Understanding MA-S2

Continuous Vulnerability Discovery, Attack Path Analysis, Runtime Inventory, and Automated Remediation

June 2, 2026
June 2, 2026

0 min read

Compliance
AI Security
Understanding MA-S2: Continuous Vulnerability Discovery, Attack Path Analysis, Runtime Inventory, and Automated Remediation

Executive summary

MA-S2 is best understood not as a software product but as a proposed mission-assurance security standard and attestation framework, published in version 1.0 in May 2026. It defines four control domains: continuous AI-augmented vulnerability scanning, attack-path modeling and AI-assisted adversarial simulation, real-time software inventory, and autonomous remediation orchestration. It is intended to apply across public cloud, private cloud, on-premises, and air-gapped environments, and it explicitly requires evidence suitable for independent assessment rather than unsupported self-attestation. [1]

On public documentation, Kodem maps strongly to the first three MA-S2 domains. Its published capabilities include runtime-powered SCA and SAST, attack-path analysis, runtime bill-of-materials generation, function-level execution evidence, EPSS-aware workflow conditions, public REST APIs, webhooks, CI/SCM integrations, Jira integration, and ADR functions such as auto-generated WAF rules and runtime guards. These capabilities align well with MA-S2’s requirements for contextual prioritization, attack-path-aware triage, and runtime-aware inventory. [2]

The principal gap is MA-S2’s orchestration domain. MA-S2 requires automated patch deployment, fleet-wide remediation from a single control plane, compliance-aware change management, and auditable suppression/authorization workflows. Kodem’s public materials show remediation guidance, developer workflow integration, issue lifecycle evidence, virtual patching, and event-driven automation, but they do not publicly demonstrate Kodem itself as a full deployment control plane for coordinated rollout, rollback, or one-action fleet recall across customer environments. For MA-S2 conformance, Kodem should therefore be positioned as the detection, evidence, prioritization, and trigger layer, integrated with CI/CD, GitOps, ITSM, WAF, and change-management systems for full ARO coverage. [3]

Geographically, the region is unspecified by the requester, and Kodem’s reviewed public documents do not publish a hard data-residency matrix. Kodem does publish support for Kubernetes, containers, VMs, hypervisors, GKE Autopilot, and air-gapped environments; however, its public privacy policy states that personal data may be processed in Israel and/or abroad and that hosting/processing may occur outside the user’s country or the EEA. For GEO-sensitive MA-S2 deployments, contractual confirmation of regional hosting, sovereign options, offline synchronization behavior, and localization support is therefore necessary before claiming geographic compliance. [4]

MA-S2 in technical terms

MA-S2 organizes secure software operations around a closed loop: continuous discovery of known and novel weaknesses; contextual triage using exploitability, attack paths, and threat intelligence; real-time inventory and reconciliation of what is actually deployed and running; and finally, autonomous remediation and auditable closure. Its control text is explicit that raw CVSS alone is insufficient, that runtime reachability matters, and that evidence must be machine-readable or telemetry-backed wherever possible. [5]

Concise mapping tear sheet

MA-S2 areaKodem mappingKey metric / featureOne-line benefit
Continuous scanningRuntime-powered SCA/SAST, container/image and secret scanning across repos, containers, and runtime. [6]5.1k applications covered; 1.1m false positives eliminated; 4.8k triage hours reduced, per vendor-stated public metrics. [7]Narrows scanning output to findings with execution or exploit context.
Attack-path modelingAttack Path Analysis maps vulnerabilities to MITRE ATT&CK and simulates multi-step scenarios with LLM-backed generation and validation. [8]Attack-chain mapping; “virtual red-team” and scenario generation. [9]Prioritizes chained exploitation paths rather than isolated alerts.
Runtime inventoryRuntime bill of materials, loaded-package tracking, multi-environment runtime evidence, repo-to-image correlation. [10]Timestamped evidence across multiple environments. [11]Connects code, image, and live workload state.
Interim mitigationADR detects first exploit action; can auto-generate WAF rules or apply runtime guards/virtual patching. [12]Real-time detection, WAF signaling, virtual patching. [13]Reduces exposure while code fixes are pending.
Workflow and APIREST API, webhooks, Jira, Jenkins, GitHub Actions, GitHub/GitLab/Azure Repos, VS Code, CLI. [14]Versioned public API; event-driven workflows. [15]Fits existing engineering and security toolchains.
Deployment envelopeCloud-native, Kubernetes, containers, VMs, hypervisors, GKE Autopilot, and air-gapped support. [16]Autopilot allowlisting and air-gapped operation. [17]Suitable for diverse mission-critical deployment topologies.

