Loading ToolBench
Hyperliquid perp market data -- order books, prices, funding rates, OI, volume, OHLCV candles in one API.
Three tools with complete input schemas and descriptions, but descriptions are verbose (200+ chars) and lack LLM-optimized structure. Tool names follow verb_noun pattern correctly. Parameters have types but lack granular descriptions. No output schemas documented. Error handling and recovery guidance absent. Naming is clear but descriptions bury actionable intent under examples and cross-references.
Use this when you need a snapshot of all Hyperliquid perpetual markets at once. Returns pricing, funding, and volume data for every listed asset in a single call. 1. markets: array of all listed perp assets 2. Each market contains: coin, midPrice, markPrice, fundingRate, openInterest, volume24h, dayChange 3. totalMarkets: number of listed perp markets 4. timestamp: data timestamp in ISO format Example output: {"markets":[{"coin":"BTC","midPrice":67450,"fundingRate":0.00008,"openInterest":1250000000,"volume24h":3400000000,"dayChange":2.3},{"coin":"ETH","midPrice":3520,"fundingRate":0.00012,"openInterest":890000000,"volume24h":1200000000,"dayChange":1.1}],"totalMarkets":148} Use this FOR screening all Hyperliquid markets, finding highest funding rates, or building a market overview dashboard. Do NOT use for a single coin order book -- use hyperliquid_get_market_data. Do NOT use for spot DEX quotes -- use dex_get_swap_quote. Do NOT use for vault performance -- use hyperliquid_get_vault_data.
Fetch OHLCV candle data for a Hyperliquid perpetual asset at specified interval and lookback period.
Use this when you need Hyperliquid perpetual futures data for a specific coin. Returns full L2 order book depth with real-time pricing and funding metrics in one call. 1. bids: array of bid levels with price and size 2. asks: array of ask levels with price and size 3. midPrice: current mid price between best bid/ask 4. markPrice: oracle mark price used for liquidations 5. fundingRate: current hourly funding rate (e.g. 0.0001 = 0.01%) 6. openInterest: total open interest in USD 7. volume24h: 24-hour trading volume in USD Example output: {"midPrice":67450.5,"markPrice":67448.2,"fundingRate":0.00008,"openInterest":1250000000,"volume24h":3400000000,"bids":[{"price":67450,"size":12.5}],"asks":[{"price":67451,"size":8.3}]} Use this BEFORE placing trades on Hyperliquid. Essential for checking spread, depth, and funding before entering a position. Do NOT use for EVM gas -- use gas_get_current_price. Do NOT use for spot DEX quotes -- use dex_get_swap_quote. Do NOT use for whale positions -- use hyperliquid_track_whale_positions.
Descriptions exceed 200 chars and lack LLM-optimized structure. First sentence should state WHAT+WHEN, then details. Current descriptions bury key intent under examples and cross-tool references.
Output schemas not documented. LLMs cannot plan downstream calls or extract fields without knowing response structure. Baseline: 100% of A+ tools document return types.
Parameter descriptions missing or minimal. 'coin' param lacks format guidance (e.g., 'uppercase symbol, 3-5 chars'). 'interval' and 'limit' in get_candles lack constraints. Baseline: 100% of A+ tools have param descriptions.
Inferred effective spec: 2025-06-18+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 61 | 2025-06-18+ | v2 |
No error handling or recovery guidance. If coin symbol is invalid or API fails, LLM has no actionable next step. Baseline: A+ tools include 'Try search_*() first' or 'Retryable: yes/no'.
Cross-tool references in descriptions ('Do NOT use for...') are anti-patterns. LLMs do not reliably parse negations. Use tool selection logic server-side or rely on LLM's own reasoning. Descriptions should focus on WHEN to use THIS tool, not when NOT to.