---
title: "Setup Audit for Claude (Free) | Guided Review of Your Company's Claude Setup"
description: "A free, guided audit you run inside your own Claude. It starts with a 10-minute check that lists what to fix, each finding marked easy, medium or hard, then works through the fixes with you and records progress on a private page. No account or email."
canonical: "https://www.atilab.io/claude-setup-audit"
publisher: "ATI Lab"
publisher_url: "https://www.atilab.io"
---

# Setup Audit for Claude (Free) | Guided Review of Your Company's Claude Setup

> A free, guided audit you run inside your own Claude. It starts with a 10-minute check that lists what to fix, each finding marked easy, medium or hard, then works through the fixes with you and records progress on a private page. No account or email.

Free guided audit

## Setup Audit for Claude

A free, guided audit you run inside your own Claude: it reviews how your company has Claude set up, fixes what it can with you step by step, and records progress on a private page.

Starts with a 10-minute check. No account, no email. Made by ATI Lab, a member of the Claude Partner Network. Not an Anthropic product.

Start here · Quick check

### 10 minutes to a list of findings, each marked easy, medium or hard

Every question below starts with the same check. Your assistant looks at what is loaded and set up, asks you a few questions, and agrees with you what to fix first. You can stop there and keep the list.

- See what is loaded in a normal session: connectors, tools and skills
- Count projects, plugins, org skills and scheduled tasks
- Ask about usage limits, onboarding, permissions and spend
- Agree the findings and what to fix first

### Start your audit

Every question starts with the same 10-minute quick check. The question sets which layers follow, and in what order.

2. Get the prompt for your assistant

No account and no email. Creating a run makes a private progress page for you.

Your prompt appears here. You can paste it into an assistant that can send web requests, or use the ATI Lab connector (https://www.atilab.io/mcp) in Claude chat or Cowork. If neither works, your assistant gives you a short code to paste back.

How it works

### Your assistant runs the audit. This site keeps the checklist.

- Pick your questionChoose the question closest to yours. It sets which layers follow the quick check, and in what order.
- Paste the promptYour assistant reads the audit instructions from this site. Nothing to install.
- Run the 10-minute checkIt looks at what is loaded and set up, asks a few questions, and lists what it found, each finding marked easy, medium or hard.
- Fix what you choose toStop there, or carry on layer by layer. It asks before it changes anything, and your private page ends in a report you can print.

What it covers

### After the check: 7 layers, each ending with something you keep

- FoundationsAbout 45 minutes **You end with:** A company context file, tidy projects, one page of usage rules and a way to bring people on. Follows the quick check for every question.
- Skills from your expertsAbout 50 minutes **You end with:** One skill built from a senior person's own method, an owner for it and a ranked list of the next ones. **Where it gets hard:** Organising many skills into plugins for each team, kept up to date from one place. Follows the quick check for every question.
- ConnectorsAbout 40 minutes **You end with:** A table of what Claude can read and change, unused access closed, and the heaviest connectors named. **Where it gets hard:** Replacing heavy general connectors with custom ones for your top services, behind one entry point. Follows the quick check for every question.
- Work that runs without youAbout 35 minutes **You end with:** The first process to automate, written as a spec someone can build from. **Where it gets hard:** Building the process outside the chat and trusting it to run on its own. Follows the quick check for 3 of the 5 questions.
- Plan and seatsAbout 20 minutes **You end with:** A plan recommendation with the reasons written down. Follows the quick check for 1 of the 5 questions.
- Right tool for the workAbout 20 minutes **You end with:** A map of your team's work against the assistant that suits it, and a two-week test. Follows the quick check for 1 of the 5 questions.
- Content from real workAbout 35 minutes · optional **You end with:** A brand voice file and a first piece written from work your experts actually did. **Where it gets hard:** Publishing straight to your website or its content system. Not tied to a question. Your assistant adds it when the quick check calls for it.

A full path is the quick check plus 4 layers: about 3 hours of work, usually over several sittings. You can stop after the quick check, and you can skip any step.

What it measures

### What the audit is trying to improve

- Hours returned each week on named tasks
- Usage spent per finished task
- Processes that run without a person starting them

What we store

### Four things, and nothing about who you are

- A random run id, which is also the link to your progress page
- The question you picked and, if your assistant sends one, its name (for example “Claude”)
- When the run was created and last updated
- For each step: done, skipped or blocked, and a summary of up to 400 characters written by your assistant

ATI Lab staff can see your progress page and read the summaries on it. Runs are deleted 90 days after their last update. We also count visits to this tool the same anonymous way as our other free tools; that count never includes your run id.

