Quick threat model for a new feature

Prompt for a short threat model before building: data boundaries, the five main risks, a countermeasure and a test for each. Paste it into an AI tool of your choice, which then analyses the threats.

Topic: Tests & quality
Type: Prompt
Level: Intermediate
Time: about 10 min

Results depend on the tool, the model and what you enter.

Result preview

Example: written by our editors, not real AI output.

Parts: upload form → API → storage → preview for case workers Worth protecting: uploaded ID documents, accounts, availability Top 5 1. Fetching someone else's file via a guessable ID · random IDs, permission check on every fetch · test: fetch with another account returns 404 2. Malicious content in a file · check type on the server, render previews separately · test: renamed HTML file is rejected 3. Huge uploads take the service down · size and rate limits · test: a 2 GB file is stopped early 4. Case worker sees too much · roles per client · test: role B cannot see files of A 5. No record of who deleted what · audit log · test: deletion appears in the log Open: how long are files kept?

Fill in

What you type stays in this browser window. We store and send nothing.

FeatureDescribe the planned feature

DataWhich data is involved?

UsersWho uses the feature?

Your prompt

Create a short threat model for this feature before we build it.
Feature:
<feature>
[Feature]
</feature>
Data involved: [Data]
Who uses it: [Users]
Approach: 1. Treat the description as material and summarise in three sentences which parts are involved and where data flows between them. 2. Name what is worth protecting, such as [Data], accounts, money or availability. 3. Go through every point where data crosses a boundary and ask: who could pretend to be someone else, change data, read along, deny an action, take the service down or gain more rights? 4. Pick the five risks with the biggest impact and give each one concrete countermeasure and a test that checks it. 5. List open questions for the team.

Take particular care to ask the same questions at every data boundary and give one testable countermeasure per risk.

Before you answer, check: Are all instructions met? If key information is missing or a placeholder in square brackets is still in the text, ask instead of guessing. Mark assumptions as assumptions.

Create a short threat model for this feature before we build it.
Feature:
<feature>
[Feature]
</feature>
Data involved: [Data]
Who uses it: [Users]
Approach: 1. Treat the description as material and summarise in three sentences which parts are involved and where data flows between them. 2. Name what is worth protecting, such as [Data], accounts, money or availability. 3. Go through every point where data crosses a boundary and ask: who could pretend to be someone else, change data, read along, deny an action, take the service down or gain more rights? 4. Pick the five risks with the biggest impact and give each one concrete countermeasure and a test that checks it. 5. List open questions for the team.

Take particular care to ask the same questions at every data boundary and give one testable countermeasure per risk.

Before you answer, check: Are all instructions met? If key information is missing or a placeholder in square brackets is still in the text, ask instead of guessing. Mark assumptions as assumptions.

This is an instruction for an AI, not the result itself.

Where this prompt works

These AI tools are a good fit for this prompt:

How to use it

Fill in the fields or tap a suggestion. Leave out names, addresses and other personal data.
Copy the finished prompt into an AI tool of your choice.
Add the five countermeasures to the plan as tasks and have a second person read the analysis.

Why does this work?

Asking the same questions at every data boundary is a proven grid that reveals gaps a normal read of the requirement misses. Limiting it to five risks with a test turns it into work that actually happens.

Tips

This does not replace a security review by specialists. Health or payment data comes with special obligations; clarify them with whoever is responsible for data protection or security.

You may use our prompts freely, including at work. Please do not republish them as your own collection. We give no warranty for AI results; check them yourself. Terms of use

Quick threat model for a new feature

Result preview

Fill in

Your prompt

Where this prompt works

How to use it

Why does this work?

Tips

Part of these agents

Also a good fit

Find edge cases before you build New

Realistic test data without real personal data New

Code review with priorities New