What is the benefit of using the Responses API over Chat Completions?

The Responses API is recommended for new projects as it offers built-in tools like web search and file search, handles server-side conversation state, and provides better performance for agentic workflows.

Does this skill support streaming for better user experience?

Yes, it provides patterns for both the .stream() helper and standard for-await-of iteration to handle real-time token delivery in chat interfaces.

Can I use this skill with Azure OpenAI?

Yes, the patterns include configuration guides for both direct OpenAI API access and Azure OpenAI service instances.

How does this skill ensure type safety for AI responses?

It utilizes the zodResponseFormat helper from the OpenAI SDK to map Zod schemas to JSON schemas, ensuring the model output is automatically validated and typed.

How are API keys managed in these patterns?

The skill enforces security best practices by requiring the use of environment variables like OPENAI_API_KEY instead of hardcoding sensitive credentials.

OpenAI SDK Integration

Name: OpenAI SDK Integration
Author: agents-inc

byagents-inc

•

数据科学与机器学习

Implements official OpenAI SDK patterns for Chat Completions, Responses API, and structured outputs in TypeScript.

This skill provides standardized patterns for integrating the official OpenAI SDK into TypeScript and Node.js projects within Claude Code. It enables developers to implement complex AI workflows, including the newer Responses API for agentic behaviors with built-in tools, type-safe structured data extraction using Zod, and low-latency streaming. By following production-grade best practices for error handling, retries, and client configuration, it ensures robust and scalable integration with the GPT-4o and o1 model families, as well as vision, audio, and embedding APIs.

主要功能

01Type-safe structured outputs using Zod schema validation

02Comprehensive multimodal support for vision, audio, and embeddings

03Production-ready error handling and client retry configuration

04Support for Responses API with server-side state and built-in tools

05Streamlined event-based and for-await streaming patterns

065 GitHub stars

使用场景

01Developing high-performance RAG pipelines using embeddings and batch processing

02Building agentic workflows with built-in web search and code execution

03Extracting validated JSON data from LLM responses into TypeScript types

主要功能

01Type-safe structured outputs using Zod schema validation

02Comprehensive multimodal support for vision, audio, and embeddings

03Production-ready error handling and client retry configuration

04Support for Responses API with server-side state and built-in tools

05Streamlined event-based and for-await streaming patterns

065 GitHub stars

使用场景

01Developing high-performance RAG pipelines using embeddings and batch processing

02Building agentic workflows with built-in web search and code execution

03Extracting validated JSON data from LLM responses into TypeScript types