Pack №CLOUDCode word: CLOUD

5 Claude Code settings I never skip «5 Claude Code settings» — the full setup guide

The exact setup that stops Claude Code from improvising, forgetting context, and burning your token limit — with direct links to every doc. Free.

CLOUD

Comment CLOUD under the video and the guide that lands in your DMs is this page — the 5-settings setup guide.

5 settingsFree guideDirect doc links

The 5 settings that stop Claude Code from messing up.

The result

The 5 settings

If you just installed Claude Code and threw a task at it — don't be surprised by the mistakes. These five settings close most of them. Do them now, not after the third screw-up.

1. CLAUDE.MDGeneral rules Claude reads in every chat — no need to explain the same thing twice. Write how you like to work, e.g. “Ask questions, don't invent what you don't know.” Docs: https://docs.anthropic.com/en/docs/claude-code/memory

2. PLAN FIRST, THEN WORKAdd it as a rule in CLAUDE.md: Claude drafts a plan first and starts only after your “ok”. You see what it's about to do before it touches code — and can correct it early. Anthropic themselves recommend this.

3. DON'T SIT IN ONE CHAT FOREVERThe longer the chat, the more tokens every message eats. Chat getting heavy? Ask Claude: “Write everything into memory.md”. Open a new chat, say “Read memory.md”, keep going. The chat is light again and the limit drains slower. Docs: https://docs.anthropic.com/en/docs/claude-code/memory

4. INSTALL CONTEXT7Without it Claude writes code from outdated docs and invents functions that don't exist. With Context7 it pulls fresh documentation as it works. Just ask Claude to install it. GitHub: https://github.com/upstash/context7

5. NOTIFICATIONSAsk Claude: “Notify me when you're done or waiting for my reply.” No more sitting by the screen — Claude calls you while you do other things. Setup docs: https://docs.anthropic.com/en/docs/claude-code/hooks

Free

Get the setup prompt

The article breaks down the whole setup in detail — and it includes the ready-to-use prompt that configures Claude Code correctly.

Read the article & grab the prompt

How to run it — the short version

Three steps. The prompt alone is not enough: the frames and the passport do half the work, and their exact build is the part we keep in the paid work.

Add CLAUDE.md

Create .claude/CLAUDE.md in your project and write your working rules there — Claude reads them in every chat.

Require a plan first

Add one rule: Claude drafts a plan and waits for your “ok” before touching code.

Rotate chats with memory.md

When a chat gets heavy, ask Claude to dump context into memory.md, open a fresh chat, and say “read memory.md”.

Install Context7

Ask Claude to install it — it pulls fresh documentation while working, so Claude stops inventing functions.

Turn on notifications

Ask Claude to notify you when it's done or waiting — stop babysitting the terminal. Docs: https://docs.anthropic.com/en/docs/claude-code/hooks

Want this done for your product instead of by hand? The offer below is built exactly for that — for you, and only today.

Free

Get an analytics agent for your channel

On our blog: how to build an agent that runs your channel analytics for you — what is working, what to post next, decided from your numbers instead of guesswork. It is the same agent-first approach that planned this ad. Look for the analytics-agent article on the AIevgen blog — it is free to read.

First order · 10 video creatives · 3–4 business days

Style Test Pack — $97

$97 first order only

10 AI ad videos for your product — the core of the pack
Bonus: a carousel for 1 product
Bonus: the Ready-to-Post Kit — everything prepared, you just publish

Made with the same system as the 5 Claude Code settings walkthrough ad above: passport, storyboard, frame-by-frame generation. Changed your mind before we start? 100% refund. The result is not your taste? One full remake — or up to 3 small fixes — until it is. The $97 price is for the first order only; after that, client terms apply.

Order the Style Test Pack — $97