Loading ToolBench
MCP server starter template for Eterna Hybrid Exchange integration, providing tools for account snapshots and market data queries
Three tools with minimal definitions. Tool descriptions are present but generic (10-50 chars), below the 194-char baseline. No error handling guidance, no output schemas, no composition hints. This server reads like a thin wrapper around an SDK with no LLM-optimization effort.
Fetch account balance information
Fetch open positions for the account
Fetch ticker/market data for a given symbol
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract fields without knowing what getBalance, getPositions, and getTickers return.
Tool descriptions are generic and under 50 chars. 'Fetch account balance information' does not explain when to use this vs getPositions, what fields are returned, or whether the call is safe to retry.
getTickers parameter 'symbol' has a description, but no enum constraint or format validation. LLMs may hallucinate invalid symbols like 'BTCUSD' or 'BTC' instead of the required 'BTCUSDT' format.
No error handling guidance. If getTickers is called with an invalid symbol or network fails, the LLM has no recovery path. Errors should guide the agent to retry, search for valid symbols, or escalate.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 39 | 2026-07-28+ | v2 |
No tool composition hints. If an agent needs to monitor positions and fetch ticker data for each, it must infer the workflow. Descriptions should hint: 'Call getPositions first, then getTickers for each symbol in the positions.'