Two tools with clear, domain-specific names (assert_read_only, is_read_only) and good descriptions (60-80 chars each). Input schemas present with typed parameters and descriptions. However, output schemas are not documented in the visible code, only exception types are mentioned. Error handling via ReadOnlyViolation exception is present but lacks recovery guidance. No tool annotations (readOnlyHint, destructiveHint). The tools are narrow and focused (single responsibility), but descriptions could be more explicit about when to use each variant and what the return value structure is.

• Document the return value for both tools. assert_read_only returns None on success; is_read_only returns a boolean. Make this explicit in tool descriptions and output schema.
• Add tool annotations: mark both tools with readOnlyHint=true and idempotentHint=true so agents know they are safe to call repeatedly without side effects.
• Enhance error messages with recovery guidance. Instead of raising a bare ReadOnlyViolation, include the reason (e.g., 'contains pg_read_file() call') and suggest 'Try removing file I/O functions.'
• Clarify the distinction between assert_read_only and is_read_only in their descriptions. E.g., 'Use assert_read_only to enforce read-only checks and raise an exception on violation; use is_read_only for a boolean result without exceptions.'
• Add an enum constraint for the 'dialect' parameter: ['postgres', 'mysql', 'tsql', 'mssql', 'sqlserver', 'sqlite', None]. Document that None enables dialect-agnostic parsing.