Gmail MCP has clear, well-structured tool definitions with complete schemas and descriptions. All 7 tools are explicitly registered with proper JSON Schema. Naming follows verb_noun convention (list_, search_, read_, create_, modify_, get_). Descriptions are concise (50 - 150 chars) and explain WHAT each tool does. However, several parameter descriptions lack actionable constraints (e.g., max_results has no min/max bounds, query lacks format guidance). Output schemas are not documented, LLMs cannot predict response structure. Error handling is generic (catch-all exception → JSON error). No tool annotations (readOnlyHint, destructiveHint) despite clear read/write distinction. Missing idempotency guidance for write operations.

• Document output schemas for all tools. Example for list_emails: 'Returns {messages: [{id, subject, from, date, snippet}], total_count}'. LLMs need to know what fields to extract.
• Add min/max constraints to numeric parameters in schema. Change max_results to {type: integer, minimum: 1, maximum: 100} instead of relying on description alone.
• Add tool annotations: mark read_email, list_emails, search_emails, list_labels, get_attachment with readOnlyHint=true; mark create_label, modify_email with destructiveHint=true.
• Enhance error messages with recovery guidance. E.g., catch 'message not found' and return 'Message ID not found. Try search_emails(query="...") to find the message first.'
• Clarify modify_email behavior or split into separate tools. If it only adds/removes labels, rename to add_remove_labels. If it also marks read, document that explicitly or create mark_as_read as a separate tool.