Financial Signals MCP

Derived financial intelligence for AI agents — not another SEC EDGAR filings server. Six of those already exist. This serves the interpreted layer: the patterns, surprises, flows, scores, and anomalies an agent actually needs to make a decision. Raw data is commodity; interpreted signals are premium.

Part of the FoundryNet Data Network. Every output carries verifiable provenance. See also: gov-contracts-mcp, brand-intel-mcp, patent-intel-mcp.

Built as a free-tier alternative to enterprise financial data (FactSet, Morningstar, S&P Capital IQ) — financial intelligence for agents without enterprise subscriptions.

Connect

• MCP endpoint (Streamable HTTP): https://financial-signals-mcp-production.up.railway.app/mcp

• Registry: io.github.FoundryNet/financial-signals-mcp

• Agent card: https://financial-signals-mcp-production.up.railway.app/.well-known/agent-card.json

Claude Desktop / Cursor / Claude Code

claude mcp add --transport http financial-signals https://financial-signals-mcp-production.up.railway.app/mcp

{ "mcpServers": { "financial-signals": { "url": "https://financial-signals-mcp-production.up.railway.app/mcp" } } }

Related MCP server: Toolstem MCP Server

Tools

Free tier: 25 paid-tool queries/day per agent (plus free macro_dashboard + mint_info). Then the tool returns an HTTP-402 with a metered payment challenge — pay it, re-call with the same args plus payment_tx=<reference>. An Authorization: Bearer fnet_… key bypasses the paywall.

The moat: composite_value_score

screen_stocks and company_profile expose a proprietary composite_value_score (0-100) — a transparent blend of: valuation vs. sector, growth, margin quality, insider sentiment, institutional momentum, and earnings consistency. Sort screen_stocks by it and you're using a ranking nobody else computes.

Sources & method

A daily run after US market close computes DERIVED tables (not raw filings) over the S&P 500 from free sources: yfinance (price, ratios, earnings, insider & institutional — Yahoo's aggregation of SEC data), FRED (macro), and SEC EDGAR (supplementary). Signal-typing + the composite score are computed at ingest.

Honesty notes: revenue surprise, guidance direction, and post-earnings move are not reliably free, so they're null/none where unavailable (roic uses ROA as a capital-efficiency proxy). Coverage grows as the daily universe is processed.

More

part of the FoundryNet Data Network.

Built by FoundryNet · forge@foundrynet.io

Live network activity

Live feed: mint.foundrynet.io/feed
Real-time verified work across 17 servers and autonomous agents, with cryptographic provenance.