Loading ToolBench
MCP server for querying Solana blockchain data including transaction history and account balances
Two tools with minimal documentation. Both have basic input schemas with type definitions and descriptions, but tool descriptions are extremely brief (under 100 chars). Parameter descriptions are present but generic. No output schemas documented. Error handling returns None with print statements instead of structured error guidance. Tool names follow verb_noun pattern but lack clarity on what data is returned. No pagination support despite tools returning lists. Missing idempotency hints, permission declarations, and recovery guidance.
Fetches the current amount balance for a given Solana address
Fetches the latest transactions for a given Solana address
Tool descriptions are too brief (under 100 chars). LLMs cannot determine when to select these tools vs alternatives or what data structure is returned. Baseline for A+ tools is 194 chars average.
No output schemas documented. LLMs cannot plan downstream operations or extract required fields. get_solana_latest_transaction returns a list but pagination, total count, and field structure are undocumented.
Error handling returns None with print() statements instead of structured error responses. LLMs receive no guidance on what failed, why, or what to do next. Should return actionable error objects with recovery hints.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | 2026-07-28+ | v2 |
get_solana_latest_transaction accepts a limit parameter but no pagination support (no offset, cursor, or next_token). Returning raw API response without capping results risks context window exhaustion.
Parameter descriptions are generic ('The Solana address to query'). Missing format constraints, validation rules, or examples of valid addresses. LLMs cannot validate input before calling.