Threat modeling use cases

Apply Threat Modeling Where Security Decisions Begin

The most important security decisions are made at design time: which components talk to each other, where data flows and what the system trusts. These are the situations where teams use Astrix to examine potential threats before a design is built or changed, and what they get out of it.

  • Application architectureStarts from an OpenAPI spec or diagram
    STRIDE
  • Cloud, API and integration designStarts from Terraform, CloudFormation, ARM or Kubernetes
    STRIDE · PASTA
  • AI and agentic systemsStarts from a description or diagram
    SAIF · MAESTRO
  • Security architecture reviewsStarts from the design under review
    Any framework
  • Consulting and client assessmentsStarts from each client's architecture
    VAST · any
Typical pairings for illustration. Any supported framework can be used with any input.

Why these use cases matter

Common security challenges, and the threat modeling response

These challenges apply to most engineering organizations, whatever tools they use. Threat modeling is the established practice for each one.

  • ChallengeSecurity is considered lateIssues surface in testing or production, when fixing them means rework.
    Threat modeling responseExamine the design firstReason about threats while the architecture is still easy to change.
  • ChallengeArchitectures are complexServices, clouds and third parties add trust boundaries that are easy to overlook.
    Threat modeling responseMap components and data flowsMake interactions and boundary crossings explicit so they can be reviewed.
  • ChallengeAttack paths are hard to seeIndividual weaknesses can look minor until they are chained together.
    Threat modeling responseReason about attacker pathsFollow how an attacker could move from an entry point to valuable assets.
  • ChallengeFindings are hard to communicateArchitects, developers and leaders need the same concern explained at different depths.
    Threat modeling responseUse shared, structured findingsDescribe each threat with standard severity and classification.
  • ChallengeDesigns keep changingA review done once goes stale with the next feature or integration.
    Threat modeling responseRevisit the model on changeTreat the threat model as a living document, not a one-off deliverable.

Use-case overview

Five threat modeling use cases for Astrix

Each use case below separates General practice (how threat modeling helps in that situation) from In Astrix (what the platform itself provides).

Use case 1

Application architecture threat modeling

  • Application architects
  • AppSec teams
  • Software engineering teams
General practice

The situation

A team is designing a new application, or a significant feature, and wants to know what could go wrong before code is written. Application threat modeling looks at the components, the data flowing between them and the trust boundaries they cross, then asks which threats apply at each point and what should be done about them.

At a glanceIn Astrix
Typical inputs
OpenAPI / SwaggerVisiodraw.ioGitHubDescription
Typical framework
STRIDE, for its six general threat categories
Who decides
Your reviewers accept, adjust or reject each finding
Typical outputA reviewed threat model with scored findings, mitigation work items in Jira, GitHub Issues or Azure DevOps, and a PDF or DOCX report.
In Astrix

How Astrix supports it

Import the API specification, a Visio or draw.io diagram, or the GitHub repository, or describe the application in plain language. Run the STRIDE analyzer, or another supported framework, to generate a draft threat model. Your team then reviews the findings, each scored with CVSS, mapped to CWE and MITRE ATT&CK, and accompanied by mitigation guidance.

Use case 2

Cloud, API and integration design

  • Solution architects
  • Cloud and platform engineers
  • Security engineers
General practice

The situation

A system is moving to new infrastructure, exposing a new API, or taking on an external dependency such as a payment or identity provider. Every new connection adds a trust boundary and a set of assumptions: who can call what, which data leaves your environment, and what happens if the other side misbehaves.

In Astrix

How Astrix supports it

Astrix imports infrastructure-as-code (Terraform, CloudFormation, ARM templates and Kubernetes YAML) alongside OpenAPI definitions, so the model starts from how the system is actually defined. Attack path visualization shows how a weakness at one boundary could lead to others, with kill-chain and lateral-movement analysis.

Astrix analyzes the definitions and descriptions you provide. It does not connect to or scan your live cloud accounts.

At a glanceIn Astrix
Typical inputs
TerraformCloudFormationARMKubernetes YAMLOpenAPI
Useful views
Attack paths, severity-by-likelihood heat maps
Typical outputFindings at each new boundary crossing, with mitigation guidance, re-analyzed when the design changes.
Example integration with trust boundaries Inside your cloud boundary, an API calls an orders service. The orders service calls an external payment provider, and the API relies on an external identity provider. Both calls to external providers cross a dashed trust boundary marked with a threat marker. YOUR CLOUDEXTERNAL Public APIOrders serviceIdentity providerPayment provider !!
Conceptual example, not a product screenshot. Markers (!) show where data crosses into external services.

