Thesma MCP Server
Gives AI assistants access to SEC filings, BLS employment, Census demographics, and SBA lending data via natural language queries.
README
Thesma MCP Server
Give your AI assistant access to SEC, Census, BLS, and SBA data.
What it does
An MCP server that wraps the Thesma API, giving AI assistants (Claude, Cursor, ChatGPT) native access to SEC EDGAR filings — US-GAAP 10-K and IFRS 20-F, covering every US-listed public company on NYSE and NASDAQ (~6,000 companies) — plus Bureau of Labor Statistics employment data, US Census Bureau demographics, and SBA 7(a) small business lending. Ask questions in plain English, get structured data back.
Quick example
"What was Apple's revenue last year?"
The AI calls get_financials and returns Apple's income statement with formatted line items.
"What was Spotify's revenue last year?"
The AI calls get_financials for SPOT and returns Spotify's income statement in EUR with taxonomy: ifrs-full and by_nature presentation metadata.
"Find high-margin S&P 500 companies where insiders are buying"
The AI calls screen_companies with margin filters and insider buying signals.
"Which funds increased their position in NVDA last quarter?"
The AI calls get_holding_changes and shows quarter-over-quarter position changes.
"What's the average wage for software developers in Texas?"
The AI calls get_occupation_wages with SOC code 15-1252 and state filter, returning median and percentile wage data.
"What does Apple's labor market look like — hiring trends, local wages, compensation benchmarks?"
The AI calls get_company for AAPL, which automatically includes BLS labor market context alongside SEC company details.
"Give me a full picture of Apple — profile, financials, ratios, insider trades, top holders, executive compensation, and board."
The AI calls get_company('AAPL', include='financials,ratios,insider_trades,holders,compensation,board,labor_context,lending_context') and returns a composed view in one call instead of eight tool invocations.
Installation
pip install thesma-mcp
Claude Desktop
Add to your config file (~/Library/Application Support/Claude/claude_desktop_config.json on macOS, %APPDATA%\Claude\claude_desktop_config.json on Windows):
{
"mcpServers": {
"thesma": {
"command": "uvx",
"args": ["thesma-mcp"],
"env": {
"THESMA_API_KEY": "your-api-key"
}
}
}
}
Cursor
Add to ~/.cursor/mcp.json:
{
"mcpServers": {
"thesma": {
"command": "uvx",
"args": ["thesma-mcp"],
"env": {
"THESMA_API_KEY": "your-api-key"
}
}
}
}
Using
pip installinstead ofuvx? If you've already installedthesma-mcpwith pip, you can use"command": "thesma-mcp"directly (noargsneeded) instead ofuvx.
Get your API key at portal.thesma.dev (free tier: 250 requests/day).
Available tools
Company Discovery
| Tool | Description |
|---|---|
search_companies |
Find US public companies by name, ticker, index tier, exchange, domicile, taxonomy, or currency |
get_company |
Get company details — CIK, SIC code, fiscal year end, index membership, exchange, domicile |
Financial Statements
| Tool | Description |
|---|---|
get_financials |
Get income statement, balance sheet, or cash flow from SEC filings — US-GAAP 10-K and IFRS 20-F, returned in native currency |
get_financial_metric |
Get a single financial metric over time for trend analysis; per-point currency honors presentation-currency transitions |
Financial Ratios
| Tool | Description |
|---|---|
get_ratios |
Get computed financial ratios — margins, returns, leverage, growth |
get_ratio_history |
Get a single ratio over time for trend analysis |
Screening
| Tool | Description |
|---|---|
screen_companies |
Find companies matching financial criteria — profitability, growth, leverage, index tier, SIC, exchange, domicile, taxonomy, currency, insider/institutional signals, labor market, SBA lending (county + industry) |
Corporate Events
| Tool | Description |
|---|---|
get_events |
Get 8-K corporate events — earnings, M&A, leadership changes, material agreements |
Insider Trading
| Tool | Description |
|---|---|
get_insider_trades |
Get Form 4 insider transactions — purchases, sales, grants, option exercises |
Institutional Holdings
| Tool | Description |
|---|---|
search_funds |
Find institutional investment managers (hedge funds, mutual funds) by name |
get_institutional_holders |
Get which funds hold a company's stock with shares and market values |
get_fund_holdings |
Get a fund's portfolio — what stocks it owns |
get_holding_changes |
Get quarter-over-quarter changes in institutional positions |
Compensation & Governance
| Tool | Description |
|---|---|
get_executive_compensation |
Get executive pay — salary, bonus, stock awards, total, CEO pay ratio |
get_board_members |
Get board of directors — age, tenure, independence, committee memberships |
Filings
| Tool | Description |
|---|---|
search_filings |
Search SEC filings by company, type (10-K, 10-Q, 8-K, etc.), and date range |
Filing Section Search
| Tool | Description |
|---|---|
search_filing_sections |
Semantic search of SEC filing section content (Risk Factors, MD&A, etc.) by natural-language query — optionally scoped by ticker, filing type, section type, year, or similarity threshold |
Webhooks
