AWS MCP server has complete input schemas for all 9 tools with proper JSON Schema structure (types, descriptions, required fields). Tool names follow verb_noun convention (create_*, list_*, describe_*, delete_*). However, descriptions are minimal (10-40 chars) and lack context about when to use each tool, prerequisites, or what the tool returns. Parameter descriptions are present but generic. No output schemas documented. Error handling is absent from visible code. Security concerns: master_password exposed as parameter in create_rds_instance. No pagination support for list_resources despite potential for large result sets.

• Expand tool descriptions to 50-200 characters. Example: 'Create an EC2 instance in a specified region. Requires an AMI ID and region. Returns instance_id, public_ip, and state. Use after creating a VPC and security group if deploying into a custom network.'
• Move master_password from create_rds_instance parameters to server-side secret injection. Accept only a secret_name parameter and resolve it to the actual password inside the tool handler using environment variables or a vault.
• Add output schema documentation for each tool. Document return types as JSON Schema (e.g., create_ec2_instance returns {instance_id: string, public_ip: string, state: string, launch_time: string}).
• Implement pagination for list_resources: add limit (1-100, default 20) and next_cursor parameters. Return {resources: [...], total_count: int, next_cursor: string|null}.
• Add error handling with recovery guidance. For delete_resource, return 'Resource not found. Available resources: [...]' on 404. For permission errors, return 'Insufficient permissions. Required: ec2:TerminateInstances'. For destructive ops, support a dry_run parameter.