MetaCore Security
Platform access and AI action boundaries
Identity, sessions, Bridge permissions, agent authority, audit and recovery principles.
Explore platform controls →MetaCore Platform · Security & Operational Resilience
Security starts with boundaries, not a promise to watch everything.
MetaCore Security is the security and resilience layer of the Persistent AI Operating Platform: identity, sessions, scope, Bridge entitlements, AI roles, audit, rollback and incident operating principles. Baseline cyber security remains a separate foundation.
IDENTITY→SCOPE→ACTION→AUDIT→RECOVERY→WRITEBACK
SECURITY PATHWAYS
One security landscape. Three clear paths.
Choose the path that fits your task: platform access and AI boundaries, practical protection for your organisation, or a protocol for a specific incident. These paths complement one another but are not the same service.
Kiberapsauga.lt
Practical protection for organisations
Risk assessment, technical protection, audits, training and a clear response plan tailored to the organisation.
Open Kiberapsauga.lt →LIVE incident protocols
From a signal to a verifiable case
Record the situation, add logs and evidence, receive AI analysis and prepare material for a human decision.
Open LIVE centre →AI helps organise and analyse. Authorised people approve critical actions; service scope depends on granted access and agreed processes.
WHAT WE PROTECT
Platform security starts with who is allowed to act.
MetaCore Security first protects the platform control plane: identity, sessions, entitlements, context, AI/tool boundaries and recoverable working state.
Identity & Sessions
Login signals, active sessions, new-IP / multi-IP risk signals, rate limits and brute-force protection.
Bridge & Entitlements
Bridge access and user entitlements are separate from the AI model; scope can be active or revoked.
Scoped Authority
A significant action should have an authenticated role, delegated mandate, explicit scope and a verifiable action trail.
Workspace & Recovery
Backup, restore, rollback, log and service hygiene are applied as deployment and incident-recovery practices.
AI / Tool Boundaries
Agents and AI roles receive tool permissions, approval gates and rollback discipline for a specific workflow.
Customer Endpoints
Server / FTP / MCP / equipment registry support exists, but full customer monitoring/action claims await real endpoint, audit and rollback evidence.
BASELINE CYBER + METACORE
NIS2, SOC and administration are not replaced by AI.
MetaCore Security is an additional control and evidence layer above baseline cyber security. Endpoint protection, firewalls, patching, admin discipline, legal duties and formal compliance remain separate responsibilities.
Area
Baseline cyber
MetaCore Security
Access
MFA, passwords, roles, admin hardening.
Session/auth signals, Bridge scope and AI/tool authority controls.
Incidents
SOC/admin process, containment and recovery.
Evidence-first incident workflow, rollback and writeback into organizational memory.
AI
Data classification, acceptable use and accountability.
Persona/agent/tool scope, approval, audit and model-neutral governance.
Compliance
Legal and organizational program.
Readiness evidence and control traces; not a substitute for certification.
Claim boundary: “NIS2-aligned direction / readiness” is not NIS2 certification or automatic compliance.
QUANTARA SECURITY PERSONA
Quantara is a security role, not a proprietary foundation model.
Quantara can be configured as a MetaCore Security persona: analyze authorized context, assist with triage, propose hardening steps, prepare rollback plans and coordinate agents or tools under delegated mandate and scope.
Diagnose
Collects available facts and separates evidence from hypothesis. It does not receive magical access to systems.
Recommend / Coordinate
Can propose the smallest safe action or coordinate agents when the relevant tools and scope are explicitly authorized.
Verify / Document
After an action, verifies outcome, rollback availability and continuity evidence.
No universal 24/7 claim: actual monitoring depends on connected signals, runtime availability, scope and the specific deployment contract.
AI PERSONA & AGENT SECURITY
Hierarchical orchestration requires hierarchical boundaries.
A Persona is a higher-level work role with context and policy. An agent is a task executor. A tool is an action channel. Each layer receives only the authority needed for its task.
PERSONA→AGENT / GROUP→TOOL→SYSTEM
Least Authority
Role, agent and tool receive minimum scope; broader authority is not the default.
Policy-defined Approval
Risky or irreversible actions follow approval / review gates defined by canonical policy; where human approval is required, it is provided by an authorized role.
Audit & Rollback
Initiator, executor, target, result and rollback should be traceable to the degree supported by the workflow evidence.
CAPABILITY STATES
What we can say today.
Auth Security Controls
Session lifecycle, auth alerts, rate-limit / brute-force signals, user security controls and Bridge entitlement states operate on the platform.
Operational Audit Primitive
A root-operation audit schema and logging path exist; coverage is expected to expand through a unified event ledger.
Hardening & Recovery
Server, web, auth and workflow hardening is performed as controlled deployment/incident work with backup and rollback discipline.
Kiberapsauga.lt Lane
Baseline cyber: endpoint protection, audits, training, ISO readiness and incident help — separate from the MetaCore AI control layer.
Endpoint Security Gateway
The registry schema is ready, but the canonical tenant registry currently has no customer endpoint evidence. Promotion gate: endpoint → scoped credential → action audit → rollback.
Unified Security Event Ledger
Connect auth, tool, agent, incident and recovery events into one canonical evidence chain.
INCIDENT OPERATING LOOP
An incident should become a verifiable sequence.
1 · DetectSignal and source.
2 · AssessScope, impact, confidence.
3 · ContainSmallest isolating action.
4 · RecoverRestore a safe state.
5 · VerifySmoke / evidence / rollback check.
6 · WritebackCause, lesson, prevention.
AI does not judge or punish. A security role collects facts, helps contain risk and executes only authorized actions; authority remains resolved through canonical policy, authenticated roles, delegated mandate and scope; AI does not self-expand it.
SECURITY REVIEW
Define the boundaries first. Then decide what should be connected to AI.
Start with identity, access, sessions, Bridge and agent/tool scope. Customer endpoints are connected only with an authenticated role, explicit mandate / scope, audit and rollback.