Loading ToolBench
A financial transaction analysis MCP server that provides tools for querying spending data by category, generating spending summaries, and analyzing monthly spending trends
Critical gaps across all dimensions. Tool descriptions are in Romanian and extremely brief (5-20 chars), violating the 10-1024 char baseline and providing no actionable context for LLM selection. Schemas are present but parameters lack descriptions entirely, 'category', 'start_date', 'end_date', 'months' have no guidance on format, constraints, or usage. No output schemas documented. No error handling guidance. Tool names are verb-noun compliant but descriptions fail to explain WHEN to use each tool or what distinguishes them. No security considerations, no pagination guidance for potentially large result sets, no idempotency markers.
Tendințe lunare
Totaluri pe categorii
Filtrează după categorie
Tool descriptions are 5-20 characters in Romanian ('Filtrează după categorie', 'Totaluri pe categorii', 'Tendințe lunare'). Baseline is 10-1024 chars with English context. LLMs cannot determine when to select each tool or how they differ.
All parameters lack descriptions. 'category' enum is clear, but 'start_date' and 'end_date' have no guidance on ISO format, timezone handling, or whether they are inclusive/exclusive. 'months' parameter has no range constraints (min/max).
No output schemas documented. LLMs cannot predict response structure (e.g., does get_spending_summary return nested objects or flat arrays?). Baseline: 100% of A+ tools document return types.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 29 | 2026-07-28+ | v2 |
No pagination support visible. get_transactions_by_category and get_spending_summary could return hundreds of records, blowing context window. No limit, offset, or cursor parameters.
No error handling guidance. If category is invalid or date parsing fails, LLM receives no recovery hint. Baseline: errors must tell LLM what to do next.