Single tool with clear naming (hyperliquid_get_vault_data) and good description (194 chars, explains WHAT, WHEN, and cross-references related tools). Input schema present with limit parameter typed as number and described. However, NO output schema documented, LLMs cannot plan downstream calls or extract fields. Parameter lacks min/max bounds (limit unbounded; default 20 but max 100 hardcoded in logic, not schema). Description is strong but schema completeness is weak.

• Add outputSchema to config.ts route definition. Document the response structure: { type: 'object', properties: { count: { type: 'number' }, totalVaults: { type: 'number' }, sortedBy: { type: 'string', enum: ['apr_descending'] }, vaults: { type: 'array', items: { type: 'object', properties: { vaultAddress, name, leader, apr, aprFormatted, totalPnl, followers, tvl, tvlFormatted } } }, cachedUntil: { type: 'string', format: 'date-time' } } }
• Add minimum and maximum to limit parameter schema: { type: 'number', minimum: 1, maximum: 100, description: 'Max vaults to return (default: 20, max: 100)' }
• Enhance error response with recovery guidance. Instead of generic 502, return: { error: 'Hyperliquid API unavailable', details: err.message, recovery: 'Retry in 30s or check https://status.hyperliquid.xyz' }
• Clarify tool description: remove references to non-existent tools or add a note like '(Note: whale_positions and funding_rates tools not yet available in this server)' to prevent LLM confusion.
• Document cache behavior in description: 'Results are cached for 60 seconds. Repeated calls within this window return identical data. Timestamp in cachedUntil field indicates when cache expires.'

Inferred effective spec: 2025-06-18+.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry