Tools
Read-only tools across News, Markets, Signals, SEC datasets, and Account — search news, resolve assets and markets, pull Oracle signals, and query every SEC filing dataset.
The Model Context Protocol (MCP) is an open standard that lets an AI agent discover and call external tools, read resources, and run prompt templates through a uniform interface — instead of hand-wiring HTTP calls into the model. The QuantConomy MCP server speaks that protocol, so any MCP-capable client can use the QuantConomy API as a first-class set of tools.
The server wraps the QuantConomy REST API and exposes it as MCP primitives, so an agent can answer market questions directly from the data without you writing any glue code:
Tools
Read-only tools across News, Markets, Signals, SEC datasets, and Account — search news, resolve assets and markets, pull Oracle signals, and query every SEC filing dataset.
Resources
Stable URIs the agent can attach as context: quantconomy://account (plan, credits, limits)
and quantconomy://stats (system-wide counts and last-update times).
Prompts
Parameterised research playbooks — market-brief and company-due-diligence — that chain
the tools into a single workflow.
See the full list on the Tools & resources page.
The server is hosted at https://mcp.quantconomy.com/mcp and speaks Streamable HTTP. It is
fully stateless: every request builds a fresh session from its own
Authorization: Bearer mtk_... header, handles the request, and disposes — so your API key is
supplied per request and is never stored between calls. There is also a GET /health
endpoint. Any MCP client that can connect to a remote server URL can use it.
The MCP server adds no data of its own. Every tool forwards to the REST API at
https://api.quantconomy.com/api/v1 using your API key, so all of the platform’s rules
apply identically whether you call REST directly or go through MCP:
Every tool is a GET, so the agent can only read — it can never create, modify, or delete.
Any MCP-capable client that can connect to a remote server URL — including Claude Desktop, Claude Code, Cursor, and ChatGPT. See Connect a client for per-client setup.