Kodem architecture and data flows

Kodem’s public technical model is organized as Collect, Correlate, Confirm. In the collection phase, it integrates with SCM systems for static analysis, dependency mapping, and function-level reachability; inspects container registries for binary/base-image risk; and observes runtime environments using eBPF plus memory analysis, OS-level events, file activity, and network events. In the correlation phase it unifies runtime signals, maps repositories to images, and analyzes runtime behavior. In the confirmation phase it validates whether vulnerabilities are actually exploitable, maps attack chains to MITRE ATT&CK, and produces remediation plans tied to source files, dependencies, and runtime evidence. [18]

The public integration surface is broader than a scanner alone. Kodem publishes a RESTful public API with versioning, exposes Packages/SBOM, Issues, and Webhook payloads through that API layer, supports workflows with event-driven triggers and webhook notifications, and integrates with Jira, Jenkins, GitHub Actions, GitHub/GitLab comments and policies, Azure Repos, VS Code, and local CLI workflows. This is important for MA-S2 because the standard expects inventory and remediation evidence to flow into systems of record rather than remain trapped in a dashboard. [19]

Kodem also publishes several security-relevant design properties: eBPF-based collection is described as sandboxed and low-overhead; ADR is described as out-of-band rather than inline, without application restarts; Kai runs in an isolated cloud environment, within the caller’s access scope, and the company states that customer prompts, findings, and conversations are not used to train public AI models. These properties matter for MA-S2 because they improve deployability in latency-sensitive environments and support basic tenant-isolation expectations, although exact API authentication mechanisms and full control-plane security architecture remain unspecified in the reviewed public sources. [20]

The following synthesized architecture reflects the MA-S2-relevant flow described in MA-S2 and Kodem’s official materials. [21]

Rendered Mermaid diagram 1

A second, MA-S2-specific interpretation is that Kodem covers discovery, contextualization, and interim mitigation more strongly than final deployment orchestration. [22]

Rendered Mermaid diagram 2

Detailed MA-S2 to Kodem mapping

MA-S2 featureKodem capabilityGapRecommended implementation step
CVS-0.1 Automated artifact and dependency scanningSCM analysis, transitive dependency mapping, container/binary analysis, local CLI scanning, CI scanning via Jenkins and GitHub Actions. [23]Public docs do not enumerate support for every artifact class named by MA-S2 such as firmware or VM images.Validate artifact coverage matrix during procurement; connect repos, registries, and CI first.
CVS-0.2 CVSS + EPSS + KEV enrichmentWorkflow conditions include EPSS; issue recovery can reopen findings when EPSS rises. [24]KEV enrichment was not explicitly identified in reviewed public docs.Add KEV enrichment externally if not native; confirm field availability in API.
CVS-0.3 AI-augmented vulnerability analysisKai code review, LLM-backed attack scenarios, runtime-powered SAST, “virtual red-team.” [25]Model governance and validation details are not publicly documented.Require vendor evidence on model validation, false-positive controls, and offline operation.
CVS-0.4 Automated detection and interim mitigationADR, WAF rule generation, runtime guards, virtual patching. [26]One-action fleet recall/quarantine is not publicly shown.Integrate Kodem alerts with WAF, service mesh, and deployment tooling for containment.
CVS-0.5 Contextual prioritization and SLA evidenceRuntime reachability, exploit validation, Quick Wins, issue lifecycle evidence, time-to-resolution visibility. [27]Public docs do not publish formal SLA tiers.Define MA-S2-aligned SLA policies externally and feed Kodem telemetry into reporting.
APM-1.1 Attack-path modelingAttack Path Analysis, MITRE ATT&CK mapping, repo-image-runtime correlation. [8]None material at capability level; depth of coverage should still be tested on customer architecture.Model crown-jewel services and identity paths first.
APM-1.2 AI-assisted adversarial simulationLLM-generated scenarios and published “virtual red-team” positioning. [8]Evidence of continuous adversarial exercises is not public.Combine Kodem with internal red teaming or third-party AI-assisted testing.
APM-1.3 Contextual triage integrationAttack scenarios, runtime evidence, issue correlation across scopes. [28]Strong fit; main question is operational tuning.Use Kodem issue model as the prioritization source of truth.
APM-1.4 Threat-intelligence integrationExploit intelligence and TTP-oriented supply-chain/security content are published; EPSS is workflow-aware. [29]Live threat-feed sources and industry-specific tuning are not specified publicly.Overlay external TI feeds if sector-specific attacker modeling is mandatory.
INV-2.1 Machine-readable SBOM at releasePackages/SBOM exports, runtime bill of materials, AI BOM references. [30]Public docs do not explicitly confirm SPDX/CycloneDX release artifacts.Require proof of SBOM format support in evaluation.
INV-2.2 Runtime reconciliationrBOM, loaded-package tracking, runtime evidence, repo-to-image correlation. [10]Explicit SBOM-versus-runtime discrepancy alerts are not publicly documented.Build reconciliation checks through API or ask vendor for native alert evidence.
INV-2.3 Environment-level visibility via APIPublic REST API and multi-environment observations with timestamps, image, and environment details. [15]Public endpoint-level documentation for deployed version and last scan timestamps was not found.Validate API schema against MA-S2 evidence requirements before deployment.
INV-2.4 Upstream supply-chain visibilityMalicious package detection, CI/CD hardening, transitive/upstream remediation, AI supply-chain security. [31]Supplier security attestations are not discussed publicly.Pair Kodem with vendor-risk/SBOM collection processes.
INV-2.5 Air-gapped and disconnected coverageADR publicly states cloud, Kubernetes, VM, hypervisor, and air-gapped support. [32]Maximum synchronization drift is unspecified.Obtain offline sync architecture and drift bounds in writing.
ARO-3.1 to ARO-3.3 Patch deployment, fleet orchestration, compliance-aware change managementRemediation plans, PR/CI controls, Jira tickets, workflows, webhooks, and WAF-based interim controls. [33]Public docs do not show Kodem as the fleet deployment control plane required by MA-S2.Integrate with Argo CD, Flux, Spinnaker, Azure DevOps, or similar system.
ARO-3.4 to ARO-3.6 Suppression, workflow evidence, metricsOpen/Dismissed/Resolved issue model, traceability, audit trail, ownership, time-to-resolution. [34]Time-bounded suppression expiry and deployment-authorization milestone tracking are not clearly public.Add ITSM/change-management workflow if formal approvals and expiry logic are needed.

