sand0vvv/hyperevm-mcp

**Ask your AI assistant about the Hyperliquid ecosystem in plain English.**
It reads. There is nowhere to put a private key, and that is the whole security model.
Four of those cells read n/a. That is the point: stHYPE publishes no yield pool, and kHYPE's weekly delta came back from the source implying the pool paid nothing seven days ago, which its own daily history contradicts. A number we cannot stand behind is not printed.
The vault row shows the same rule twice over. HLP has no pool on DefiLlama at all,
so a table built from that source alone silently omits the largest single place
to deposit on the chain — it is read from Hyperliquid instead. And Hyperliquid's
own APR field for it is *not* shown, because the same value means one thing as
an annual rate and something 365 times larger as a daily one, and the vault's
history does not settle which. What is shown is realised profit and loss per
window, divided here, where the arithmetic is ours to defend.
Then ask whether to believe one of those rates:
Zero earned, every day, for a month.
Install
**Claude Code**
**Cursor, Claude Desktop, or any MCP client** — add to the config file:
That is the whole setup. Nothing to fill in afterwards — there is no account to authenticate against.
Try asking
- *what's the best HYPE yield right now*
- *compare lending rates on HyperEVM*
- *which HYPE pools are just token emissions and which actually earn*
- *has that 40% APY held up, or is it decaying*
- *is BTC funding actually stable, or was that one hour*
- *who earns the most fees on this chain*
- *could I actually get out of $500k of HYPE*
- *which protocols grew this week*
- *tell me about HyperLend*
- *what's in this wallet: 0x…*
- *where should I stake HYPE and what does the commission cost me*
Tools
Nine tools. Five report the present, three report how it got there, one reads an address.
| Tool | What it answers |
| --- | --- |
| hyperevm_yields | Every yield on HYPE in one table: liquid staking, lending with supply/borrow/utilisation/max LTV, LP and looping, and the HLP vault — with earned yield separated from token emissions |
| hyperevm_protocols | What is deployed on HyperEVM: chain TVL, per-protocol TVL, 1d/7d change, sortable by size or weekly growth; a detail card for any one protocol |
| hyperevm_fees | Fees users actually paid, ranked over 24h/7d/30d. TVL says how much money sits somewhere; this says whether anyone is paying to use it |
| hl_market | Perp mark price, funding hourly and annualised, open interest, 24h volume; spot pairs; predicted funding across Hyperliquid, Binance and Bybit |
| hl_staking | HYPE validators ranked by predicted APR, with commission, uptime and stake share — what the network pays before any wrapper takes its cut |
| hyperevm_pool_history | One pool's APY and TVL over weeks, split into earned versus emitted — so a headline rate can be checked against its own history |
| hl_funding | How funding actually behaved over days: average, range, how much of the time it held its sign, and what the position would have paid |
| hl_orderbook | Spread and resting depth, and whether a given trade size fills inside the visible book — or a plain statement that it does not |
| hyperevm_wallet | Any public address: account value, open positions with entry, unrealised PnL and liquidation price, spot balances, HYPE staking |
The pairs are deliberate. hyperevm_yields and hl_market tell you what a number is now; hyperevm_pool_history and hl_funding tell you whether to believe it. A 47% pool that has been 100% emissions for a month and a 20% pool earning fees are not the same product, and only the second pair can tell them apart.
Slash commands