exchange-data-mcp
Provides read-only market data from Binance, including tickers, order books, recent trades, and candlesticks, normalized into a common format.
Provides read-only market data from Coinbase, including tickers, order books, recent trades, and candlesticks, normalized into a common format.
Provides read-only market data from KuCoin, including tickers, order books, recent trades, and candlesticks, normalized into a common format.
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
exchange-data-mcp
Read-only market data MCP server for crypto exchanges with public endpoints only and no API keys and no accounts. Binance, KuCoin and Coinbase normalized into one shape and serialized for LLM context.
Works as an MCP server for agent clients or as a plain Python library with zero dependencies.
Why this exists
Agents that work with markets need order books, trades and candles, and the existing options hand the model raw exchange JSON. Raw payloads often carry additional useless stuff the model doesn't read, while this option doesn't. and on top of that it unifies info across exchanges. This package normalizes the dialects once then serializes them more easily.
Related MCP server: PredMCP
Notable stuff about it:
Symbol dialects. BTC-USDT on KuCoin and Coinbase, BTCUSDT on Binance, are aggregated in one canonical form (BASE-QUOTE).
Timestamp units change per exchange and per endpoint. Binance is milliseconds, KuCoin stats and books are milliseconds, KuCoin trade history is nanoseconds, candles on KuCoin and Coinbase are seconds, Coinbase tickers are ISO strings. Every model comes out in ms since epoch.
Coinbase candle rows are [time, low, high, open, close, volume], newest first. That is the documented column order. KuCoin candles use [time, open, close, high, low, ...] instead. Positional mapping with tests, or the parser inverts candles.
Rate limits are per IP and agent tools wake up in bursts. The cache holds per-endpoint TTLs and collapses concurrent misses into one request (single-flight) so that five tools reading the same book in the same second cost one HTTP call. Failures aren't cached and a caller whose leader raised retries immediately after.
Spread math runs on Decimal end to end.
Slim serializers keep exact decimal strings under short keys. Measured on the recorded fixtures: the 24h ticker drops from 605 bytes of raw envelope to 207 bytes (34%). On the already-compact book snapshot the size is roughly neutral; the win is one schema in three exchanges.
Usage
As a library (zero dependencies):
from exchange_data_mcp import MarketDataService
from exchange_data_mcp.transport import urllib_fetcher
svc = MarketDataService(urllib_fetcher())
svc.ticker("kucoin", "BTC-USDT")
svc.compare("BTC-USDT") # same pair on every venue, executable spread from top of bookAs an MCP server:
pip install -e ".[mcp]"
python -m exchange_data_mcp.serverTools exposed: get_ticker, get_orderbook, get_recent_trades, get_candles, compare_price.
Errors are profiled: retryable (RateLimited carrying retry_after, ExchangeUnavailable), fix-the-input (SymbolNotFound, UnsupportedRequest), or the-exchange-changed-something (BadResponse).
Tests
37 tests, fully offline against recorded fixtures:
python3 -m unittest discover -s tests -t .Status
For portfolio purposes. REST with a TTL cache. For the live 10 ms streaming side of the same engineering, see kucoin-fast-stream.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Live crypto + IDX stock market intelligence for AI agents — 19 tools, no API key.
Crypto market data for AI agents: live prices, OHLCV, sentiment, indicators and exchanges.
Market and on-chain crypto data for AI agents. Pay per call in USDC on Base (x402).
Aligned market data for AI agents: crypto, stocks, FX and macro on one clock. Local or hosted.
Related MCP Servers
- AlicenseAqualityCmaintenanceProvides live cryptocurrency market data from over 100 exchanges, enabling AI agents to fetch prices, order books, funding rates, and more for trading analysis and arbitrage opportunities.132MIT

PredMCPofficial
AlicenseNot gradedqualityDmaintenanceSafe, read-only market data for AI trading agents, offering 44 tools to query prediction markets, perpetuals, and cross-venue signals without the ability to execute trades.1MIT- AlicenseNot gradedqualityCmaintenanceEnables AI agents to query cryptocurrency network and market data through tools for time-series metrics, OHLCV candles, bid/ask quotes, index levels, and asset or market metadata.MIT
- AlicenseAqualityBmaintenanceEnables AI agents to fetch live crypto prices and market data, convert between crypto and fiat, track EVM gas fees, value portfolios, calculate profit/loss, and view historical charts and trending metrics with zero configuration required.1075 npmMIT