Use when the user asks how to build with OpenAI products or APIs and needs up-to-date official documentation with citations, help choosing the latest model for a use case, or model upgrade and prompt-upgrade guidance; prioritize OpenAI docs MCP tools, use bundled references only as helper context, and restrict any fallback browsing to official OpenAI domains.
The `openai-docs` skill provides users with authoritative guidance and documentation for working with OpenAI products and APIs. It is designed to solve the challenge of staying up-to-date with rapidly evolving OpenAI technologies by directly leveraging the official developer documentation through the MCP (Managed Content Platform) server. By prioritizing official OpenAI sources over generic web searches, the skill ensures that developers and users receive accurate, current, and reliable information for building applications, choosing appropriate models, and upgrading prompts or models effectively.
This skill offers several key features and capabilities. It enables precise searches within OpenAI's developer docs, allowing users to fetch specific sections, quote accurately, or explore doc pages without a clear query. For model selection, upgrade, and prompt-guidance tasks, it prioritizes the latest-model documentation and handles explicit target models while providing optional newer guidance when relevant. The skill also includes a structured workflow for handling cases when the MCP server is missing, including installation instructions, retry procedures with escalated permissions, and guidance for re-running searches after restarting the Codex environment. Additionally, it covers multiple OpenAI products, including the Apps SDK, Responses API, Chat Completions API, Codex, gpt-oss models, Realtime API, and Agents SDK, supporting a range of development and agentic workflows.
Target users include developers, engineers, and technical teams who are building with OpenAI technologies and need precise, up-to-date documentation for implementation, migration, and optimization tasks. Common use cases include selecting the latest model for a project, migrating existing code or prompts to newer models, exploring API capabilities, integrating multimodal interactions, and developing agentic applications that require real-time interactions or tool usage. By focusing on official guidance and structured workflows, the skill helps users reduce errors, maintain compatibility, and accelerate development with OpenAI products.
You can use the MCP tool `mcp__openaiDeveloperDocs__search_openai_docs` to find the most relevant pages. Once identified, use `mcp__openaiDeveloperDocs__fetch_openai_doc` to pull exact sections or quotes.
The skill provides documentation support for Apps SDK, Responses API, Chat Completions API, Codex, gpt-oss models, Realtime API, and Agents SDK.
Run `codex mcp add openaiDeveloperDocs --url https://developers.openai.com/mcp`. If permissions prevent installation, retry with escalated permissions and a brief justification. Restart Codex afterward and rerun searches.
Yes. It fetches guidance from the latest-model documentation and preserves explicit target models. Optional newer guidance is provided when the user does not specify a model.
Fallback to web search is only permitted if the MCP server is installed but returns no meaningful results. Any fallback must be restricted to official OpenAI domains.
Quick Setup:
.claude/skills/Repository
openai/skills