Verified against ChatGPT · 2026-07-10
Set up a ChatGPT Project so every new thread inherits the same context
Project-level instructions and file-handling rules that make every new chat inside a ChatGPT Project start with the right standing context, instead of you re-uploading files and re-explaining the brief in every thread.
The prompt
Ready to copy — highlighted parts are example details you can swap.
PROJECT INSTRUCTIONS — paste into this Project's instructions field so every thread inside it inherits this context automatically. PROJECT Q3 Pricing Launch WHAT THIS PROJECT IS FOR Rolling out usage-based pricing tiers for the SaaS product this quarter STANDING CONTEXT (true for every thread in this project, do not re-ask for this) Current plans are Starter, Growth, and Scale. Growth is the plan we are converting Starter users to. FILES IN THIS PROJECT Treat the uploaded files as the current source of truth. If a new thread's question conflicts with something in an uploaded file, flag the conflict instead of silently trusting whichever is more recent in the conversation. RULES FOR EVERY THREAD IN THIS PROJECT - Assume I am mid-work on Rolling out usage-based pricing tiers for the SaaS product this quarter, not asking a one-off question. Carry the standing context into your first response without me repeating it. - If a thread's request would be better served by something already established in another thread in this project, say so, since threads in this project do not automatically share each other's conversation history. - Always flag any pricing number you suggest as a draft, never state it as decided FIRST THREAD IN THIS PROJECT Draft the customer-facing FAQ explaining the new tiers and the migration timeline
Customize the highlighted detailsoptional — the prompt above already works
Why this works
A Project is a distinct container from both a single chat and from Custom Instructions: files and project-level instructions attached to it are available to every new thread opened inside that Project, which is what lets you run several separate standing contexts, a launch, a hiring process, a client, without them bleeding into each other or into your global Custom Instructions. Explicitly naming the uploaded files as the source of truth matters because a model will otherwise give recency more weight than provenance, treating a claim made three messages ago in the current thread as more authoritative than a fact sitting in a document, simply because it is closer in the context window. The line noting that sibling threads do not automatically share each others conversation history is a real, easily-missed mechanic: a Project gives every thread the same files and instructions, but it does not give thread B the transcript of thread A, so a rule reminding the model to say so when relevant prevents it from assuming continuity that is not actually there. Writing the standing context once, at the project level, is also what stops the earliest and most common Project failure: restating the same three paragraphs of background in every new thread you open.
Verified against
ChatGPT GPT-5.1 (Projects) · 2026-07-10
Changelog
- 2026-07-10 — Initial publish, verified against ChatGPT GPT-5.1 Projects.
Building this for real?
This is a free starting point. If you'd rather have what Scult builds built and running for your business, that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
