
Setting up Claude Desktop: Environment Configuration, External Tool Integration and Plan Overview
About the author
I'm a member of the global sales team at Classmethod and today I'll cover three things: installing the Claude Desktop app on Windows, configuring the settings and files you need before your first project, and connecting Claude to external tools like Slack and Google Workspace.
Note: The installation process detailed below was done in 2026 using a Windows device. Refer to the official Claude documentation for the installation process for Linux and Mac devices.
Installing Claude
The quickest way to get access to all Claude tools (Chat, Cowork and Code) would be via the desktop app. The desktop app is available for macOS and Windows devices. They also provide a mobile app for use with the desktop app for iOS and Android. You can download the installers for your specific operating system directly from the Claude download page.
Tip: The download page also shows you possible integration options for Claude Cowork and Claude Code for apps such as Chrome, PowerPoint, Slack or environments such as your terminal or favored IDE.
The process for installing Claude on a Windows device after downloading the installer took about 3 minutes. I did encounter an AddPackage error during the installation but restarting the device solved the problem. It also automatically carries over any conversation history from the Claude web interface to the app as long as you sign in with the same account.


Claude Pricing and Available Plans
Claude offers two primary pricing structures: Subscription plans (a flat fee for a set usage amount) and API plans (pay-as-you-go pricing for building apps).
Below is an overview of the subscription plans available for individuals and businesses.
Pro-Tip: Always double-check Claude’s official pricing page, as Anthropic occasionally updates token allocations.
| Plan | Price (Monthly) | Usage Limits | Recommended For |
|---|---|---|---|
| Free | $0 | Standard usage (resets every 5 hours) | Beginners and casual users asking basic questions. |
| Pro | $20 | 5x more than Free plan | Heavy users, writers, and students who use AI for hours daily. |
| Max 5x | $100 | 5x more than Pro plan | Professionals needing deep research and complex logic. |
| Max 20x | $200 | 20x more than Pro plan | Power users and coders who want the absolute highest limits. |
| Team (Standard) | $25/seat | 1.25x more than Pro plan (Shared pool) | Small businesses or startup teams collaborating on projects. |
| Team (Premium) | $125/seat | 6.25x more than the Pro plan | Tech teams and developers |
| Enterprise | $20/seat + usage rate | Pay-as-you-go (Scaleable) | Large corporations requiring high security and admin controls. |
*Pricing as of April 2026.
These usage limits apply across all Claude tools (Chat, Cowork, and Code combined). Paid plans let you purchase additional usage if you hit your cap. Limits reset on two cycles: a rolling 5-hour window (short-term cap) and a weekly cap (long-term cap).
How to Configure Your Claude Desktop Environment
After signing in, you can change your settings to be more appropriate for your use case or work environment. Some settings you can consider checking out include:
Personal preferences
Things you would like Claude to consider in its responses (e.g., if you are a non-engineer, you could try asking Claude to make sure to include simple explanations for error messages or snippets of code).
Memory
Enables Claude to recall and reference your past chats. This means you don't have to re-explain your role or project context every time you start a new conversation: Claude already knows.
You can toggle this off from Settings → Capabilities if you prefer a clean slate each time. There's also an incognito mode (the ghost icon in the upper right) for quick one-off questions you don't want saved to your history.

Global instructions (Cowork specific)
General instructions you would like to apply for all your Cowork sessions. These can include your preferences or any context you would like Claude to know when you use Cowork (e.g., specific text or data formatting, context about your work)
Once you’re done looking through your settings, it is generally recommended that you do the following before starting with your very first project.
- Set the working directory
- Create your global CLAUDE.md file
Setting your working directory
For Cowork, you can set your working directory from the dropdown inside the prompt window. The default “work in a project” asks you to create or choose a project folder for your task but you can also just select the “choose a folder” option to set a general folder for your tasks.

Claude can read, write, and delete files from the working directory if you grant those permissions so don't point it at your Documents folder or desktop. I created a dedicated claude_work folder inside my home directory. This keeps Claude's files separate from everything else and makes it easy to find what it created. This general Claude folder will suffice for one-shot tasks but for any recurring projects, it is recommended that you create a project folder so that you can add a CLAUDE.md file to the project for any recurring instructions or context relevant to the project.
This brings us to the next thing you would want to set up before starting your first project: a CLAUDE.md file.
What is a CLAUDE.md file?
The CLAUDE.md file is a markdown file that Claude reads at the start of every session. This tells Claude about the what, how, and why of your project which will serve as important context for how it should behave. Skipping this setup is like hiring an assistant who forgets everything they learned the moment they clock out. Providing it with the bare essentials about what role you are in as well as what conventions they should follow or what they definitely should not do makes all the difference at the start of your day.
You can have one CLAUDE.md file for each project or subproject but for most non-engineers, a single global CLAUDE.md file is usually enough to give Claude high-level guidance on your role, preferred tone, and company standards.
You can generate a starter CLAUDE.md file using the /init command for Claude Code at the start of your project but this is made mostly for codebases and will produce a file with unnecessary build commands. In general, you could ask Claude (both Chat and Code should work) to draft a CLAUDE.md file for you and the general recommendation is to include who you are, what you do, and how you want it to be done. I went with the below prompt for drafting my CLAUDE.md file.
“Help me draft a global CLAUDE.md file for me as a non-technical user. I work for the global sales team for an IT company with cloud and AI services. I mainly use Claude for drafting sales reports, sales research and for checking communication in Japanese and English. Do not use technical jargon or always provide an explanation for any jargon or messages. Always go straight to the point without AI throat-clearing. Ask me a few questions if you need additional context before creating the file.”

