Live — Processing market data now

Markets Brought to You
by AI

The MCP server for market intelligence. Connect Claude, GPT, Gemini, or any AI agent to 14 live data sources — cross-exchange signals, whale detection, and natural language narratives in ~300 tokens.

>"Any whale activity on BTC in the last hour?"

Get Started — FreeorRead the Docs →

Data Sources

~300

Tokens / Symbol

Signal Types

<1ms

Analytics Latency

The Problem

Raw Exchange Data Wastes Your Context Window

AI agents spend most of their token budget parsing verbose JSON instead of reasoning about market conditions.

Without Market.ai

Every exchange speaks a different format. A single orderbook snapshot consumes 10,000+ tokens of nested arrays and metadata.

-10,000+ tokens for one orderbook snapshot
-12 different JSON schemas to parse
-No cross-exchange aggregation
-No signal detection or anomaly alerts

Raw Exchange Data

10,000+

tokens / symbol

With Market.ai

One normalized API delivers pre-digested context packets. Price action, orderbook intelligence, cross-exchange signals — all in ~300 tokens.

+~300 tokens per symbol, fully analyzed
+One unified schema across all exchanges
+Cross-exchange aggregation + arb detection
+10 signal types with severity levels

Market.ai Context

~300

tokens / symbol

Why Market.ai

Intelligence No Single Exchange Can Provide

Cross-exchange signals, AI-native delivery, and natural language narratives. The market intelligence layer your AI agent is missing.

Cross-Exchange Signals

10 signal types no single exchange can show: whale movements across venues, liquidation cascades, cross-exchange funding spikes, orderbook imbalances, spread anomalies, and volume surges. Rolling-window anomaly detection with severity levels.

AI-Native via MCP

Not just an API — a Model Context Protocol server. Your AI gets market intelligence as native tool calls. 10,000+ tokens of raw exchange data compressed into ~300-token context packets your LLM can actually use.

Arbitrage Detection

Real-time cross-exchange price divergence monitoring across 14 venues. Spot arb opportunities by spread in basis points, with the exact exchange pair and direction. Filter by minimum spread threshold.

Polymarket Predictions

Crypto Up/Down probabilities for BTC, ETH, SOL, and XRP across 5-minute to daily timeframes. Real-time orderbook, pricing, and resolution data from Polymarket's CLOB — a sentiment signal no exchange provides.

Natural Language Narratives

Every data point comes with a human-readable market narrative. "BTC rallying +2.1%, strong buy pressure (OBI 0.23), 42.5 BTC bid wall at 67k." Your AI reads context, not raw numbers.

Conversational Trading

"Buy 0.5 BTC at market" — natural language trade execution with confirmation flows, portfolio queries, and safety rails. Chat via web, Discord, or Telegram with broker-connected AI.

Bring Your Own AI

Works With Every AI Platform

Market.ai is an MCP server. Connect it to Claude, ChatGPT, Gemini, or any MCP-compatible client. Your AI gets live market intelligence as native tool calls — no custom integration code needed.

Claude

Native MCP support in Claude Desktop and Claude Code

ChatGPT / OpenAI

Connect via MCP or use the REST & WebSocket APIs directly

Gemini

Google's AI with Market.ai context via MCP integration

Any MCP Client

Cursor, Windsurf, custom agents — anything that speaks MCP

claude_desktop_config.json

{
  "mcpServers": {
    "market-ai": {
      "url": "https://mcp.market-ai.dev/sse",
      "headers": { "Authorization": "Bearer mk_live_..." }
    }
  }
}

API

One Request, Complete Market Context

A single API call returns everything your AI agent needs — in ~300 tokens instead of 10,000+.

Request

curl -H "Authorization: Bearer mk_live_YOUR_KEY" \
  https://api.market-ai.dev/v1/quotes/BTC/USDT

MCP — Claude Desktop Config

{
  "mcpServers": {
    "market-ai": {
      "url": "https://mcp.market-ai.dev/sse",
      "headers": {
        "Authorization": "Bearer mk_live_YOUR_KEY"
      }
    }
  }
}

Response — ~300 tokens

{
  "ok": true,
  "data": {
    "sym": "BTC/USDT",
    "px": 67452.31,
    "chg": 2.14,
    "vol": "1.2B",
    "obi5": 0.23,
    "spread": 0.8,
    "walls": {
      "bid": [{"px": 67000, "qty": "42.5", "x": 8.2}]
    },
    "xchg": [
      {"n": "binance", "px": 67451.80, "vol": "485M"},
      {"n": "coinbase", "px": 67453.10, "vol": "312M"},
      {"n": "bybit", "px": 67452.00, "vol": "198M"}
    ],
    "trend": "Bullish",
    "narrative": "BTC rallying +2.1%, strong buy pressure (OBI 0.23), 42.5 BTC bid wall at 67000 (8.2x median)."
  }
}

How It Works

Three Layers, One Feed

Direct exchange connections feed into a Rust analytics engine that publishes compressed context to your AI agent.

Connect

Direct WebSocket connections to 12 crypto exchanges, NASDAQ via Alpaca, and Polymarket. No CCXT. No third-party aggregators. Raw data at the source.

Process

Rust core engine manages orderbooks, computes analytics (OBI, VWAP, walls, depth), aggregates cross-exchange data, and detects anomalies in sub-millisecond time.

Deliver

Token-efficient context packets via MCP Server for Claude/Cursor, WebSocket push for real-time feeds, or REST API for on-demand queries. Narratives included.

Data Sources