QuantOracle

63 deterministic quantitative finance calculations via MCP. Options pricing, Greeks, implied volatility, exotic derivatives, risk metrics, portfolio optimiza...

Security Scan

The package and runtime instructions broadly match the stated purpose (an MCP-compatible quantitative-finance toolset), but there are a few configuration/name mismatches and small surprises you should review before running it in production.

Purpose & Capability

Name/description promise (63 quant tools via MCP) is implemented by the included Node code which exposes an MCP server built from an OpenAPI spec; declared required binary (node) and dependencies (@modelcontextprotocol/sdk, express) are consistent with that purpose.

Instruction Scope

SKILL.md tells agents to run `npx quantoracle-mcp` or connect to a remote MCP URL. The code implements an MCP server that fetches an OpenAPI spec from BACKEND_URL and exposes tool endpoints; the instructions do not ask the agent to read unrelated local files or secrets. However, SKILL.md and config-schema reference remote endpoints (mcp.quantoracle.dev / api.quantoracle.dev) while the code defaults BACKEND_URL to http://localhost:8001 — this mismatch could cause different runtime behavior depending on environment variables.

Install Mechanism

There is no high-risk remote download; sources are present in the package (package.json, dist/, src/). Dependencies are standard npm packages. No URL-shortener or extract-from-untrusted-URL install steps were found.

Credentials

The skill declares no required env vars but the code reads optional env vars (BACKEND_URL, PORT, FREE_DAILY_LIMIT, WALLET_ADDRESS). Those are plausible for configuring a service but the registry metadata does not document them; names and defaults also conflict with config-schema.json (which uses daily_limit/backend_url keys). A default WALLET_ADDRESS is present (likely for payment routing) — not a secret, but worth noting if you expect no crypto interaction.

Persistence & Privilege

The skill is not marked always:true and does not request elevated system-wide privileges. It runs a local MCP server and keeps in-memory rate counters; nothing in the package attempts to modify other skills or system-wide agent settings.

Assessment

This package appears to implement what it claims (an MCP server that maps an OpenAPI spec to 63 deterministic finance tools). Before installing or running: 1) Note the BACKEND_URL/port configuration — if you do not set BACKEND_URL, the code defaults to http://localhost:8001 (but SKILL.md/config mention api.quantoracle.dev / mcp.quantoracle.dev). That means the server you contact for tool definitions can change behavior based on your environment; review and set BACKEND_URL explicitly. 2) The code fetches openapi.json from BACKEND_URL at runtime; that remote API defines the tool schemas and endpoints — you should inspect that backend (or self-host it) if you need to ensure the tool definitions are safe and do not exfiltrate data. 3) A default WALLET_ADDRESS is baked into the code (likely for payments); it's not a secret but check it if you expect different billing behavior. 4) Run in an isolated/test environment first or override BACKEND_URL to a self-hosted backend if you need full network control. If you want, provide the fetched openapi.json or the truncated remainder of src/dist files and I can review the request/response handling and any forwarding/proxying logic in more detail.

63 deterministic quant computation tools for AI agents. Every tool accepts JSON and returns JSON. Same inputs always produce same outputs.

Install

npx quantoracle-mcp

Or connect directly via MCP:

https://mcp.quantoracle.dev/mcp

Tools

Options Pricing: Black-Scholes pricing with 10 Greeks (delta, gamma, theta, vega, rho, vanna, charm, volga, speed, color), implied volatility solver, multi-leg strategy builder, payoff diagrams.

Exotic Derivatives: Binomial tree, barrier options, lookback options, Asian options, volatility surface, option chain analysis, put-call parity.

Risk Metrics: Portfolio risk (Sharpe, Sortino, max drawdown, VaR, CVaR), Kelly criterion, position sizing, correlation analysis, stress testing, parametric VaR, transaction cost modeling.

Portfolio Optimization: Mean-variance (max Sharpe, min variance, target return), risk parity weights.

Monte Carlo Simulation: Geometric Brownian Motion with configurable paths, steps, and confidence intervals.

Statistics: Linear/polynomial regression, cointegration, Hurst exponent, GARCH forecasting, distribution fitting, correlation matrix, realized volatility, probabilistic Sharpe ratio, z-scores, normal distribution.

Technical Indicators: RSI, MACD, Bollinger Bands, ATR, Fibonacci retracement, crossover detection, regime detection.

Crypto/DeFi: Impermanent loss (v2/v3), liquidation price, funding rate analysis, DEX slippage, APY/APR conversion, vesting schedules, rebalance thresholds.

FX: Interest rate parity, purchasing power parity, forward rates, carry trade analysis.

Macro: Taylor Rule, Fisher equation, inflation-adjusted returns, real yield.

Time Value of Money: Present value, future value, NPV, IRR, CAGR.

Pricing

1,000 free calls per day. After that, pay-per-call via x402 (USDC on Base):

• $0.002 — Simple formulas (z-score, APY convert, TVM)
• $0.005 — Medium computation (Black-Scholes, Kelly, indicators)
• $0.008 — Complex computation (exotic derivatives, regression, GARCH)
• $0.015 — Heavy optimization (Monte Carlo, portfolio optimize, vol surface)

Usage

Ask the agent to use QuantOracle tools for any quantitative finance calculation. Examples:

• "Price a call option on AAPL at strike $200, spot $195, 30 days to expiry, 25% vol"
• "Calculate the optimal Kelly fraction for a strategy with 55% win rate, 1.2:1 reward-to-risk"
• "Run a Monte Carlo simulation of a $100 stock with 20% vol over 1 year"
• "What's the implied volatility if this option is trading at $5.50?"
• "Calculate impermanent loss for an ETH/USDC v3 position between $2000-$4000"