ThreatModeler
Also known as: ThreatModeler Nexus, IriusRisk
Governed agentic threat modeling platform whose multi agent system builds, maintains, and reports on threat models grounded in a persistent Secure Design Graph, threat modeling applications, cloud, OT, and AI agents at enterprise scale with a deterministic framework and Bring Your Own AI.
ThreatModeler, based in Jersey City and founded in 2010, builds agentic threat modeling and secure design software that uses AI to find design flaws before software ships. In January 2026 ThreatModeler acquired IriusRisk, another early automated threat modeling vendor, and iriusrisk.com now describes IriusRisk as part of ThreatModeler. The combined company draws on a curated library of threats, security requirements and modeled components. Its premise is that as AI writes a growing share of production code, finding flaws has become cheap, and the harder work is confirming what matters, catching what is missing and proving it to auditors.
Its flagship, ThreatModeler Nexus, pairs three named agents with a deterministic framework. A System Mapping Agent builds a system map from architecture artifacts or infers one from code, a Graph Agent grounds and updates that work in the customer's environment, and a Reporting Agent produces audit evidence, all working on the Secure Design Graph, a maintained representation of systems, threats, controls and compliance mappings.
The platform threat models applications, cloud infrastructure, operational technology and AI agents, imports from ticketing, diagramming and enterprise architecture tools, runs in CI/CD pipelines, and takes a Bring Your Own AI approach so the customer chooses the model. Every threat, control and decision traces back to architecture, and deployment is offered as SaaS, on premises or in a private cloud. ThreatModeler names an MCP Server that signs clients in with OAuth 2.1, but its endpoint and tools are not published.
ThreatModeler Nexus carries ISO 27001:2022 and SOC 2 Type II, and it received FedRAMP Moderate authorization on 16 September 2026 through its partnership with Knox Systems, whose federal cloud it runs on. There is no published SSO, role model or audit log for customers. There is no approval step before agent output is committed, so human review sits in the customer's own process. It fits security organizations that need repeatable, auditable threat models across a large portfolio of applications, cloud and AI systems. A small team wanting a quick one off diagram will find it built for enterprise process.
Vendor details
Canonical URL
https://www.threatmodeler.ai
Category
Security / SOC agent
Subcategory
Agentic threat modeling and secure design
Funding status
Independent and privately held, headquartered in Jersey City, New Jersey, founded in 2010. In January 2026 ThreatModeler acquired IriusRisk, combining the two companies' threat libraries and technology; iriusrisk.com now describes IriusRisk as part of ThreatModeler. ThreatModeler has cited thirteen granted patents, more than a decade of curated research, and customers in regulated industries including Charles Schwab.
Company status
acquired
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
ThreatModeler imports from ticketing, diagramming and enterprise architecture tools, runs in CI/CD pipelines and against Infrastructure as Code, and connects to live cloud infrastructure for continuous risk identification. It names an MCP Server for IDEs and AI coding agents, whose page returns a 404, and pushes security requirements into developer sprints.
In practice
Your AI copilots ship code faster than your security team can review the designs behind it, so threat modeling becomes a stale, manual bottleneck. ThreatModeler's agents build and continuously update models from your architecture and code automatically.
You need to prove your security posture to auditors across many systems and frameworks, but evidence is scattered and inconsistent. ThreatModeler's Secure Design Graph is a system of record where every threat and control traces to architecture and maps to compliance.
You want to use AI for threat modeling without model lock in or inconsistent, unrepeatable results. ThreatModeler's Bring Your Own AI and deterministic framework ensure the same architecture in produces the same threats out, every time.
