Local first
The first MCP server is a local stdio proxy. It forwards tool calls to the client API with your API key.
The QuantConomy MCP track gives AI assistants a read-only path into market news, SEC filings, assets, markets, prediction markets, and ranked signals. It is being built as a separate access layer on top of the client API, with source trails visible enough for humans to audit what the agent used.
The first MCP server is a local stdio proxy. It forwards tool calls to the client API with your API key.
A hosted remote MCP server should be treated as a separate launch track with auth, rate limits, and security review.
The early tools read market context. Execution, orders, and write actions stay out of the public MCP beta.
MCP should call the same client API as direct users, so plans, credits, and permissions stay consistent.
The first commercial value is research speed: ask an assistant to find the relevant filings, signals, and entries, then inspect the source trail yourself.
Search news entries with sentiment and entity context
List Oracle signals with direction, strength, and expiry
Read SEC insider trades, 8-K items, 13F filings, and Form 144 records
Fetch market candles, assets, clusters, publishers, and prediction markets
Lock the external API base URL and remove placeholder domains.
Publish an install path for the local MCP package.
Return structured MCP content where clients can parse it reliably.
Add resources for reusable records such as signals, filings, clusters, and assets.
Add prompts for common research flows such as "explain this signal" and "summarize this filing trail."
Prepare a remote MCP server only after OAuth or a secure API-key connection flow is ready.
Document tool scopes, credit costs, and safe usage examples.