Server has 12 tools with complete input schemas and descriptions. Naming follows verb_noun convention (search_, list_, get_, send_, download_). Descriptions are present (avg ~100 chars) but lack actionable context for LLM selection. Parameters are typed but many lack format/constraint details. Output schemas are not documented. Error handling is minimal, no recovery guidance or categorization. Security concern: send_message and send_file are destructive operations with no confirmation pattern. Missing idempotency guidance for retry scenarios.

• Add output schema documentation for all tools. Example: 'Returns: {messages: [{id, sender, content, timestamp, media_type}], total_count, has_more}'. This enables LLM to chain tools correctly.
• Implement confirmation pattern for send_* tools. Add optional 'dry_run' parameter or require explicit 'confirm=true' to prevent accidental message sends.
• Document JID and phone number formats in parameter descriptions. E.g., 'recipient: Phone number (country code + digits, no +) or JID (user@s.whatsapp.net for DM, group@g.us for group).'
• Add error categorization and recovery guidance. E.g., 'If send_message fails with "Not connected", the agent is offline, retry after reconnection. If recipient is invalid, call search_contacts() first.'
• Specify bounds for pagination: 'limit: 1-100 (default 20)', 'page: 0-based integer'. Prevents LLM from passing absurd values.