An AI stock screener lets you describe the stocks you want in plain English and get back a ranked table built from real financial data. This guide shows you how to build one in Claude without writing code, by connecting Claude to the FinancialData.Net MCP server.
- Time to complete: about 10 minutes
- Skill level: beginner, no programming required
- You will build: a Claude connector that screens stocks on valuation, profitability, balance-sheet strength, insider activity and more
Why does Claude need a data connection? A chatbot without a data connection answers from training data, which goes stale. With the MCP connector, Claude calls live data tools and builds the answer from the numbers they return.
What is MCP, in one paragraph
The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external services and use their tools. The FinancialData.Net MCP server makes financial data available directly in Claude, including stock prices, company financial statements, ETF data, and more. When you ask a question, Claude automatically selects the relevant tools, retrieves the data, and uses it to generate an answer. You don't need to write code or manually select data sources.
What you need
| Requirement | Details |
|---|---|
| Claude account | Any plan that supports custom connectors. On Team and Enterprise plans, an organization owner adds the connector first. |
| FinancialData.Net subscription | Professional or Enterprise. The MCP server is not available on other plans. |
| Your API key | A 32-character key from your FinancialData.Net dashboard. |
Step 1: Get your API key
- Sign up or sign in to FinancialData.Net.
- Subscribe to a Professional or Enterprise plan on the pricing page.
- Open your dashboard and copy your API key.
Step 2: Add the connector in Claude
Open Claude on the web or desktop app and go to Settings.
Find the Connectors section. At the time of writing, this is under Customize → Connectors.
Choose Add custom connector (the button may be labeled "Add" or "Add custom connector").
Enter the following:
Field Value Name FinancialDataMCP server URL https://financialdata.net/mcp?key=API_KEYReplace
API_KEYwith the key from Step 1, then select Add.In a new chat, select the + button, open Connectors, and turn on FinancialData.
Approvals and permissions: Claude asks for approval before using tools from a custom connector. You can review or change which tools are enabled in the connector's settings.
Can't find it? Claude's interface changes over time. See Anthropic's custom connector guide for the current steps.
Step 3: Verify the connection
Send this prompt:
List the available tools from the FinancialData server.
A working connection returns a list of tools such as getStockPrices, getKeyMetrics and getIncomeStatements. If you see them, the screener is ready.
Step 4: Run your first screen
The screener works best when you give Claude three things: a universe (which stocks), criteria (what to filter on) and an output format (how to show results).
Screen these stocks: MSFT, AAPL, GOOG.
Criteria: PEG ratio below 1, return on equity above 20%, debt-to-equity below 0.5.
Use the latest fiscal year available. Show a table with P/E, forward P/E,
PEG, ROE, debt-to-equity and free cash flow, then list which stocks pass
and which fail each criterion.
Behind the scenes, Claude calls getKeyMetrics for each ticker, applies your filters and builds the table.
Example output
The figures below were retrieved through the FinancialData.net MCP using the latest fiscal year available for each company at the time. Your results will differ as the data updates.
| Ticker | Fiscal year | P/E | Forward P/E | PEG | ROE | Debt/Equity | Free cash flow | Result |
|---|---|---|---|---|---|---|---|---|
| MSFT | FY2026 | 20.7 | 19.0 | 0.66 | 30.2% | 0.11 | $36.3B | Pass |
| GOOG | FY2025 | 28.8 | 15.2 | 0.84 | 31.8% | 0.12 | $68.0B | Pass |
| AAPL | FY2026 | 37.9 | 35.7 | 1.87 | 117.9% | 0.79 | $129.4B | Fail (PEG, debt/equity) |
Build your universe
To screen a group of stocks, ask Claude to build the list first:
Get the constituents of the S&P 500 and show me the Information Technology companies.
Then screen the first 20 for earnings growth above 15% and debt-to-equity below 0.5.
This uses getIndexConstituents to define the universe, then getKeyMetrics per company.
How screening works: MCP tools return data per security. For large universes, split the work into batches of 20 to 50 tickers per prompt, and name the universe explicitly.
10 screening prompts you can copy
| Screen | Prompt | Tools Claude uses |
|---|---|---|
| Value | "For AAPL, MSFT, GOOG and AMZN, rank the stocks by PEG ratio and show P/E, price-to-book and free cash flow." | getKeyMetrics |
| Valuation vs. history | "Compare Eli Lilly's current P/E to its five-year historical average. Is it above or below?" | getKeyMetrics |
| Profitability | "Show gross margin, operating margin and return on assets for MSFT over the last five fiscal years. Is profitability improving?" | getProfitabilityRatios |
| Balance-sheet strength | "Among Apple, Ford, AT&T, Boeing and Coca-Cola, flag any company with a current ratio below 1 or a debt-to-equity ratio above 1." | getLiquidityRatios, getSolvencyRatios |
| Growth | "Compare revenue and net income growth for AAPL and MSFT over the last four years." | getIncomeStatements |
| Dividends | "Show dividend yield and payout ratio for KO, PEP and PG. Which has the most room to raise its dividend?" | getKeyMetrics |
| Insider activity | "Show recent insider transactions for NVDA and summarize buying versus selling." | getInsiderTransactions |
| Institutional ownership | "For Vanguard, BlackRock and State Street, show their most recent AAPL position and how it changed from the prior quarter." | getInstitutionalHoldings |
| Upcoming catalysts | "Which S&P 500 companies report earnings next week?" | getIndexConstituents, getEarningsCalendar |
| Relative performance | "Compare NVIDIA's trailing 12-month return to the S&P 500's and calculate the excess return." | getStockPrices, getIndexPrices |
Every endpoint in the FinancialData.Net documentation is available as a tool, including ETFs, mutual funds, crypto, forex, commodities, ESG scores, analyst consensus and economic data. You can combine them in a single prompt.
Make your screener repeatable
Retyping criteria each time is slow. Save them once:
- Create a Claude Project for your screener.
- Add project instructions with your screening rules, for example: "Use the FinancialData connector for all financial data. Always state the fiscal year used. Show results as a table. Flag any metric that is missing."
- Reuse your prompts. Paste the criteria that work into the project and rerun them on new universes.
What data is behind the screener
| Dataset | Coverage |
|---|---|
| Symbols (stocks, ETFs, options, futures, crypto, forex, mutual funds) | 252,000+ tickers across 20+ exchanges |
| End-of-day and intraday prices | Since 2015 |
| Company information and fundamentals (statements, ratios) | Since 2010 |
| Press releases, market news | Since 2006 |
| Insider and institutional trading | Since 2010 |
| Economic data | Since 1854 |
International fundamentals cover exchanges including Toronto, London, Frankfurt, Euronext Paris, Euronext Amsterdam, Tokyo, Hong Kong, Singapore, Korea, Brazil, Mexico and India.
FAQ
Do I need to know how to code? No. You add one URL in Claude's settings and write your screens in plain English.
Which plan do I need on FinancialData.Net? Professional or Enterprise. The MCP server is not included in the Free, Standard or Premium plans.
Does it work with AI tools other than Claude? Yes. The same server URL works in any MCP-compatible client.
Can Claude screen the whole market in one prompt? MCP tools return data per security, so screen a defined universe such as an index, a sector or your watchlist in batches.
Is this investment advice? No. Screens produce candidates for further research. Verify data, read company filings and consider your own situation before making decisions.