Developers Hub
Build with Options Analysis Suite
Three programmatic surfaces for the OAS analytics engine. Pick whichever one fits your runtime: a typed Python client for notebooks, scripts, and research pipelines; a Model Context Protocol (MCP) server that drops into ChatGPT, Claude, Perplexity, and Grok so an AI agent can call OAS analytics directly; or a REST API with an OpenAPI 3.1 spec for anything that speaks HTTP.
Python SDK
Type-safe httpx + Pydantic v2 client wrapping 17 pricing models (Black-Scholes, Heston, SABR, Variance Gamma, Jump Diffusion, Local Volatility, FFT, PDE, Binomial, and 8 more), full 17-Greek output, GEX/DEX exposure, IV surfaces, and 5-model calibration. 49 typed methods, drift-checked against the deployed OpenAPI spec so a stale SDK fails CI before it can ship.
- Install:
pip install options-analysis-suite - Package: PyPI
- Repository: github.com/Options-Analysis-Suite/options-analysis-suite-python
- Detailed docs: Python SDK page
MCP Server
The Model Context Protocol server exposes the OAS analytics engine as native tools that ChatGPT, Claude, Perplexity, and Grok can call directly. When a user asks an AI assistant a question about options pricing, GEX, or model calibration, the assistant queries OAS in real time instead of guessing from training data.
- Public mirror: github.com/Options-Analysis-Suite/options-analysis-suite-mcp
- Detailed docs: MCP Server page
REST API
Direct HTTP access for any runtime that speaks JSON. Same endpoints the Python SDK wraps. OpenAPI 3.1 spec is published at data.optionsanalysissuite.com/openapi.json; the interactive docs are at data.optionsanalysissuite.com/docs. Authentication is API-key based.
- Base URL:
https://data.optionsanalysissuite.com - OpenAPI spec: openapi.json
- Interactive docs: Scalar-rendered API reference
- Reference page: API access overview
Which surface should I use?
The three surfaces sit at different levels of the same stack. The REST API is the lowest layer - every other client eventually calls it. The Python SDK is the most ergonomic for any Python code; type signatures, structured errors, calibration object persistence, BYOK credential helpers. The MCP server is for AI-agent integration: drop OAS into ChatGPT, Claude, Perplexity, or Grok and the agent gets options analytics as native tools without writing any glue.
- Notebooks, scripts, backtests, research: Python SDK. Type-safe, persistent calibration, full coverage.
- AI agent workflow, ChatGPT, Claude, Perplexity, Grok: MCP server. Native tool integration; no code required.
- Non-Python runtime, custom integration, raw HTTP: REST API. Same endpoints the SDK wraps.
What you can build
- Calibration pipelines. Fit Heston, SABR, Variance Gamma, Jump Diffusion, or Local Vol against the live chain via your broker's data (BYOK). Persist fitted parameters to disk; reload anywhere; evaluate fair value or full Greeks at any strike/expiration without re-fitting.
- Daily watchlist analytics. Pull IV rank, max pain, expected move, GEX, P/C ratio, and 50+ metrics across a watchlist in one batched call. Cached at the edge so the same call costs minimal compute on subsequent fetches.
- Backtesting frameworks. Historical Greeks, IV surface snapshots, regime classifications, ATS/OTC volume, FINRA short data, SEC EDGAR filings, FRED rates - typed access to the full data layer.
- AI agent tools. Drop the MCP server into Claude and an LLM can answer "what's the GEX flip on SPY today" with real numbers, fit Heston against AAPL's chain, or compute the iron-condor breakevens for any custom strikes.
Authentication and pricing
The Python SDK and REST API are both backed by the API tier and use API-key authentication. The MCP server is different: it is included with a Pro (Professional) subscription, not gated to the API tier, and authenticates via your Options Analysis Suite email and password rather than an API key (OAuth for Claude, ChatGPT, and Grok; base64 credentials for Perplexity). See pricing for tier details and the MCP server page for per-client setup. End-of-day data is included on both tiers; real-time data requires a BYOK broker integration (Tradier, tastytrade, or Public).