SPNT

Platform Architecture

Ten modules. One graph. One substrate.

Detection, hardening, offense, governance, investigation, identity oversight, data classification, AI-security, agentic/MCP control, and sovereign inference share one canonical data model. Agents act over the graph; the brain (Claude in the cloud, Jadro on-prem) reasons on it. This is not integration theater. This is architecture.

The Problem

Why dashboards are not platforms

Most security platforms are just aggregation layers with API connectors. That is integration theater.

Connector fatigue

Every tool needs configuration. Every connector breaks. Every upgrade requires re-mapping. You spend more time maintaining integrations than using them.

Context collapse

The same vulnerability appears differently in three tools. Your team triages it three times. Remediation happens without knowing validation status.

Data silos persist

Aggregating data in a dashboard does not make it unified. Context is still lost. Evidence is still duplicated. Compliance still requires manual assembly.

The result: manual correlation, delayed validation, duplicated evidence, and security teams who spend more time managing tools than managing risk.

The Solution

A canonical data model

Serpentine does not aggregate external data. It generates its own findings from its own scanners, all normalized into one schema.

Assets

Written byPostava
Used byAll modules

Single inventory across all security operations

Findings

Written byOdbrana
Used byNapad, Regulativa, Istraga

Normalized vulnerabilities regardless of source

Validations

Written byNapad
Used byPostava, Regulativa, Istraga

Exploit confirmation drives real prioritization

Remediations

Written byPostava
Used byRegulativa

Hardening actions become compliance evidence

Evidence

Written byAll modules
Used byRegulativa

Every action produces audit-ready proof

Controls

Written byRegulativa
Used byAll modules

Framework requirements flow to operations

Identity

Written byNadzor
Used byAll modules

Human and non-human identity surfaces privilege and access risks

Classification

Written byPodatoci
Used byOdbrana, Istraga, Regulativa

Data sensitivity drives finding prioritization

AI-Incidents

Written byProtivAI
Used byIstraga, Regulativa

AI/LLM risks on the same register as infrastructure

Agents

Written byZastapnik
Used byNadzor, Istraga, Regulativa

AI agents and MCP tools governed as first-class identities

Research

Written byIstraga
Used byAll modules

Autonomous reasoning grounds prioritization in real threat context

Inference

Written byJadro
Used byIstraga, Napad, Zastapnik

Sovereign brain the agentic modules reason on — cloud or air-gapped

Every entity is first-class. Every relationship is explicit. Every module reads and writes the same state.

The Loop

How data flows through ten modules

Security becomes continuous when data flows naturally between disciplines — agents act on the graph, the sovereign brain reasons on it.

Odbrana

detects

Postava

hardens

Napad

validates

Istraga

investigates

Regulativa

governs

Nadzor

oversees

Podatoci

classifies

ProtivAI

AI-secures

Zastapnik

controls agents

Jadro

reasons

All modules read and write toone substrate

Intelligence feeds

OSINT

Feeds from public breach, leak, and threat intelligence sources

Threat Intel

IOC and TTP feeds mapped to your asset inventory

Exploit DB

Known exploit availability integrated into validation

Compliance KB

35 frameworks, 3,144 obligations, continuous mapping

Adversary Models

Threat actor profiles for attack path simulation

Agent & MCP Telemetry

Live signals from AI agents and MCP tools acting in your environment

Research Corpus

Academic and industry security research for reasoning

This is not a one-time assessment. The loop runs continuously, with every action updating the shared security graph.

What This Enables

What unified architecture unlocks

Capabilities that fragmented tools cannot provide.

Evidence reuse

A vulnerability scan satisfies SOC 2 CC7.1, ISO A.12.6.1, and GDPR Art. 32 simultaneously. Upload or generate evidence once, map it everywhere.

Example: One Odbrana finding maps to 4 frameworks without manual re-entry

Validated prioritization

Do not prioritize by CVSS alone. Prioritize by actual exploitability. Napad validation feeds directly into remediation queues.

Example: Critical finding downgraded after Napad confirms no exploit path

Continuous operating state

Security is not point-in-time. The platform maintains live state across assets, findings, validation, remediation, and evidence.

Example: Auditor queries current posture, not a 90-day-old report

Example

One vulnerability, the full platform

See how a single finding moves through the entire platform.

Example Finding

Exposed Administrative Endpoint

CRITICAL
Discoveredby Odbrana

Web scanner detects /admin/api endpoint responding without authentication

Finding created: ODB-2024-0142, severity CRITICAL, CVSS 9.8
Validatedby Napad

Offensive agent confirms endpoint returns user PII without credentials

Validation record: EXPLOITABLE, reproduction steps captured
Remediatedby Postava

Hardening policy applied: endpoint moved behind auth, rate limiting added

Remediation record: APPLIED, before/after evidence captured
Mappedby Regulativa

Finding and remediation linked to ISO A.9.4.1, SOC 2 CC6.1, GDPR Art. 32

Evidence package: audit-ready, control status SATISFIED

Total time from discovery to audit-ready evidence: automated, continuous, traceable.