Learn by using GetRuthless
Keep this guide beside your Ruthless chat. Copy one prompt, add the example it asks for and paste it into “Message Ruthless”. Nothing is sent until you choose to send it.
The short answer
A useful checklist describes what must happen, what must never happen and how someone will prove it. Give Ruthless your app’s purpose and fake examples, ask it to explain unfamiliar terms, and turn each important rule into a check with a visible result and an owner.
Describe the app before collecting security labels
An AI-built app can have attractive screens while its private information is poorly protected. Start with a short description: who uses it, what they save, what they pay for and who can manage other people’s accounts. “Secure the backend” is too vague to evaluate. “One customer must never download another customer’s invoice” gives someone a rule to inspect.
The backend means the service working behind your screens. A database is where the app saves information. A permission rule decides who may read or change that information. A secret key lets the app use another service. Understanding these four terms is more useful than collecting a long list of unfamiliar acronyms. OWASP’s testing guide provides broader coverage when you or your builder need more detail. [1]
Build a short checklist in GetRuthless
Open GetRuthless chat, inspect Choose model and select an available model suited to explaining your app. Paste an ordinary description, not private customer files. Name the kind of data without sharing its contents. Include an example of the expected rule and explain which features are real today. A checklist for a fictional future payment feature does not establish anything about the app you have now.
Ask for a few high-value checks first. Each should have an owner and evidence you can recognise. If the model asks for code, have your builder supply only the relevant redacted excerpt. GetRuthless cannot see your project files or live configuration unless you bring that context.
Help me understand security in my own appointment app. Customers save an address and book a visit; staff see assigned visits; customers pay for visits. Use fake accounts and payment test mode in a private copy, not real customers. Give me five plain-English checks covering who can see data, private service keys, paid status, deletion and restoring a backup. For each explain the term, the expected rule, who should run it and the observable evidence. Identify unanswered product questions. Do not call the app secure from this description.Make each check answerable
For privacy, a useful row in your own notes says: “Fake customer A cannot open fake customer B’s address; builder checks the returned data and B’s unchanged record.” For money, it says: “An unpaid fake booking does not become paid because someone changed the screen; builder checks the payment provider’s test result and the saved booking.” Test mode simulates payments without charging a real card.
For keys, ask which credentials are private and whether any reach the visitor’s browser or shared project history. For deletion, check what the delete button removes, what remains and whether that matches your stated policy. Do not let the model invent a retention promise for your business.
For recovery, a backup is a saved copy intended to help recover lost information. A restore check means loading it into a separate private environment and proving a known fake record can be opened there. “A backup job ran” does not prove restoration works. Keep real customer data out of the material pasted into chat.
Own one unanswered item
Your exercise is to choose the most important checklist item and ask Ruthless: “What exact result would disprove this claim?” If the claim is that only assigned staff can see addresses, the disproof is an unassigned fake staff account receiving one. Record the actual result and date, along with anything untested. A checklist is a map of evidence and gaps; ticking boxes without that evidence does not improve the app.
Common questions
Should I fix everything the model mentions?
First establish whether it applies and whether there is a reproduced problem. Ask for the relevant code or observable check rather than accepting speculation as a finding.
Is this a complete security audit?
No. It is a beginner method for organising a few important checks. Sensitive data, money and complex sharing can need specialist testing.
Sources & further reading
The example prompts and workflows are GetRuthless guidance. Source references support the concepts explained; they are not an endorsement of GetRuthless.
Put the guide into practice.
Open Ruthless and choose an available standard or abliterated model suited to your task. Copy a prompt above, add the context it asks for and check the result. You can compare models in the same workspace; see current pricing before sending requests.