fAlpha MCP¶

Model-scored US equity research for agents, strictly read-only. fAlpha serves its own model signal for a stock together with the drivers behind it, plus the public data around it: signal history and flips, news sentiment, analyst coverage, SEC filings, FRED macro and fundamentals. Every result states which source produced it and the date the data is for. The server never places orders.

Server type: Remote (Streamable HTTP)
Auth: OAuth 2.1 (fAlpha account) or personal access token
Endpoint: https://agent.falpha.ai/mcp
Tools: 17, all read-only
Pricing: no-card trial token; account plans after
Category: Finance
Built by: fAlpha (falpha.ai); agent kit on GitHub (MIT)

Why This Matters for Operators¶

Equity research workflows usually mean juggling terminals, EDGAR and news feeds. fAlpha packages them behind one endpoint with a distinctive addition: the vendor's own model signal with the drivers behind it, signal history and flips, and a screener over the covered universe. Each result is dated and source-attributed, and the server is read-only by design, so an agent cannot trade.

For operators and advisors, the practical workflows are comparison and monitoring: compare two tickers, check when a signal last flipped and what changed, screen sectors with a positive signal and improving sentiment, and summarize a 10-Q alongside its signal. No-card trial tokens are available, which keeps first-run evaluation free.

Tools & Capabilities¶

Capability group	What it covers
Signal	Current card, compare two tickers, history on a past date, performance, flips, term structure
Screener	Screen the covered universe, coverage check
News sentiment	News sentiment and its trend
Analyst coverage	Analyst coverage and price targets
SEC filings	Filings and XBRL company facts
FRED macro	Macroeconomic backdrop
Company profile	Company profile and ratios
Glossary	fAlpha terms, and a link to open the view on falpha.ai

Data sources: fAlpha's own model signal and news sentiment, Financial Modeling Prep (prices, analyst targets, fundamentals), SEC EDGAR (filings, XBRL) and FRED (macro).

Installation¶

claude mcp add --transport http falpha https://agent.falpha.ai/mcp

Then run /mcp to sign in, or attach the account token in the Authorization header (Bearer scheme). A no-card trial token is issued at falpha.ai/mcp/trial. Claude Desktop connects via Customize, then Connectors; ChatGPT via Developer mode, then Plugins.

Configuration¶

{
  "mcpServers": {
    "falpha": {
      "type": "http",
      "url": "https://agent.falpha.ai/mcp"
    }
  }
}

Business Relevance¶

Operators and founders track a watchlist against model signal and news sentiment
Advisors compare tickers and summarize filings alongside their signal in one call
Analysts screen sectors by signal and sentiment before deeper work
Compliance-conscious teams get read-only access by construction: the server cannot place orders

Integration with CorpusIQ¶

fAlpha adds public-market context to the private business data CorpusIQ answers from. A composed workflow: CorpusIQ reports revenue and margin trends from QuickBooks or Stripe, and fAlpha supplies the same company's public signal, analyst coverage and latest 10-Q summary. The operator reads private and public pictures side by side in one session.

Limitations¶

Read-only by design: no trading, no portfolio actions
US equity coverage only
The signal is model output, not investment advice
Ongoing use requires an fAlpha account beyond the trial token

FAQ¶

Does fAlpha trade?¶

No. All 17 tools are read-only and the server never places orders.

Is there a free way to try it?¶

Yes. A no-card trial token is issued at falpha.ai/mcp/trial; ongoing use needs an fAlpha account.

Is the signal investment advice?¶

No. The signal is model output and every result states which source produced it and the date the data is for.