It asked me a few questions about my language preference, tone and anything to avoid. Here's the full CLAUDE.md file Claude generated (also visible in the screenshot above):
# Global CLAUDE.md — Personal Instructions
## About Me
- Work in **sales** at an IT company specializing in cloud and AI services
- Non-technical background — avoid jargon, or explain it plainly if necessary
- Use Claude primarily for:
- Drafting and reviewing sales reports
- Sales research on clients and markets
- Checking and writing business communication
---
## Communication Style
- **Match my language** — respond in the same language I write in, or the language of the document I share
- **Be direct** — no greetings, no apologies, no filler phrases like "Great question!" or "Certainly!"
- **Business tone by default** — unless I say otherwise, keep all drafts and communications formal and professional
- **Use bullet points for long answers** — short answers can be plain text
---
## Tools I Use
Claude may have access to the following — use them when relevant to my request:
- **Gmail** — reading, drafting, and searching emails
- **Slack** — reading messages and drafting replies
- **Google Workspace** — Docs, Sheets, Calendar, Drive
- **HubSpot** — CRM for client and deal information
---
## How to Handle Errors
When something goes wrong:
- State clearly what failed
- Explain in plain language why it likely happened
- Give one or two concrete suggestions to fix it
---
## Before Making Changes
Always describe what you are about to do and ask for my confirmation before:
- Creating or editing any file
- Sending or modifying anything in Gmail, Slack, or HubSpot
- Taking any action that cannot be easily undone
You can further instruct Claude to customize the CLAUDE.md file for you but I went ahead with this draft and had it make the file for me. You can add context that Claude should always know (e.g., territory, key competitors etc. for sales) or anything it should definitely not do (e.g., do not use casual language for Japanese communication output) but the above draft should be sufficient to get started with using Claude.
Setting Up Claude Desktop Connectors (Slack, Google Workspace, HubSpot)
A big part of my expected tasks for Claude involve pulling sales or customer data from company tools to generate a high-level or project-specific report so the ability to integrate these tools automatically gives the Claude Desktop app a significant edge over the standard web interface.
Claude connects with external tools such as Slack and Google Workspace via what it calls “Connectors”. They have a full list of the available connectors on the Claude official site but I was able to find most of the tools I use in the default list from the desktop app.
To connect and configure your connectors, click Settings→Connectors or you can also access it from Customize→Connectors at the upper left menu of the app. Below are the connectors I’ve set up for now.

Each connector has its own permission settings, divided into read-only tools (e.g., searching emails) and write/delete tools (e.g., creating or deleting calendar events). You can set each permission to "always allow," "needs approval," or "blocked." I'd recommend starting with "needs approval" for write/delete actions, this way Claude asks before making changes, and you can loosen permissions later once you're comfortable. You can also get granular: for example, I allowed Claude to create Google Calendar events but blocked it from deleting them.

Another thing to note is that you might have to set up other Desktop connectors depending on the tools you use. For example, if you only use Google Drive via the Chrome browser, you will have to also set up the “Claude in Chrome” extension to allow Claude Desktop to access your browser.
Note: Site-level permissions for Claude are inherited from the Chrome extension and you will need to configure its permissions from the Chrome extension settings.

Summary and Next Steps
This setup was enough to get me started since I mostly use Claude through Cowork. If you plan to use Claude Code as your primary tool, you'll also probably need to set up your terminal, IDE, and Git. Refer to the official Claude Code quickstart for those steps.
I expect the above CLAUDE.md file and my overall setup to change over the course of learning about Claude and starting new projects so I’ll briefly go through what changed in my environment after a few months.
For my next post, I’d like to go over the concept of Prompt Engineering vs Context Engineering before starting to familiarize myself further with Claude’s features and models to ensure I have a more solid understanding of cost- and time-efficient Claude usage.
References
Claude Support- Install Claude Desktop
Download Claude | Claude by Anthropic
Plans and Pricing | Claude by Anthropic
Connectors | Claude
Claude Code Docs - Quickstart