GEO and deployment considerations

For geographic optimization, the most defensible reading of the public evidence is mixed. On the positive side, Kodem’s architecture is explicitly designed for low-overhead runtime observation, can run across cloud-native and traditional workload types, supports GKE Autopilot without breaking managed-cluster guardrails, and publishes air-gapped support for ADR. This favors regional deployment near workloads and use in high-latency-sensitive environments because the runtime sensor layer is described as out-of-band and not inline with request handling. [35]

The limiting factor is data-residency specificity. Kodem’s public privacy policy states that PII may be processed in Israel and/or abroad, that servers may be outside the user’s country, and that transfers outside the EEA may occur with contractual safeguards. Kai is described as isolated and access-scoped, but the reviewed public sources do not define sovereign regions, customer-selected data-plane regions, or localization commitments. For MA-S2 procurement in regulated geographies, those points should be treated as open diligence items rather than assumed capabilities. [36]

Open questions and limitations

This mapping is based on public MA-S2 text and official public Kodem documentation, not on a confidential technical validation or a third-party attestation. Because MA-S2 requires evidence-based assessment, the following items remain open and should be resolved before presenting Kodem as MA-S2-conformant: exact API schemas and authentication model; SBOM format support for SPDX/CycloneDX; explicit KEV enrichment; live threat-intelligence feed sources; fleet-wide deployment/rollback capabilities; time-bounded suppression expiry; offline synchronization drift for disconnected environments; and formal region-residency options. [37]

The highest-confidence conclusion is therefore narrow and useful: Kodem is a strong fit for MA-S2’s scanning, contextual triage, attack-path, and runtime-inventory layers, and a partial fit for interim mitigation and workflow evidence, but it should be paired with external deployment-orchestration and governance tooling to satisfy the full autonomous remediation requirements of MA-S2. [38]

Table of contents

Related blogs

Runtime Observability in the Post‑Claude Code Security Era

The adoption of large language models (LLMs) as coding assistants has accelerated rapidly. GitHub’s 2024 developer survey found that 97% of developers have used AI coding tools, with many organizations now relying heavily on these technologies for rapid prototyping, MVP development, and production releases [1]. This increased reliance on AI-generated code introduces non-trivial security risks.

April 10, 2026

8

Turn the Lights On: AI Governance Through Runtime Enforcement

Enterprise AI governance is rapidly evolving from discovery to visibility. Organizations have begun identifying where AI exists and, more recently, illuminating how AI behaves at runtime. Nevertheless, true governance demands more than just visibility, it requires enforcement.

March 26, 2026

5

Turn the Lights On: From AI Discovery to Runtime Illumination

Enterprise AI governance has rapidly converged on discovery mechanisms centered around traffic inspection and external observation. While these approaches provide partial visibility into model usage, they rely on inference rather than direct observation of execution. Recent research (2025 - 2026) demonstrates that critical AI security risks, including prompt injection, agent hijacking and tool-level exploitation, manifest primarily at runtime and are often invisible to boundary-based monitoring. This post argues for a shift from discovery to runtime illumination, a model that treats execution as the primary source of truth for AI governance.

March 20, 2026

5

Stop the waste.
Protect your environment with Kodem.

Get a personalized demo
Get a personalized demo

A Primer on Runtime Intelligence

See how Kodem's cutting-edge sensor technology revolutionizes application monitoring at the kernel level.

5.1k
Applications covered
1.1m
False positives eliminated
4.8k
Triage hours reduced

Platform Overview Video

Watch our short platform overview video to see how Kodem discovers real security risks in your code at runtime.

5.1k
Applications covered
1.1m
False positives eliminated
4.8k
Triage hours reduced

The State of the Application Security Workflow

This report aims to equip readers with actionable insights that can help future-proof their security programs. Kodem, the publisher of this report, purpose built a platform that bridges these gaps by unifying shift-left strategies with runtime monitoring and protection.

Get real-time insights across the full stack…code, containers, OS, and memory

Watch how Kodem’s runtime security platform detects and blocks attacks before they cause damage. No guesswork. Just precise, automated protection.

Combined author
Mahesh Babu
Publish date

0 min read

Compliance

AI Security