Your progress page has an optional form for asking ATI Lab to look at your results. Your email address reaches us only if you send that form, and it goes to our CRM, not into the run. The [privacy policy](https://www.atilab.io/privacy#setup-audit) has the details.

What we do not store

### Your conversation stays with your assistant

- Your name, email address or company name
- Your conversation with your assistant
- Files, context documents or skills you create during the audit
- Your IP address with the run

Summaries are written by your assistant, which is told to show you each one first and to leave out names and confidential details. We also remove anything shaped like an email address. Your progress page can be opened by whoever has its link and by ATI Lab, so share the link with care.

FAQ

### Questions about the audit

#### What is the Setup Audit for Claude?

A free, guided audit you run inside your own Claude: it reviews how your company has Claude set up, fixes what it can with you step by step, and records progress on a private page. It starts with a 10-minute quick check that ends in a list of findings, each marked easy, medium or hard, and the order to fix them in. After that, the question you pick sets which layers you work through. Each layer ends with something you keep, such as a company context file or a skill built from one of your own procedures.

#### How long does it take?

The quick check takes about 10 minutes, and you can stop there with your list of findings. A full path is about 3 hours of work, which most people spread over several sittings. Each layer shows its own time on this page.

#### Which assistants does it work with?

It is written for Claude: Claude Code, Claude chat and Cowork. The prompt works in any assistant that can read a web page. To update your progress page the assistant must be able to send web requests, or you add the ATI Lab connector (https://www.atilab.io/mcp) in Claude chat or Cowork. If neither is possible, the assistant prints a short code when you stop and you paste it into your progress page.

#### What do you store?

A random run id, the question you picked, timestamps, and for each step a status and a summary of up to 400 characters. The run holds no name, email, company, IP address or conversation. ATI Lab staff can read the summaries. Runs are deleted 90 days after their last update. If you send the optional form on your progress page, your email goes to our CRM so we can reply, not into the run.

#### Can ATI Lab read my conversation?

No. The audit runs inside your own assistant. We only receive the status of each step and the short summary your assistant sends, and ATI Lab staff can read those on your progress page. The instructions tell the assistant to show you each summary before sending it and to leave out names, email addresses and confidential details.

#### Do I have to finish in one sitting?

No. Stop after the quick check, or after any step. Your progress page keeps your place, and pasting the same prompt again continues from the next open step. The page is kept for 90 days after its last update.

#### Can I do all of it myself?

Most of it, yes. Some layers have a part that is hard to do alone, and this page says which under “Where it gets hard”. When you reach one, your report says so and describes what ATI Lab does for that part. Nothing in the audit requires it.

#### Who can see my progress page?

Anyone who has the link, and ATI Lab. The link contains a long random id and is not listed or indexed anywhere, so treat it like a password and share it only with people who should see the report. ATI Lab staff can open the page and read its summaries. We do not know who you are unless you send the optional form on that page.

#### Is it really free?

Yes. There is no account, no email gate and no charge. If the report shows steps where you got stuck and you would like help, you can book a call with us, but nothing in the audit requires it.

#### Is this an Anthropic product?

No. The Setup Audit for Claude is made by ATI Lab, a member of the Claude Partner Network. It is not an Anthropic product. It is a structured review of your setup, not a security or compliance certification.

### Need help understanding your results or planning next steps?

Book a call and bring your report, or write to [contact@atilab.io](mailto:contact@atilab.io).

More free tools

### Keep going

- [AI Business PlannerTurn the task that eats your team's week into a 90-day AI plan.](https://www.atilab.io/how-can-i-use-ai-for-my-business)
- [AI ROI CalculatorTurn hours lost to manual work into a savings range you can defend.](https://www.atilab.io/ai-roi-calculator)
- [AI Agent Readiness CheckerSee whether ChatGPT, Claude and AI agents can actually use your website.](https://www.atilab.io/ai-agent-readiness-checker)
- [AI Policy GeneratorGet an AI acceptable use policy your team can adopt today.](https://www.atilab.io/ai-policy-generator)
- [Claude ExplorerSee how Claude Code works across a dev team, safely.](https://www.atilab.io/claude-code-for-companies)

---
Source: https://www.atilab.io/claude-setup-audit

## About the publisher

ATI Lab is an AI consultancy and implementation agency (Lisbon and Sofia). It audits workflows, then builds and runs AI automations, agents and software, measured against the client's baseline.

- Services: https://www.atilab.io/services
- Case studies by industry: https://www.atilab.io/solutions
- About ATI Lab: https://www.atilab.io/about
- Book a strategy call: https://www.atilab.io/booking
- Contact: contact@atilab.io
- Site guide for agents: https://www.atilab.io/llms.txt