Prerequisites
Cryptohopper MCP configured in an MCP client — see the setup overview.
An active Cryptohopper subscription. Orderbook data is available on all tiers, but multi-exchange comparisons need Explorer or higher for more than three venues — see subscription tiers.
Setup steps
Open your MCP client
Run the single-venue slippage check
Use the prompt below. A good response gives you a concrete average fill price and a basis-point slippage number.Using the Cryptohopper MCP, pull the current orderbook for BTC/USDT on Binance. Assume I want to buy 1.5 BTC at market.
Walk the ask side from the top:
- Report the average fill price for my full order.
- Compute slippage from midpoint in basis points.
- Report how deep into the book my order takes me (in price terms and level terms).
- Flag if the book depth is insufficient for my size.
Keep the answer numeric and concise.Extend to multi-venue comparison
Use the prompt below:
Repeat the same analysis for BTC/USDT on Coinbase, Kraken, OKX, and Bybit. Present the results in a table with columns: exchange, average fill price, slippage (bp), depth consumed. Sort by best fill price first.
Save both prompts as reusable templates
Parameterize pair, size, and side (buy/sell) so you can reuse them for any order.
Wire into your execution workflow
For manual orders: run the multi-venue check, pick the best venue, then place manually. For end-to-end agents: use the venue ranking to drive the execution step — see research via MCP, execute via REST API .
Sample output
What the multi-venue prompt returns for a 1.5 BTC market buy:
Exchange | Avg fill | Slippage (bp) | Depth consumed |
Binance | 66,182.40 | 2.1 | 3 levels |
OKX | 66,186.15 | 2.7 | 5 levels |
Bybit | 66,190.80 | 3.4 | 4 levels |
Kraken | 66,201.50 | 4.9 | 7 levels |
Coinbase | 66,245.30 | 11.5 | 14 levels |
Actual numbers vary with live conditions. Coinbase slippage is notably elevated here — the ask side appears thin within 50 bp of mid.
Cost profile
Orderbook calls cost 1 unit each on all tiers.
Check | Call units |
Single-venue check | 1 |
Five-venue comparison | 5 |
10 checks/day × 5 weekdays, 5 venues | 250/week |
Comfortably within any tier. See rate limits explained.
Troubleshooting
The agent returns the book but doesn't walk it
Be explicit in the prompt — demand the computation step by step: "Starting from the best ask, sum the sizes at each level until you have at least 1.5 BTC. Weight each level's price by the size consumed at it. Report the weighted average." Without explicit instructions, the model may summarize instead of compute.
The slippage number looks too low or too high
Orderbooks go stale within seconds on active markets. Re-run the check immediately before acting — do not rely on a result from 30 seconds ago. If the number remains implausible, check the reported depth consumed. A result based on too few levels usually means the book is thin and the estimate is optimistic.
EXCHANGE_NOT_SUPPORTED when you add exchanges
Your tier does not include all requested venues. Pioneer covers three exchanges (Binance, Coinbase, Kraken); Explorer covers eight. See supported exchanges and subscription tiers.
Depth is insufficient on every exchange
You're trying to move more size than any single book can absorb without meaningful impact. Options: split the order across exchanges, reduce the size, use a limit order instead of a market order, or trade in tranches over time. If the output doesn't flag this automatically, add to your prompt: "if any exchange has insufficient depth for my full size, say so explicitly and suggest split execution."
The comparison picks a venue that seems wrong
Thin-book venues can show a low slippage number on a small calculation but be a bad choice in practice. Add a sanity check to your prompt: "also report 24h volume and spread for each exchange, so I can spot suspiciously thin venues."