A funding rate of -0.191% per hour means nothing until you know what it pays on your size, and how long it takes to earn back the fees you paid to get in. This returns both, along with the fee-recovery time, so the scan output is already the comparison you were going to do by hand. It reads market data only — the position, and the risk that the rate flips while you hold it, are yours.
An MCP server over Hyperliquid's perpetual markets, aimed at one trade: collecting funding. Every hour one side of a perp pays the other, and when the rate goes far enough negative or positive there is a known edge in taking the paid side. This scans all 229 markets for those cases and projects what the position would earn.
- All 229 perps scanned for rates past a threshold you set, each result carrying the hourly and annualised rate, projected dollars per hour and per day for your equity and leverage, how long until fees are recovered, mark price and 24h volume — `scan_funding_rates`
- The current funding rate for one named asset — `get_rate`
- The markets currently sitting at extreme rates, listed on their own — `list_assets`
No account and no key — it reads Hyperliquid's public market data. The project calls itself `funding-rate-mcp` (0.1.0 in pyproject) and runs over stdio as a Python module; PyPI does not have that name. Acting on what it finds is a separate matter: this server reads rates, it does not place orders.