Use case 3

AI, LLM and agentic systems

  • Product security teams
  • AI and ML engineering teams
  • Security architects
General practice

The situation

A product is adding an LLM-powered feature, or a set of agents that call tools and each other. These designs raise threats that traditional categories do not fully describe, such as manipulated model inputs, over-privileged agents and unintended actions between autonomous components. Frameworks written specifically for AI systems help teams ask the right questions.

At a glanceIn Astrix
Frameworks
SAIFMAESTROSTRIDE
Typical inputs
A system description or architecture diagram of the AI feature
Typical outputAI-specific threats with scores, mappings and mitigation guidance, alongside conventional findings.
In Astrix

How Astrix supports it

Astrix includes dedicated analyzers for SAIF, Google's Secure AI Framework for AI and ML systems, and for MAESTRO, an emerging seven-layer framework for multi-agent architectures. Teams can analyze the same design with STRIDE as well, to cover both conventional and AI-specific threats.

Use case 4

Security architecture design reviews

  • Security architects
  • Enterprise architects
  • Architecture review boards
General practice

The situation

An architecture review board or security architecture review must weigh security alongside function, cost and operations. Reviews work best when threats are presented in a consistent structure, when the discussion is recorded, and when decisions can be traced afterwards.

At a glanceIn Astrix
Review aids
Heat mapsAttack pathsCommentsRolesAudit trail
Shared with
Reviewers in the platform, or stakeholders via PDF, DOCX and scheduled email reports
Typical outputA documented review: findings, decisions, owners and the model version they apply to.
In Astrix

How Astrix supports it

Astrix gives reviewers a shared threat model to discuss:

  • severity-by-likelihood heat maps and attack paths to set priorities
  • comment threads and owner, editor and viewer roles for each model
  • an audit trail of changes
  • PDF and DOCX reports, including scheduled email reports for stakeholders outside the platform
  • snapshots that keep earlier versions of the model for reference

Astrix supports the review discussion and its record. It does not automatically approve or validate an architecture, and it does not connect to enterprise architecture repositories.

Use case 5

Security consulting and client assessments

  • Cybersecurity consultancies
  • Independent security architects
  • Channel partners
General practice

The situation

Consultants assess architectures for many clients. Each engagement needs a consistent method, a clear deliverable, and strict separation of one client's information from another's.

At a glanceIn Astrix
Separation
One isolated database per client organization
Administration
Partner admin role across managed organizations
Deliverables
PDFDOCX
Typical outputA repeatable threat model and report for each client, using the same methodology.
In Astrix

How Astrix supports it

Astrix is multi-tenant: each client organization is kept in its own dedicated database, and a partner administrator role can manage the organizations a partner serves. The same five framework analyzers give each engagement a consistent method, and PDF or DOCX reports provide the client deliverable.

How Astrix fits into the workflow

One engagement, five stages

Whatever the use case, a threat modeling engagement follows a similar path. Each stage is labelled to show whether it is a practitioner activity or a step Astrix supports directly.

  1. Understand the system

    Agree scope and gather the specs, IaC, diagrams or description.

    Astrix imports these
  2. Map components and flows

    Capture components and data flows; confirm where trust boundaries sit.

    Astrix models; you confirm
  3. Examine threats

    Apply the chosen framework to the model.

    Astrix analyzer drafts
  4. Analyze and document

    Score, classify and prioritize findings; agree what to accept or mitigate.

    Astrix scores; you decide
  5. Review and act

    Assign mitigations, share the report and revisit the model when the design changes.

    Your team, with Astrix tracking

For a detailed breakdown of who does what at each stage, see the threat modeling workflow on the platform overview.

Use cases by role

What each professional needs from threat modeling

  • Application security teams

    Their challengeMore applications to review than reviewers.

    Why threat modeling helpsFocuses scarce review time on design-level risk.

    Where Astrix relatesDraft threat models from existing artifacts, so experts start from findings rather than a blank page.

  • Security architects

    Their challengeApplying a consistent method across very different systems.

    Why threat modeling helpsGives every review the same structure.

    Where Astrix relatesFive methodology-specific analyzers, plus attack paths for cross-component reasoning.

  • Software and product engineering

    Their challengeSecurity feedback that arrives late or is hard to act on.

