Loading ToolBench
CryptoCompare prices, OHLC history, social stats, news, exchanges.
This server has 15 tools with basic schemas and minimal descriptions. All tools are READ_ONLY and properly typed (string, number, boolean). However, descriptions are extremely terse (5-20 chars), parameter descriptions are completely absent, and output schemas are undocumented. No error handling guidance, no enum constraints for categorical parameters (e.g., 'lang', 'sortOrder'), and no pagination documentation despite tools returning lists. Tool names lack action verbs (e.g., 'price' instead of 'get_price'). The codebase shows explicit tool registration with inputSchema, but parameter-level descriptions are missing entirely, violating the baseline that 100% of A+ tools have parameter descriptions.
Full coin list.
Exchange list.
Daily OHLC.
Hourly OHLC.
Minute OHLC.
News feed.
News categories.
Tool names lack action verbs. 'price', 'news', 'social_stats' do not convey intent. Should be 'get_price', 'list_news', 'get_social_stats' to help LLMs infer action before reading description.
Descriptions are critically short (5-20 chars). Baseline is 194 chars. Examples: 'Current price (one coin → many fiat).' (38 chars), 'Minute OHLC.' (12 chars). Too brief for LLM selection and context. Must explain WHAT, WHEN, and WHAT IT RETURNS.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 33 | 2026-07-28+ | v2 |
News feeds.
Current price (one coin → many fiat).
Full snapshot.
Current price (many → many).
Social stats.
Top by market cap.
Top trading pairs.
Top by volume.
Parameters lack descriptions entirely. Baseline: 100% of A+ tools have parameter descriptions. Example: 'fsym' parameter has no description explaining it is a cryptocurrency symbol. LLMs cannot infer meaning from names alone.
No enum constraints for categorical parameters. 'lang' (language code), 'sortOrder', 'feeds', 'categories' accept free-form strings, inviting hallucinated values. Should declare valid options as enums.
Output schemas are undocumented. No indication of what fields are returned, their types, or structure. LLMs cannot plan downstream calls or extract data without knowing response shape.