The Open Source Security Testing Methodology Manual is the science of security. Formally derived from quantum information theory and thermodynamics and translated into practical, real-world operations as an engineering discipline. Build your security on quantified measurements. Peer reviewed. Written for humans and machines.
OSSTMM tells you if what you have actually works and meets all necessary compliance requirements. No more guessing. Real metrics for true liability protection.
OSSTMM 4 is formally derived from quantum information theory and thermodynamics. The companion paper Security as a Natural Law (Herzog, Entropy 2026) provides the peer-reviewed physics. The methodology provides the translation to engineering. Security is not a best practice. It is a measurable property of any system, grounded in the same laws that govern quantum states.
OSSTMM 4 is the first security methodology designed for AI parsing as a consumer. Definitions are machine-parseable. Test modules are structured as discrete, executable instruction sets. All formulas and calculations written for AI application. Your AI security tools can read the methodology directly.
Many new channels join the original five: Human, Physical, Wireless, Telecommunications, and Data Networks. Additional testing modules for Web, Cloud, Mobile, AI, Source Code, OT/ICS, IoT, AI/Agents, PQE, and more. The methodology applies identically across all. All measurable the same way and all metrics comparable. What changes between channels is the tools, the techniques, and analysis.
Map your OSSTMM findings directly to NIS2, DORA, EU AI Act, GDPR, HIPAA, PCI DSS, ISO 27001, NIST CSF, and other regulatory frameworks. One security test produces evidence for multiple compliance requirements. Stop testing once for security and again for compliance.
They solve different problems. Both are needed. But only one measures what's actually happening and provides verifiable scores to solve liability issues and support decision-making.
| Capability | OSSTMM | NIST CSF / ISO 27001 |
|---|---|---|
| Tests actual operational security | Yes, primary purpose | No, assesses policy compliance |
| Quantifiable security metrics | RAV score, Attack Surface measurement | Maturity levels, qualitative ratings |
| Theoretical foundation in physics | Derived from quantum information theory | Best practices and expert consensus |
| AI system security testing | Dedicated AI Facet + Trust Controls for AI agents | AI governance guidance (emerging) |
| Channel coverage | Many channels: human to IoT, including cloud, mobile, AI, OT/ICS | Primarily IT-focused |
| Repeatable, peer-reviewed results | STAR reporting standard | Auditor-dependent |
| Open source | Free, peer-reviewed | Paid standards documents |
| Measures trust in systems | Natively | No |
The theoretical foundation behind OSSTMM 4.
O4's theoretical foundation comes from the peer-reviewed physics paper Security as a Natural Law (Herzog, 2026), which presents the hypothesis that security properties derive from quantum information theory and thermodynamics. Rather than treating security as a human construct of policies and best practices, the paper demonstrates that security behaviors emerge from the same physical laws that govern waves and information at the quantum level.
The five attributes that underpin the entire OSSTMM 4 framework (Characteristics, Context, Force, Resources, Narrative) are derived from five quantum information properties. The two fundamental forces that determine every security outcome (entropy and latency) are thermodynamic realities, not management abstractions. This means OSSTMM test results measure actual physical properties of a system's security posture, not compliance with arbitrary checklists.
The practical consequence: security testing becomes reproducible, comparable, and falsifiable across any channel, any organization, any regulatory framework. A RAV score computed in one engagement means the same thing in every other engagement because it measures the same underlying physics.
OSSTMM 4 is published as one Foundation volume which carries the theory, measurement, and workflow and additional channel-specific Facets that carry the testing methodology. Each Facet stands alone as a testable specification.
The master volume. Carries the load-bearing theory, the Attack Surface Mapping and Rules of Engagement, the RAV measurement model, the STAR reporting format, compliance mapping, and the four-phase workflow that every channel-specific Facet references. This is where OSSTMM starts.
25 chapters + 6 appendices · 102 pages · Includes RAV calculation companion and compliance mapping to NIS2, DORA, EU AI Act, GDPR, HIPAA, PCI DSS, ISO 27001, NIST CSF
The Social Engineering and Human Factors Channel
Test methodology for human-mediated engagements: social engineering, pretexting, physical-access-via-social, and the human factors that constitute a target's most porous perimeter. 17 test modules.
The Built Environment and Physical Access Channel
The built environment and physical access controls: barriers, surveillance, locks, environmental controls, and the operational security of physical spaces.
The RF and Unbounded-Medium Channel
Radio frequency and unbounded-medium communications: Wi-Fi, Bluetooth, RFID, NFC, satellite, and any signal that propagates without a physical conductor.
The Switched-Circuit and Telco-Perimeter Channel
Switched-circuit networks and the telco perimeter: PBX, VoIP, SIP, SS7, and the convergence of traditional telephony with data networks.
The Network-Layer and Protocol-Stack Channel
The network layer and protocol stack: routing, switching, firewalls, IDS/IPS, DNS, and the infrastructure that carries data between endpoints.
The HTTP-Borne Application Channel
HTTP-borne applications: authentication flows, session management, input handling, API security, and the application logic exposed through web interfaces.
The Tenanted-Infrastructure and Shared-Responsibility Channel
Tenanted infrastructure and shared-responsibility boundaries: IAM, storage, compute, network segmentation, and the security assumptions baked into cloud service models.
The Handheld-Endpoint and Mobile-Platform Channel
Handheld endpoints and mobile platforms: app sandboxing, data-at-rest, transport security, platform permissions, and the unique attack surface of devices that travel with the user.
The Model, Agent, and Retrieval-Augmented-System Channel
AI-system engagements: model prompt/context surfaces, training-data provenance, retrieval augmentation, agent action envelopes, and the new attack primitives (jailbreak, prompt injection, tool-use exploitation) specific to generative and agentic AI. 17 test modules including hallucination propagation, memory manipulation detection, and model replication testing.
The Pre-Deployment Source-Artifact Channel
Physics based analysis for testing and analyzing all source code before it runs in production.
The Operational-Technology and Industrial-Control-System Channel
Operational technology and industrial control systems: SCADA, PLCs, DCS, safety systems, and environments where availability outranks confidentiality.
The Constrained-Device and Embedded-System Channel
Constrained devices and embedded systems: firmware, update mechanisms, sensor integrity, and the security of devices too small to run traditional defenses.
Everything you need to run an OSSTMM-compliant security test.
Reformulated for V4. Log base-10 normalization, per-dimension scoring, JSON/spreadsheet companion for tool integration.
Security Test Audit Report. Machine-readable companion output. Mandatory report contents linked to specific compliance claims.
Map OSSTMM findings to NIS2, DORA, EU AI Act, GDPR, HIPAA, PCI DSS, ISO 27001, NIST CSF. One test, every framework.
Professional certifications for security analysts and penetration testers. Prove you can test security, not just talk about it.