Loading ToolBench
MCP server for Solana blockchain interactions via QuickNode RPC endpoint
Single tool 'get_slot' has critical definition gaps. Tool naming follows verb_noun pattern (get_slot) but lacks context about what 'slot' means in Solana. No output schema documented. No error handling guidance. This server exposes only one read-only operation with minimal metadata, insufficient for production use.
Fetch the current slot number from the Solana blockchain
Even read-only tools should document expected inputs (e.g., commitment level, min_context_slot).
No output schema documented. LLMs cannot plan downstream calls or extract slot number without knowing response structure (e.g., is it {slot: number} or {result: {slot: number}}?).
Tool description 'Fetch the current slot number from the Solana blockchain' (48 chars) lacks context: What is a slot? When should this be called vs other Solana queries? What does the response contain? Baseline descriptions average 194 chars.
No error handling guidance. What happens if the RPC endpoint is down? If the network is unreachable? LLM receives no recovery path.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 39 | <=2025-11-25 | v2 |
Single tool severely limits domain coverage. Solana agents need: get_balance, get_account_info, send_transaction, get_transaction_status, list_tokens, etc. Current server is a stub.