Threat Modelling in the SDLC, 20 min read

3.R Review and references

In this lesson you will:

  • Recap the main points of Module 3
  • Test yourself on six review questions
  • Find the sources cited in this module

This page closes Module 3, Threat Modelling in the SDLC. Citation numbers in square brackets in the sections refer to the reference list at the end of this page.

Module summary

  • Threat modelling connects to every SDLC phase: abuse cases and security requirements, design decisions, build tasks, tests, release sign-off of residual risk, and feedback from operations and incidents.
  • Traceability links each threat to a requirement, a ticket, a test and monitoring, so anyone can check that a mitigation exists and works.
  • NIST SSDF PW.1.1, OWASP SAMM's Threat Assessment practice and the Microsoft SDL require threat modelling; sector standards and product-security regulation increasingly expect it.
  • Traditional threat models are deep but go stale. Continuous threat modelling keeps a baseline current with incremental, team-owned updates.
  • Update the model when the Level-1 diagram would change.
  • In agile teams, use three cadences: epic sessions, a five-question story check, and continuous upkeep. Write security acceptance criteria that can be tested.
  • Threat modelling as code keeps models versioned and reviewable, and can gate pipelines, but cannot find business-logic flaws.
  • Scale with risk tiers, security champions, templates, training and metrics focused on coverage, freshness, flow and effectiveness.

Review questions

1. Which artefact is written from the attacker's point of view during requirements?

  • A. A security acceptance criterion
  • B. An abuse case
  • C. A threat register entry
  • D. A residual risk acceptance

2. What is the MAIN weakness of traditional, document-centric threat models?

  • A. They cannot use STRIDE
  • B. They go stale as the system changes and are rarely owned by the delivery team
  • C. They are too shallow to find design flaws
  • D. They cannot be shown to auditors

3. Which change should MOST clearly trigger a threat model update?

  • A. Upgrading a logging library to a patch release
  • B. Adding an AI agent that can call the refund API
  • C. Changing the wording of an error message
  • D. Renaming a database column

4. What can a threat-modelling-as-code tool NOT reliably do?

  • A. Show new boundary-crossing flows in a pull request diff
  • B. Generate baseline threats for unencrypted flows
  • C. Detect that refunds may exceed the original payment amount
  • D. Fail a build when a High threat has no linked ticket

5. A security team of three cannot keep up with requests from twelve squads. What is the BEST structural fix?

  • A. Require every feature to wait for a security-led session
  • B. Introduce risk tiers and trained security champions in each squad
  • C. Replace sessions with an automated scanner
  • D. Threat model only after each release

6. Which metric BEST indicates threat modelling effectiveness?

  • A. Total number of threats recorded per quarter
  • B. Number of sessions held
  • C. Share of pen-test and incident findings that models had anticipated
  • D. Average length of threat model documents
Show answers
  1. B. Abuse (misuse) cases describe what an attacker wants to do and produce security requirements.
  2. B. Point-in-time documents fall behind the system and sit outside the team's tools.
  3. B. A new AI component with a money-moving tool adds capability and crosses boundaries.
  4. C. Business-logic rules are invisible to rule-based generators.
  5. B. Tiering focuses experts on the highest risk; champions distribute routine work.
  6. C. It measures whether models predict real findings. Counts of threats or sessions measure activity, not quality.

References

  1. NIST. SP 800-218: Secure Software Development Framework (SSDF) Version 1.1. 2022; and SP 800-218A: SSDF Community Profile for Generative AI and Dual-Use Foundation Models. 2024.
  2. OWASP Foundation. Software Assurance Maturity Model (SAMM) v2, Design → Threat Assessment. https://owaspsamm.org/
  3. Microsoft. Microsoft Security Development Lifecycle and Microsoft Threat Modeling Tool. https://www.microsoft.com/securityengineering/sdl
  4. OWASP Foundation. pytm: a Pythonic framework for threat modeling. https://owasp.org/www-project-pytm/
  5. Threagile. Agile threat modeling toolkit. https://threagile.io/
  6. OWASP Foundation. Threat Dragon. https://owasp.org/www-project-threat-dragon/
  7. Synopsys/Black Duck. Building Security In Maturity Model (BSIMM), Architecture Analysis practice.