Create a defensive threat model for a fictional document-upload workflow with assets, trust boundaries and prioritized mitigations.
A fictional workshop portal lets an authenticated organizer upload a PDF and a reviewer approve it before public download. The design has a browser, application, private file store and public download endpoint. The exercise evaluates the design only; no exploit or live scan is performed.
| Asset | Required property |
|---|---|
| Draft document | private before approval |
| Published file | integrity and correct version |
| Organizer account | authorized access |
| Review record | traceable approval |
Start with the intended workflow and the assets that matter. Draw data flows and trust boundaries without assuming every component is equally trusted. Confidentiality, integrity and availability help frame consequences. Keep the model specific enough that a reviewer can trace a threat to an actual flow.
Draw the four components and mark draft versus public data.
Data-flow model
Record your result and evidence here or in a separate file.
The private store is not treated as a public download directory.
Label each arrow with data and direction.
Authentication establishes an identity; authorization decides whether that identity may perform an action on a resource. Check upload, review and download separately. A role name alone does not prove enforcement. The case should include a question about whether one organizer can access another organizer’s draft.
Create an action-by-role matrix and three authorization questions.
Access assumptions
Record your result and evidence here or in a separate file.
Every sensitive action has a server-side authorization requirement.
Avoid relying on a hidden button as the access control.
Write threats as a cause, an affected asset and a consequence. Examples include unapproved draft exposure, file-type confusion and replacement of an approved version. Keep the review defensive: describe the condition to prevent and evidence to request, without attempting attacks against a real portal.
Record six threats tied to the diagram and existing or proposed controls.
Threat register
Record your result and evidence here or in a separate file.
Each threat references a component or flow and a specific consequence.
Replace generic hackers may attack statements with a bounded failure scenario.
Prioritization combines impact, plausible exposure and uncertainty. Scores can organize discussion but should not imply precision unsupported by evidence. Assign owners and verification methods to mitigations. Distinguish a proposed control from one confirmed implemented and tested.
Prioritize three threats and write acceptance evidence for each mitigation.
Mitigation plan
Record your result and evidence here or in a separate file.
No proposed control is marked verified.
Add a status and evidence link field to every control.
A threat model is a living design artifact. Record assumptions, open questions and the changes that would trigger review. A new public-sharing feature can alter the boundary even if the code change is small. The handoff should make unresolved risk visible rather than ending with an unsupported secure verdict.
Write a review summary and triggers for re-evaluation.
Threat-model handoff
Record your result and evidence here or in a separate file.
The summary distinguishes design review from security testing.
List the unverified assumptions beside the recommendation.
Deliver the data-flow diagram, asset list, access matrix, six threats, three prioritized mitigations and review triggers. Add a public-share-link feature in a copy and identify which boundaries change.
The exercise stays within fictional defensive design and never claims that proposed controls were tested.
| Criterion | Points | Evidence |
|---|---|---|
| Evidence and correctness | 30 | Calculations, source use and results match the supplied case; limitations are honest. |
| Required artifacts | 30 | All five stage outputs and the capstone handoff are present and inspectable. |
| Reasoning and decisions | 20 | Choices, assumptions, alternatives and the case-specific critical condition are explained. |
| Reproducibility | 10 | A reviewer can follow the stated steps and distinguish expected from observed results. |
| Clear communication | 10 | Labels, units, structure and evidence references make the work easy to review. |
Score at least 80/100, supply every required artifact, and satisfy the path-specific critical condition. Rework missing evidence before marking complete.
Draft confidentiality requires more than authentication: the application must authorize the requester for the specific draft, and the storage path must not bypass that decision. Approval must bind to a particular file version so later replacement cannot inherit stale approval.
This is a design threat model, not a penetration test, compliance certification or security guarantee.