Why is input schema validation important for AI tools?

Strict JSON schemas prevent the model from sending malformed data and ensure that all required constraints—like string length or numeric ranges—are respected before execution.

Can MCP tools trigger specific UI components?

Yes, by using the 'openai/outputTemplate' meta extension, you can link tool outputs directly to specific HTML widget UIs for a more interactive user experience.

What are MCP tool annotations used for?

Annotations like readOnlyHint and destructiveHint inform the AI model about the consequences of a tool call, enabling smarter auto-approval for safe tasks and requiring confirmation for risky ones.

What is the recommended format for MCP tool names?

Tools should follow a lowercase 'verb_noun' format, such as 'get_user' or 'create_order', keeping the name under 64 characters and avoiding promotional language.

How do I accept file uploads in an MCP tool?

You can use the '_meta' extension field 'openai/fileParams' to specify which input schema properties should be treated as file objects by the OpenAI Apps SDK.

MCP Tool Design for OpenAI Apps SDK

Name: MCP Tool Design for OpenAI Apps SDK
Author: mberto10

bymberto10

API開発

Guides the creation of robust, secure, and well-documented MCP tools for the OpenAI Apps SDK using standardized schemas and behavioral hints.

概要

This skill provides a comprehensive framework for designing the interface between ChatGPT and Model Context Protocol (MCP) servers. It equips developers with the patterns needed to define precise tool names, complex JSON input schemas, and critical behavioral annotations such as read-only or destructive hints. By following these best practices, developers can ensure their tools exhibit natural model behavior, support advanced features like file uploads and UI widget integration, and maintain high standards for security and error handling within the OpenAI Apps ecosystem.

主な機能

Behavioral annotations for safety hints and auto-approval workflows
Structured response patterns for standard data and widget integration
OpenAI Apps SDK extensions for file parameters and UI visibility
Standardized tool naming conventions and descriptive patterns
0 GitHub stars
Advanced JSON Schema design for complex and nested input parameters

ユースケース

Designing secure interfaces for sensitive database or cloud infrastructure operations
Building custom MCP servers to extend ChatGPT with domain-specific capabilities
Developing interactive tools that utilize custom UI widgets via the Apps SDK

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add mberto10/mberto-compound mcp-tool-design

For use in Claude.ai and ChatGPT

Download Skill

GitHub

概要