Loading ToolBench
Ask any blockchain data question in plain English and get live data back — 15,500+ subgraphs, Token API, Polymarket and Hyperliquid trader intelligence. Paid calls settle in USDC on Base over x402.
Graph Advocate exposes 5 tools with schemas and descriptions, but critical gaps undermine production readiness. Tool names lack action verbs (route_data_request, check_quota, polymarket_trader_score, hyperliquid_trader_score, preflight_price). Descriptions are present but inconsistent in quality, some are verbose and include implementation details (payment mechanics, pricing) rather than user-facing intent. Parameter descriptions are minimal or absent: 'sender' in check_quota has no description; 'wallet' in trader_score tools lacks format guidance (EVM address validation). No output schemas are documented, LLMs cannot predict response structure or chain tools. Error handling is absent: no guidance on payment failures, invalid addresses, or rate limits. The server mixes concerns (pricing, payment settlement, data routing) without clear separation. STDIO transport caps protocol readiness at 50.
Free. How much free daily quota a wallet has left on the A2A endpoint. Note the quota does NOT apply to route_data_request, which is charged every call — so a non-zero remaining_today here does not make the next routing call free.
Skill score for a Hyperliquid perps trader — win rate, sizing discipline, drawdown. $0.02 USDC on Base.
Skill score and classification (sharp / retail / neutral) for a Polymarket wallet, derived from its trading history. $0.01 USDC on Base.
Free. Read any Graph Advocate endpoint's 402 challenge WITHOUT paying: exact price, asset, network and payTo. Use to budget a run, or to drive a payment from your own wallet. Never settles anything.
Ask any blockchain data question in plain English and get back the right service, a ready-to-run query, and — when the answer carries a runnable subgraph query — the live rows themselves. Covers 15,500+ subgraphs across 20+ chains, Token API (EVM/Solana/TON), Aave, Uniswap, ENS, Polymarket, Hyperliquid and ERC-8004 agent discovery. Always costs $0.01 USDC on Base over HTTP — the free daily tier applies to the A2A endpoint, not to this one — and returns a payment_required object when unpaid.
Tool names lack action verbs. 'route_data_request' is vague (route to where?); 'check_quota' is generic. LLMs cannot infer intent from names alone. Rename to 'ask_graph_advocate', 'get_quota_remaining', 'score_polymarket_trader', 'score_hyperliquid_trader', 'get_endpoint_price'.
No output schemas documented. LLMs cannot predict response structure (e.g., does route_data_request return query, rows, cost, payment_required as separate fields or nested?). Document return types for all tools.
Parameter descriptions are missing or minimal. 'sender' in check_quota has no description; 'wallet' in trader_score tools lacks format guidance (EVM address, 0x prefix required?). Add descriptions explaining format, constraints, and examples.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
Descriptions conflate implementation details with user intent. 'Always costs $0.01 USDC on Base over HTTP' and payment mechanics belong in docs, not tool descriptions. Rewrite to focus on what the tool does for the user: 'Ask any blockchain data question in plain English and get back the right service and a ready-to-run query.'
No error handling guidance. What happens if wallet address is invalid? If quota is exceeded? If payment fails? LLMs need actionable recovery steps, not silent failures.