Blockstream Esplora Bitcoin block explorer (mainnet + testnet)
Server defines 8 tools with basic schemas and descriptions, but falls short of production quality. All tools have descriptions (10-70 chars, mostly terse), and input schemas are present with type declarations. However, parameter descriptions are sparse or missing entirely. Most parameters lack guidance on format, constraints, or valid values. No output schemas documented. Error handling is minimal, generic HTTP status codes without recovery guidance. Tool names lack action verbs (e.g., 'tip_height' instead of 'get_tip_height'). No tool annotations (readOnlyHint, etc.). The server is read-only and stateless, which is good, but the interface is underdeveloped for LLM reasoning.
Address summary (chain + mempool stats).
Recent transactions for an address (mempool + confirmed).
Block by 64-char hex hash or by height.
Recommended sat/vB by confirmation target.
Mempool size + fee histogram.
Current block height of the chain tip.
Transaction details by txid.
Tool names lack action verbs. Names like 'tip_height', 'block', 'address' do not start with verbs (get_, list_, search_). LLMs infer intent from the verb, these names are ambiguous and force the LLM to read descriptions to understand what action occurs.
Parameter descriptions are missing or minimal. 'network' parameter appears in 7 tools with description 'mainnet (default) | testnet', no guidance on what happens if invalid value is passed, or whether it is truly optional. 'hash_or_height' in 'block' tool lacks format guidance (64-char hex vs numeric string). LLMs cannot infer constraints from names alone.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 45 | 2026-07-28+ | v2 |
Confirmation status for a txid (block_height, block_hash, confirmed).
No output schemas documented. Tools return JSON from Blockstream API, but the server does not declare what fields to expect (e.g., 'block' returns {hash, height, timestamp, ...}). LLMs cannot plan downstream calls or extract data without knowing the response structure.
Error handling is generic and non-actionable. Errors like 'Blockstream: 404' or 'Blockstream: 500' do not guide the LLM on recovery. No distinction between retryable (network timeout) vs user-fixable (invalid address) vs fatal errors. No suggestions for alternative actions.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All tools are read-only and idempotent, but this is not declared in the schema. LLMs cannot infer safety properties and may treat these as stateful operations.