What You Actually Get When You Buy a Connector

199 euros for API code. If that sentence makes you suspicious, good. Most software is sold as a picture of a dashboard and a promise. You cannot see what you are buying until you have already paid, and often not even then.

A Toqool connector is the opposite. It is a folder of files on your machine, and you can open every one of them. This post walks through that folder, file by file, so you know exactly what 199 euros buys before you spend it.

The Whole Thing Is Readable

A connector is a set of Python scripts. Not compiled, not minified, not hidden behind a service. Plain text files you can open in Notepad. If you can read at all, you can skim them. If you know a little Python, or you paste them into Claude or ChatGPT and ask what they do, you can audit them line by line.

There is one external dependency: requests, the standard Python library for talking to web APIs. One. There is no framework, no SDK of ours, no package that pulls in forty other packages. This is deliberate. Every dependency is code you did not read, running with your credentials. We kept it to the one that half the internet already runs on.

• Readable Python scripts, one per data area of the platform
• A single dependency: requests
• verify_auth.py, which checks your credentials before anything runs
• A one-click run script and an optional daily scheduled pull
• A runbook that walks an AI assistant through your setup

verify_auth.py Is the Best Part

Every connector ships with verify_auth.py. It runs before any data is pulled, and its only job is to check your credentials against the platform and tell you precisely what is wrong if something is.

Anyone who has set up an API integration knows the failure mode. You get a 401 or a blank screen, and now you are guessing. Wrong key? Expired secret? Missing permission? Consent not granted? Typo in the tenant ID? The platform will not tell you, so you burn an afternoon on trial and error.

verify_auth.py replaces the guessing. It checks each link in the chain separately and reports per failure. Not "authentication failed" but which credential, which permission, which step, and what to do about it. When it passes, you know the pull will work before you run it. When it fails, you know exactly where, and the fix is usually minutes instead of an afternoon.

Here is the honest admission: this script exists because API setup genuinely goes wrong, on every platform, for everyone. We did not engineer the problem away. We engineered the diagnosis, which in practice is what saves you the afternoon.

Read-Only, and It Says So

The toolkit is read-only by design. It pulls data out. It never writes anything back to your shop, your marketplace account, or your tenant. No script in the folder can change a price, cancel an order, or touch a mailbox. The toolkit prints this posture when it runs, and because the code is readable, you can confirm it yourself rather than take our word.

For Microsoft 365 this pairs with admin consent: you grant read permissions, and read permissions are all the code can use. The blast radius of a mistake is a CSV file on your own disk.

What Comes Out: Dated CSV Snapshots

Each run writes clean, dated CSV files to your own disk. Orders, products, whatever the platform holds, as plain tables with a date in the filename. Each new run adds a snapshot next to the old ones, so history builds up locally without you doing anything. After three months of daily pulls you have three months of your own data, in a format that will still open in twenty years.

The format is deliberately AI-legible. When you have a question, you hand one file to Claude, ChatGPT or Excel, and you share only the slice that question needs. Not your whole account, not a live API key, one CSV. The free dashboards at Toqool Local read the same files directly in your browser, and nothing is uploaded anywhere.

Setup: About 45 Minutes, Honestly

Every connector includes a runbook, a step-by-step document written for an AI assistant to walk you through setup. You open it with Claude or a similar tool, and it guides you through creating credentials on the platform, placing them, and running verify_auth.py to confirm everything works.

For Microsoft 365, budget about 45 minutes including admin consent. That is a real estimate, not a marketing one. Registering an app in Entra and granting consent has irreducible steps, and we would rather tell you that now than have you find out at minute 30. Shop platforms like Shopify and bol.com are simpler. After setup, the one-click run script pulls fresh data whenever you want it, and the optional daily schedule does it without you.

That is the whole anatomy. Readable scripts, one dependency, a credential checker that diagnoses instead of guessing, a read-only posture you can verify, dated CSVs on your own disk, and a setup that takes an honest 45 minutes. Nothing in the box is hidden, which is exactly the point. Read it first. Trust it second.