Sources & related URLs
Research sources
Agentic Index coverage score
10.0 / 14 capabilities · 71%
| Integrations & Tool Calling | Full |
|---|---|
|
Integrations cover ticketing, diagram and enterprise architecture import, CI/CD and cloud, and the ThreatModeler MCP Server brings threat modeling into AI assistants, GitHub and IDEs. The MCP Server works with ChatGPT and Claude, with editors such as Cursor, VS Code with Copilot and Windsurf, and with CI/CD systems including GitHub Actions, GitLab CI, Azure DevOps and Jenkins. SourceThreatModeler, threatmodeler.ai/platform and /why-threatmodeler/developersread 2026-10-05 |
|
| Workflow Orchestration | Full |
|
Three agents hand work along under a deterministic framework. The System Mapping Agent turns documents and diagrams into models, the Graph Agent enriches the Secure Design Graph with components and threats, and the Reporting Agent produces audit ready reports. System mapping also reads infrastructure as code and cloud context. The Graph Agent adds the controls and patterns specific to the customer's business, and the Reporting Agent writes business reports from the same governed data that produced the model. SourceThreatModeler, threatmodeler.ai/platformread 2026-10-05 |
|
| Knowledge Grounding & RAG | Full |
|
The Secure Design Graph is a persistent single source of truth connecting the customer's architectures, components, threats, controls and security decisions, grounded in a curated threat library, and every agent works on it. ThreatModeler calls it a living system of record that changes as the software does. Models follow the STRIDE, PASTA and MAESTRO methods. SourceThreatModeler, threatmodeler.ai and /platformread 2026-10-05 |
|
| Human Oversight & Guardrails | Partial |
|
A deterministic framework governs what the agents produce, so the same architecture gives the same threats and controls. There is no step where a person reviews or approves a generated model before it is saved. When a pull request opens a risk, the threat and the fix go back to the developer in the pull request, in a form a coding assistant can act on. SourceThreatModeler, threatmodeler.ai/platform and /why-threatmodeler/developersread 2026-10-05 |
|
| Security, Identity & Governance | Partial |
|
ThreatModeler Nexus carries ISO 27001:2022 and SOC 2 Type II. It is now FedRAMP Moderate authorized, where it earlier offered FedRAMP Moderate through a partner environment. There is no published SSO, SCIM or role model for the platform. The MCP Server signs clients in with OAuth 2.1. SourceThreatModeler, threatmodeler.ai and /why-threatmodeler/developersread 2026-10-05 |
|
| Observability & Auditability | Full |
|
Every decision the agents make traces back to the architecture, controls and compliance requirements behind it, with audit ready evidence produced by the Reporting Agent. Every pull request is checked against the model, and reports map to frameworks such as NIST, ISO 27001, PCI DSS, HIPAA, DORA and the EU AI Act. SourceThreatModeler, threatmodeler.ai/platformread 2026-10-05 |
|
| Memory & State Persistence | Not documented |
|
The Secure Design Graph, which ThreatModeler says improves every future model, is the platform's knowledge base, not memory the agents keep. The agents keep no memory of their own, with set contents or a set lifetime. SourceThreatModeler, threatmodeler.ai/platformread 2026-10-05 |
|
| Deployment & Data Residency | Full |
|
ThreatModeler Nexus is offered as SaaS, on premises or in a private cloud, with regional flexibility. Sourcethreatmodeler.ai/platformread 2026-09-28 |
|
| Prebuilt Agents, Templates & Packs | Full |
|
Customers adopt items from a curated library of threats, controls and components into their models. Reports map to more than 180 compliance frameworks, including PCI, NIST, HIPAA and ISO, and three named agents do separate jobs. The product line also includes CloudModeler and ThreatModeler MCP alongside the Nexus platform. SourceThreatModeler, threatmodeler.airead 2026-10-05 |
|
| Triggers & Channel Coverage | Full |
|
Threat modeling runs from the CI/CD pipeline and on diagram and architecture imports, so changes start the work, not a person. Every pull request is checked against the model automatically, with governance applied to each change, in GitHub Actions, GitLab CI, Azure DevOps or Jenkins. When a change opens a risk, the threat and the fix come back in the pull request. SourceThreatModeler, threatmodeler.ai/platform and /why-threatmodeler/developersread 2026-10-05 |
|
| Model Flexibility & Routing | Full |
|
Customers bring their own model, and the platform's agents run on it under the same governance, so the choice of model is the customer's. ThreatModeler does not list which providers it supports. Bring your own model sits among the platform's governance controls, beside the deterministic framework. SourceThreatModeler, threatmodeler.ai/platformread 2026-10-05 |
|
| APIs, SDKs & MCP Extensibility | Partial |
|
