GPT WORLD / INDEPENDENT CODEX FIELD MANUAL Source review · 2026-07-31

Retrospective edition / PROMPTS

Version prompts like application code

Separate durable developer instructions from user input, give the task an explicit output contract, and keep prompt changes reviewable.

01 / DECIDE

Architecture decision

Use it when behavior must remain stable across releases, models, languages, or multiple product surfaces.

Separate durable developer instructions from user input, give the task an explicit output contract, and keep prompt changes reviewable. Map input, output, state, and side effects as one observable path before deciding which layer owns version prompts like application code.

02 / BUILD

Three-step implementation

  1. 01

    Frame the contract

    Write down the caller, data classification, success condition, timeout, cancellation, and ownership. Use it when behavior must remain stable across releases, models, languages, or multiple product surfaces.

  2. 02

    Build one narrow path

    Implement one end-to-end path with request correlation, typed state, and reversible failure handling. Separate durable developer instructions from user input, give the task an explicit output contract, and keep prompt changes reviewable.

  3. 03

    Prove the outcome

    Turn acceptance into a repeatable fixture, contract test, or browser test. A fixture suite compares expected task outcomes and refusal handling before each prompt revision ships.

03 / BOUND

Production boundary

Prompts are not authorization. Server policy must still validate tools, data access, and consequential actions.

04 / PROVE

Acceptance evidence

A fixture suite compares expected task outcomes and refusal handling before each prompt revision ships.

SOURCE / HTTP

Reproducible source probe

curl -fsSI 'https://developers.openai.com/api/docs/guides/prompt-engineering' | sed -n '1,5p'