I. The market moved. The tooling didn't.
Language models can reason about volatility surfaces, greeks, and skew better than
most junior traders. But ask one to actually fetch a chain and it's stuck
choosing between three incompatible MCP servers, two of which rate-limit at the worst
possible moment and one of which was abandoned the week it shipped.
II. Wrapping isn't a workaround. It's the fix.
OptionsMCP doesn't ask you to pick a winner. It sits above every options MCP server
that already exists — CBOE feeds, retail broker APIs, third-party greeks engines,
historical IV archives — and speaks one contract to your agent. Add a source behind
the wrapper, not a new integration in every agent you own.
III. One schema. Every strike, every expiry, every greek.
Normalized contracts. Normalized greeks. Normalized error handling. When a source goes
down, OptionsMCP fails over to the next one in the chain — silently, before your
agent even notices. Your strategy code never learns a new vendor's quirks again.
IV. Built for agents, not dashboards.
This isn't a UI bolted onto an API. It's a Model Context Protocol server, designed
from the first line for tool-calling loops: cheap tokens, typed responses, and
pagination that doesn't blow your context window on a single SPY chain.
V. We believe consolidation is inevitable.
Every data category in finance ends up with one dominant access layer. Market data
had it. Fundamentals had it. Options data — the most fragmented corner of derivatives
infrastructure — is next. We're building the layer, not another silo.