| Tool | Description |
|---|---|
list_webhook_event_types |
List the 5 event types you can subscribe to |
list_webhooks |
List your webhook subscriptions |
create_webhook |
Create a new subscription (returns one-time HMAC secret) |
get_webhook |
Get details for one subscription |
update_webhook |
Modify url / events / filing_types / is_active / description |
delete_webhook |
Delete a subscription |
list_webhook_deliveries |
Recent delivery attempts (debugging) |
rotate_webhook_secret |
Generate a new HMAC secret (invalidates old) |
send_webhook_test |
Enqueue a synthetic test delivery |
replay_webhook_delivery |
Re-queue a past delivery (within 7-day window) |
Industry Lookup
| Tool | Description |
|---|---|
search_industries |
Find BLS industries by name or NAICS level |
get_industry_detail |
Get industry details — child industries, data availability across CES/QCEW/OEWS |
Industry Employment (CES)
| Tool | Description |
|---|---|
get_industry_employment |
Get employment, earnings, and hours data for an industry by NAICS code |
County Employment & Wages (QCEW)
| Tool | Description |
|---|---|
get_county_employment |
Get quarterly employment data for a US county by FIPS code |
get_county_wages |
Get county wage snapshot with location quotients vs. national average |
Occupation Wages (OEWS)
| Tool | Description |
|---|---|
search_occupations |
Find BLS occupations by name or SOC group |
get_occupation_wages |
Get occupation wage data — mean, median, and percentile distribution |
Labor Market Turnover (JOLTS)
| Tool | Description |
|---|---|
get_industry_turnover |
Get job openings, hires, quits, and layoffs for an industry |
get_state_turnover |
Get state-level labor market turnover (total nonfarm) |
get_regional_turnover |
Get turnover data for a Census region (Northeast, South, Midwest, West) |
Local Area Unemployment (LAUS)
| Tool | Description |
|---|---|
get_county_unemployment |
Get monthly unemployment rate, labor force, and employment for a US county (never seasonally adjusted) |
compare_county_unemployment |
Compare unemployment metrics across up to 10 counties in a single period |
get_state_unemployment |
Get monthly state unemployment with labor force participation rate and employment-population ratio (SA or NSA) |
compare_state_unemployment |
Compare unemployment metrics across up to 10 states in a single period |
SBA 7(a) Lending
| Tool | Description |
|---|---|
get_county_lending |
Get quarterly SBA 7(a) loan aggregates for a US county (loan count, total amount, charge-off rate) |
get_state_lending |
Get quarterly SBA 7(a) loan aggregates for a US state |
get_industry_lending |
Get quarterly SBA 7(a) loan aggregates for a NAICS industry, optionally scoped to state or county |
get_lenders |
List SBA 7(a) lenders ranked by loan count, total amount, or average loan size |
get_lender |
Get details and quarterly history for a single SBA lender by lender_id |
get_lending_characteristics |
Get loan size, term length, interest rate, sub-programme, and business-type distributions for a quarter |
get_lending_outcomes |
Get vintage cohort charge-off rates and maturity status |
explore_sba_metrics |
Browse available SBA metrics by category or keyword |
get_sba_metric_detail |
Get full definition, unit, cadence, and data availability for a single SBA metric |
BLS Discovery
| Tool | Description |
|---|---|
explore_bls_metrics |
Browse available BLS metrics by category, source, or keyword |
Configuration
| Variable | Required | Description |
|---|---|---|
THESMA_API_KEY |
Yes | API key from portal.thesma.dev |
THESMA_API_URL |
No | Override API base URL (default: https://api.thesma.dev) |
Data coverage
- ~6,000 US-listed public companies on NYSE and NASDAQ — US-GAAP filers plus IFRS 20-F filers (Spotify, Nu Holdings, ASML, and other US-listed IFRS reporters). Native-currency reporting; no USD normalization.
- SEC EDGAR: financial statements (2009-present), insider trades, institutional holdings, executive compensation, board data, corporate events, filings
- Bureau of Labor Statistics: industry employment (CES), county wages (QCEW), occupation wages (OEWS), job openings and turnover (JOLTS), local unemployment (LAUS)
- Small Business Administration: 7(a) loan program data — quarterly lending aggregates by county/state/industry, lender rankings, loan characteristics (size, term, rate, sub-programme), vintage cohort charge-off outcomes
- Labor & lending market enrichment:
get_companyautomatically includes BLS labor context and SBA 7(a) lending context;screen_companiessupports labor market and SBA lending filters - Known gaps: IFRS bank filings (Nu Holdings, HSBC, Deutsche Bank, etc.) — the core banking-statement extensions (
net_interest_income, etc.) are US-GAAP-specific today; IFRS banking fields land in a future release. - All data sourced from US federal public-domain sources: SEC EDGAR, US Census Bureau, Bureau of Labor Statistics, Small Business Administration
Links
License
MIT
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。
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。
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。
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。