    Why threat modeling helpsSurfaces requirements during design.

    Where Astrix relatesFindings with mitigation guidance, delivered as Jira, GitHub Issues or Azure DevOps work items.

  • Cybersecurity consultants

    Their challengeRepeatable assessments across many clients.

    Why threat modeling helpsProvides a defensible, documented method.

    Where Astrix relatesIsolated client organizations, partner administration and PDF/DOCX deliverables.

  • Security leaders and CISOs

    Their challengeLimited visibility into design-level risk across the portfolio.

    Why threat modeling helpsMakes design risk visible and comparable.

    Where Astrix relatesConsistently scored findings, remediation tracking and reports across teams.

When to use threat modeling

Six moments that call for a threat model

These are recommended practices, not events Astrix detects automatically. Your team decides when to create or re-analyze a model.

  • Initial designBefore a new application or system is built.
  • Significant architecture changeA redesign, migration or new deployment model.
  • Sensitive data flowsWhen personal, financial or regulated data moves somewhere new.
  • New dependencyA third-party service, API or identity provider.
  • Architecture reviewAhead of a design review board or security sign-off.
  • Changed assumptionsAfter an incident, a new threat or a change in who uses the system.

Expected value

What teams can expect

Results depend on how a team uses threat modeling. Used consistently, Astrix is intended to help in these ways.

  • More structured design discussions

    A shared model and framework give reviews a common starting point.

  • Earlier view of architecture-level risk

    Threats can be examined before implementation choices are fixed.

  • Better visibility into potential threats

    Attack paths and heat maps show where findings cluster.

  • Clearer communication

    CVSS, CWE and MITRE ATT&CK give security and engineering a common vocabulary.

  • More consistent practice

    The same analyzers apply the same methodology across teams.

  • Better-informed decisions

    Mitigation guidance and a record of what was accepted or fixed support each decision.

Astrix supports threat modeling. It does not guarantee that a system is secure or compliant, and it does not replace security testing.

Why Astrix

Why teams choose Astrix for these use cases

Five methodologies, including AI-specific ones

Dedicated analyzers for STRIDE, PASTA, VAST, SAIF and MAESTRO let one platform cover conventional and AI systems.

Starts from your real artifacts

Specs, IaC, diagrams and repositories become the model's input, with plain-language description for early designs.

People stay in control

AI-assisted output is advisory. Reviewers accept every result, with roles, comments and an audit trail.

Built for sensitive design data

A dedicated database per customer, EU hosting, and no training on customer data. See platform security.

Resources

Threat modeling use cases FAQ

What is threat modeling?

Threat modeling is a structured way to identify what could go wrong with a system's security, usually at design time. It describes the system, examines potential threats against its components and data flows, and decides how to address them.

When should an organization perform threat modeling?

Ideally during initial design, and again whenever the architecture changes significantly: a new dependency, a new sensitive data flow, a migration, or ahead of an architecture review. See the six common moments.

Which teams can benefit from threat modeling?

Application and product security teams, security and enterprise architects, software engineering and DevSecOps teams, security consultants, and security leaders who need visibility into design-level risk. See use cases by role.

How does Astrix fit into an application or security architecture review?

Astrix provides the threat model the review works from:

  • it imports the design
  • it generates findings with the selected framework
  • it scores and maps each finding
  • it supports discussion through comments, roles, attack paths and reports

Reviewers remain responsible for decisions. Astrix does not approve or validate an architecture on its own.

What inputs and outputs does Astrix support?

Inputs: OpenAPI/Swagger; Terraform, CloudFormation and ARM templates; Kubernetes YAML; Visio and draw.io diagrams; architecture images; GitHub repositories; and plain-language descriptions.

Outputs:

  • scored findings with CWE and MITRE ATT&CK mapping and mitigation guidance
  • attack paths and heat maps
  • compliance gap analysis
  • work items for Jira, GitHub Issues and Azure DevOps
  • PDF and DOCX reports
How can a team evaluate whether Astrix suits its requirements?

Start with one of the use cases above that matches a real upcoming design, and ask for a demo using that system. If a requirement is not covered on this site, such as a specific integration or deployment model, raise it with the Cognexaa team so it can be confirmed before you commit.

How can I request a demo?

Email sales@cognexaa.com to book a demo, or use the demo request form.

Explore how Astrix fits your threat modeling needs

See how Astrix aligns with your application security and architecture review process, using a system that matters to you.

sales@cognexaa.com · Read the platform overview