Market and company
Delayed/end-of-day quotes, price history, symbol search, profiles, fundamentals, earnings, analyst ratings, and ETF holdings.
Official public reference
Financial data MCP for AI agents.
Connect one hosted Streamable HTTP endpoint for delayed market data, company fundamentals, financial news, US macroeconomics, and US Congress transaction disclosures.
28published MCP tool contracts
22documented dataset contracts
17datasets currently available
6hpublic discovery monitor cadence
Research coverage
One endpoint, four financial research domains.
The public catalog distinguishes live datasets from planned coverage. Every response should preserve observation dates, reporting periods, units, freshness, warnings, and known limitations.
Financial news
Deduplicated summaries, ticker news, general market news, and generated direction, importance, and confidence fields.
Macroeconomics
Economic calendars, historical indicator observations, and generated current US macroeconomic themes.
US Congress disclosures
Normalized reported transactions and ticker-level aggregates with disclosed value ranges and filing-lag context.
Fastest connection
Give your agent one URL.
The hosted implementation remains private; no local package or third-party wrapper is required.
claude mcp add --transport http unquant https://unquant.ai/mcpGeneric remote MCP: use https://unquant.ai/mcp as the Streamable HTTP server URL in the client's native remote-MCP configuration.
Access: public discovery and a bounded anonymous Free allowance are available. OAuth may be required by the current dataset contract or usage state.
For machines and evaluators
Contracts that agents can inspect before connecting.
These public files contain metadata and schemas, not returned financial records. The production implementation and provider credentials remain private.
server.json
Canonical hosted MCP identity, version, transport, endpoint, and website.
tool-contracts.json
Names, descriptions, schemas, and read-only/idempotency annotations for 28 tools.
datasets.json
Availability, freshness, filters, limits, schemas, and known limitations for 22 datasets.
llms-full.txt
A static, AI-readable synopsis of connection, coverage, contracts, limitations, and canonical links.
Frequently asked questions
Clear boundaries for trustworthy research.
What is a financial data MCP server?
It exposes typed financial-data tools through the Model Context Protocol so an AI agent can discover inputs, call the appropriate dataset, and interpret a structured response.
Does Unquant provide real-time prices?
No. Quote and price-history responses are delayed or end-of-day unless a tool response explicitly states otherwise. They are not for execution.
Can an AI agent research US Congress transactions?
Yes. Available datasets expose normalized reported transactions and aggregate activity. Disclosures do not establish current holdings, intent, or causality.
Does Unquant execute trades or provide investment advice?
No. Unquant is a research-data service. It does not execute trades or provide personalized investment advice.
Where can an agent inspect tools before connecting?
Use the machine-readable snapshots on this page or call the standard MCP tools/list method at the hosted endpoint.