Server exposes 9 read-only crypto/web/company data tools via STDIO. Tool names lack action verbs (honeypot_check vs verify_token_security; token_market vs get_token_price). Descriptions are present and moderately detailed (avg ~120 chars), but lack LLM-optimized guidance on WHEN to use each tool vs alternatives. Input schemas use Zod with type declarations and descriptions, but lack enums for chain/token parameters (free-form strings invite hallucination). Output schemas are undocumented, LLMs cannot predict response structure. Error handling is minimal: 402 payment errors return structured guidance, but API failures return raw JSON with no recovery hints. No tool annotations (readOnlyHint, idempotentHint) despite all being read-only. Parameter descriptions are adequate but could specify constraints (e.g., 'chain must be one of: ethereum, base, arbitrum…' as enum, not prose).

• Rename tools to start with action verbs: 'honeypot_check' → 'verify_token_honeypot', 'token_market' → 'get_token_market_data', 'company_kyb_fr' → 'check_french_company_kyb', 'us_company' → 'get_us_company_profile'.
• Convert chain and token parameters to enums in Zod schema: z.enum(['ethereum', 'base', 'arbitrum', 'optimism', 'polygon', 'bsc', 'avalanche', 'solana']) instead of z.string().default(). This prevents hallucinated chain names.
• Document output schemas in tool descriptions. E.g., honeypot_check returns {verdict: 'OK'|'CAUTION'|'HIGH_RISK'|'AVOID', isHoneypot: boolean, buyTax: number, sellTax: number, holderCount: number, flags: string[]}.
• Add tool annotations to all tools: inputSchema.readOnlyHint = true and inputSchema.idempotentHint = true. This signals to agents that calls are safe to retry and cache.
• Enhance error handling: wrap API responses in a consistent error envelope. E.g., if res.status >= 400 and res.status !== 402, return {error: 'API_ERROR', message: '<parsed error from API>', retryable: true/false, suggestion: '<actionable next step>'}.