OpenAI's Official Prompting and Codex Guide
赛博禅心 · wechat · 2026-07-11
This article translates and compiles OpenAI's official prompting guide, authored by Eric Provencher, who oversees Codex developer experience. It focuses on how to effectively instruct ChatGPT/Codex, emphasizing the need to state the goal first before adding context, output formats, and constraints, rather than starting with lengthy steps.
The guide also summarizes common scenarios:
- For large tasks: Provide goals, context, output expectations, and constraints
- For summarizing/comparing/converting: Attach documents, tables, slides, or PDFs
- For latest information: Enable web search and request sources
- For shared files or local assets: Organize context using projects and connected sources
The latter half covers the division of labor between plugins and personalization: cross-task preferences go into settings, while task-specific requirements belong in the prompt. Finally, it explains Codex's Steer and Queue follow-up behaviors. Steer injects a new message into the current run to change direction immediately, whereas Queue holds it for the next turn; both have corresponding controls in the desktop app and CLI.
Related event: OpenAI Releases Official Prompting and Workflow Guide for Codex(3 posts)→
More from coding & agent
- HeyGen adds a media-sourcing skill for coding agents with 75k images and 10k tracks — HeyGen · 2026-07-22
- Agent search bottlenecks are now about variance, not raw latency — rohanpaul_ai · 2026-07-22
- LangSmith adds tracing for Pipecat, LiveKit, OpenAI Realtime, and Gemini Live — LangChain · 2026-07-22
- An MCP server signs every AI agent tool call into a verifiable Merkle chain — Funky_Chicken_22 · 2026-07-22
- Annotated transcript of a Claude Code team interview is now available — trq212 · 2026-07-22
- Claude Code skill uses 10 Markdown rules to make outputs ADHD-friendly — alex_verem · 2026-07-22