Claude

Verified against Claude · 2026-07-29

Set up a Claude Project so you stop re-explaining context every chat

A structured project-knowledge document plus custom-instructions block for Claude Projects, so every new chat inside the project inherits your context automatically instead of you re-pasting it.

Claude (Projects)Claude Enterprise

The prompt

Ready to copy — highlighted parts are example details you can swap.

PROJECT KNOWLEDGE DOCUMENT — Q3 Product Launch — Northwind App

Paste this as a text file into this Project's knowledge, then paste the CUSTOM INSTRUCTIONS section below into the Project's custom instructions field. Everything here should be true for every future chat in this Project, not just the current one.

WHAT THIS PROJECT IS
Coordinating copy, positioning and internal FAQs for the Northwind app launch on October 14. Done means every asset is consistent with the same three positioning pillars.

AUDIENCE AND STAKEHOLDERS
Marketing team drafting external copy, and support leads writing internal FAQs for the same launch.

STANDING FACTS
Treat the following as ground truth for every chat in this Project. If a request in a chat conflicts with one of these, flag the conflict instead of silently picking a side.
- Launch date is October 14, not negotiable.
- Product is called "Northwind," never "Northwind App" in customer-facing copy.
- Pricing is not final and must never be stated as a number in any draft.

TERMINOLOGY
"NW" always means Northwind internally, never spelled out in Slack drafts. Customers are called "members," never "users."

---
CUSTOM INSTRUCTIONS (paste into the Project settings field)

When responding inside this Project:
- Assume the project knowledge above is already true; do not ask me to re-explain it.
- Default output register: Plain, confident, no exclamation points, short paragraphs over bullet walls
- Never invent a fact about this project that is not in the knowledge document or in the current chat — say what is missing instead of guessing.
- If a new fact emerges in a chat that should apply to every future chat, say so explicitly at the end of your reply so I know to add it to the knowledge document, rather than quietly assuming I will remember.
Customize the highlighted detailsoptional — the prompt above already works

Why this works

Claude Projects has two persistence mechanisms that this prompt deliberately targets separately: project knowledge (documents Claude retrieves from on every chat in that Project) and custom instructions (a standing system-level directive applied to every chat in that Project). Splitting standing facts from custom instructions matters because they serve different jobs — knowledge is retrieved content Claude reasons over, while custom instructions are behavioral rules about how to use that content, and conflating the two into one blob makes both weaker. The explicit instruction to flag a conflict rather than silently pick a side exists because project knowledge can go stale — a launch date changes, a decision gets reversed — and a model that quietly trusts old knowledge over a newer in-chat correction will confidently restate something that stopped being true. Asking Claude to name new facts worth promoting to the knowledge document at the end of a reply turns the Project into something that gets deliberately maintained over weeks rather than slowly drifting out of sync with reality, which is the actual failure mode of most team knowledge bases.

Verified against

Claude Sonnet 4.6 (Projects) · 2026-07-29

Changelog

  • 2026-07-29 Initial publish, verified against Claude Sonnet 4.6 in 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
All Claude prompts

Check your AI visibility

One URL in, a 0–100 score and the exact fixes out.

RUN THE CHECK

Browse all the tools

15 tools across six categories
13 of them never send your data anywhere

Free · No signup · No trial clock

SEE THE DIRECTORY