday corrections, historical depth for non-US jurisdictions.
Not Financial Advice
This tool provides factual corporate event data from official and proxy sources. It does not constitute financial or investment advice.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。