Server has 12 tools with complete input schemas and descriptions. Naming follows verb_noun convention (lookup_, register_, list_, get_, update_, delete_, emit_, cancel_, query_, get_). Descriptions are present (avg ~80 chars) but lack actionable context for LLM selection. Parameters are typed but many lack format constraints or validation guidance. Output schemas are not documented in the visible code. Error handling is minimal, no recovery guidance or categorization. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels. Missing pagination for list_clients and query_issued_nfse. No confirmation pattern for destructive operations (delete_client, cancel_nfse).

• Document output schemas for all tools. Specify field names, types, and what data each field contains. E.g., emit_nfse should return {success: bool, numero_nfse: string, numero_rps: int, errors: []}.
• Add tool annotations: mark delete_client and cancel_nfse with destructiveHint=true; mark read-only tools (lookup_cnpj, list_clients, get_client, get_nfse_urls, get_nfse_log, get_provider_info) with readOnlyHint=true; mark idempotent tools with idempotentHint=true.
• Expand descriptions to 100-200 chars. Include WHEN to use the tool, WHAT it returns, and any prerequisites. E.g., 'Register a new client (tomador) for NFS-e emission. Auto-looks up CNPJ data if provided. Returns client_id for use in emit_nfse. Call this before emitting invoices for a new company.'
• Add pagination to list_clients, query_issued_nfse, and get_nfse_log. Include limit (default 20, max 100), offset/cursor, and total_count in responses.
• Implement confirmation pattern for delete_client and cancel_nfse. Add a dry_run parameter or require a confirmation_token from a separate confirm_* tool before executing destructive operations.

Inferred effective spec: <=2025-11-25.

• Relies on Dynamic Client Registration (deprecated) - use Client ID Metadata Documents (CIMD)