The platform has its own MCP Server for AI assistants, GitHub and IDEs. It signs clients in with OAuth 2.1 and works with ChatGPT, Claude, Cursor, VS Code with Copilot and Windsurf, but no endpoint, tool list or API reference is published. When a change opens a risk, the fix comes back in a form a coding assistant can act on. SourceThreatModeler, threatmodeler.ai/platform and /why-threatmodeler/developersread 2026-10-05 |
|
| Testing, Debugging & Optimization | Partial |
|
Outputs are deterministic. The same architecture in gives the same threats and controls out, so results repeat. There are no test cases or scores for checking the agents, and no check that a change to the agents or model must pass. Customer results it cites include half the threat modeling effort at Charles Schwab, models built five times faster and ten times as many models produced. SourceThreatModeler, threatmodeler.ai and /platformread 2026-10-05 |
|
| Browser & Computer Use | Not documented |
|
No agent drives a browser, desktop or remote computer. The agents work from design, code and cloud artifacts. They take in documents, diagrams, infrastructure as code and cloud context, and return threats and fixes through reports and pull requests. SourceThreatModeler, threatmodeler.ai/platform and /why-threatmodeler/developersread 2026-10-05 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Product updates
NexusAI, the new agentic experience inside ThreatModeler Nexus, builds a threat model from a diagram, design document, screenshot, infrastructure as code file, repository or plain description, asks about missing details and matches components to the customer's libraries. It recommends which threats and requirements apply with evidence strength and confidence signals, and the team decides which recommendations to apply. Scheduled daily or weekly checks of GitHub, GitLab, Google Workspace and Jira or Confluence through MCP detect architectural drift and stage a model update for review.
Bears on: Human approval / guardrails
View sourceThreatModeler updated its Intelligent Threat Engine (ITE) to detect advanced AI-specific threats at the architectural design stage. The release introduces curated threat libraries that map agentic tool compromise and Model Context Protocol (MCP) vulnerabilities to industry frameworks such as MITRE ATLAS and OWASP Agentic AI Threats.
Bears on: Security / enterprise
View sourcePricing
Not public; quoted through enterprise engagement after a solutions engineering session
enterprise subscription; basis not disclosed
What is public
No list pricing. ThreatModeler sells to enterprises through direct engagement and a solutions engineering session, but rates are not public.
Billing mechanics
Rates and billing basis are not disclosed. Inference, not stated by the vendor: an enterprise subscription scaled to the applications, users and modules deployed across applications, cloud, OT and AI coverage.
Cost watchouts
Inference, not stated by the vendor: cost may scale with the number of applications and systems modeled, users, and modules, so broad enterprise rollouts could grow the total.
Variable cost rationale
Inference, not stated by the vendor: likely priced against applications modeled, users and modules, so cost would grow as threat modeling scales across a larger portfolio.
Additional watchouts
Confirm whether pricing is per application, per user, or a platform subscription, and how coverage across cloud, OT, and AI agents affects the quote.
Sales call required
Yes, required for paid access
Free / trial
Book a session with a solutions engineer; no public free tier
Key ambiguities
No public rate is disclosed, and whether pricing is per application, per user, or a platform subscription with modules is not clear.
Related vendors
- 7AI — Agentic SOC from the Cybereason founders: domain specialized agents…
- Abnormal AI — Behavioral AI email security with three named autonomous agents: a…
- AirMDR — AI native MDR whose AI analyst investigates every alert and writes…
- Airrived — Agentic OS that unifies SOC, GRC, IAM, vulnerability management, IT,…
- Andesite — Human-AI SOC platform where customers build their own agents and…
- Anvilogic — Agentic security operations platform that works on top of existing…
Alternatives to ThreatModeler
The closest documented capability profiles to ThreatModeler among security and SOC agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Airrived10.0 / 14Fuller documented coverage on Human Oversight & Guardrails
- Crogl9.0 / 14A lighter documented profile than ThreatModeler
- UnderDefense8.5 / 14A lighter documented profile than ThreatModeler
- ContraForce11.0 / 14Fuller documented coverage on Human Oversight & Guardrails and Security, Identity & Governance
- CrowdStrike12.0 / 14Adds documented Memory & State Persistence
- Cyware10.0 / 14Fuller documented coverage on Human Oversight & Guardrails and APIs, SDKs & MCP Extensibility
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded