Skip to main content

Financial data MCP server for Claude, Cursor, and Continue

Jintel runs a hosted Model Context Protocol (MCP) server at https://api.jintel.ai/mcp over the Streamable HTTP transport. It exposes 15 typed finance tools that any MCP-compatible runtime — Claude Desktop, Cursor, Continue, Cline, Zed, Windsurf, Raycast AI, the OpenAI Responses API, LangChain, LlamaIndex — can call natively, with the same schema and the same Bearer / x402 auth.

What you get

  • 15 tools covering market quotes, batch entity enrichment, regulatory filings (10-K / 10-Q / 13F / Exhibit 21), OFAC sanctions screening, macro indicators (GDP / CPI / interest rates / FRED), technicals (RSI / MACD / Bollinger Bands), news, social sentiment, short interest, campaign finance, S&P 500 valuation multiples, and the full GraphQL fan-out for nested queries the catalog doesn't pre-bake.
  • Streamable HTTP transport — no local install for the remote variant; agents discover the server at /.well-known/mcp.json and the richer server card which inlines the full tool catalog with input schemas.
  • Read-only annotations on every tool (readOnlyHint: true, destructiveHint: false) so Claude Desktop and other hosts can skip the human-confirmation prompt that fires for unannotated tools.
  • Free initialize and tools/list so agents can browse the catalog before binding a key — the only paid path is tools/call.

Install in one step

The five most popular runtimes:

Claude Desktop

{
"mcpServers": {
"jintel": {
"transport": "streamable-http",
"url": "https://api.jintel.ai/mcp",
"headers": { "Authorization": "Bearer jk_live_your_key_here" }
}
}
}

Cursor

{
"mcpServers": {
"jintel": {
"command": "npx",
"args": ["-y", "@yojinhq/jintel-mcp"],
"env": { "JINTEL_API_KEY": "jk_live_your_key_here" }
}
}
}

Continue

mcpServers:
- name: jintel
type: streamableHttp
url: https://api.jintel.ai/mcp
headers:
Authorization: Bearer jk_live_your_key_here

Zed

{
"context_servers": {
"jintel": {
"command": {
"path": "npx",
"args": ["-y", "@yojinhq/jintel-mcp"],
"env": { "JINTEL_API_KEY": "jk_live_your_key_here" }
}
}
}
}

OpenAI Responses API

const res = await openai.responses.create({
model: 'gpt-4.1',
tools: [{
type: 'mcp',
server_url: 'https://api.jintel.ai/mcp',
server_label: 'jintel',
authorization: 'Bearer jk_live_your_key_here',
}],
input: 'Latest quote and 30-day RSI for NVDA',
});

The full table of platforms (Cline, Windsurf, Raycast AI, LangChain, LlamaIndex, generic OpenAPI) is in Install Jintel in your AI tools.

Tools at a glance

ToolPurpose
quotesBatch quotes by ticker (equities + crypto)
searchEntitiesFan-out search across market connectors
priceHistoryBatch OHLCV history
entityByTickerSingle entity with sub-graph traversal
entitiesByTickersBatch entity enrichment
marketStatusEquity / crypto market open-close state
sanctionsScreenOFAC SDN name-match screening
institutionalHoldingsSEC 13F holdings by filer CIK
gdp / inflation / interestRatesOECD macro series
macroSeriesArbitrary FRED time-series by series ID
shortInterestShort interest data via FINRA
campaignFinanceUS PAC / committee filings via OpenFEC
sp500MultiplesS&P 500 valuation multiples via Multpl

Why MCP

  • Native tool calls. Claude / Cursor / Continue treat MCP servers as first-class function-calling targets. No glue code, no JSON parsing, no rate-limit retries — the runtime handles all of it.
  • Typed inputs. Every tool ships a JSON Schema for its arguments. Agents see exactly which tickers, dates, or filter knobs are valid before issuing the call.
  • Discoverable catalog. tools/list returns descriptions and schemas. Combined with /.well-known/mcp.json, agent registries (orank, openrouter, mcp.so) can index Jintel without bespoke connectors.

Auth options

  • Bearer API keyAuthorization: Bearer jk_live_…. Issue via Google OAuth or wallet sign-in (/auth/siwx/verify).
  • x402 pay-per-query — drop the Bearer header, sign each tools/call payment quote with EIP-3009 USDC on Base mainnet. No signup; the wallet identifies the caller.

See Authentication and Agents & x402.