AICORTEXMy AICORTEX (1)

Technology behind the platform

The implementation inventory.

Verified running components, available upstream capability, and AICORTEX-native integration are kept distinct so buyers can inspect what exists today.

61 containers15 pods46 workloads0 orphans

Verified inventory / 27 Aug 2026

Inspect each implementation boundary.

Implemented runs in AICORTEX today. Available upstream is documented vendor capability and is not automatically enabled.

29 of 29
AI Workspace

Open WebUI

RUNNING - v0.11.0 in census
Optional

Implemented in AICORTEX

  • Private multi-model chat UI
  • PostgreSQL openwebui schema
  • pgvector-backed RAG
  • SearXNG search integration
  • Kokoro/Gemini speech integrations
  • CC Sessions custom pipe
Available upstream (36)
  • Multi-model chats and mid-conversation model switching
  • File/image uploads
  • Web search with citations
  • Browser/Python or terminal-backed code execution
  • Cross-conversation memory
  • Folders/tags/pins
  • Voice/audio and voice/video workflows
  • Image generation/edit integrations
  • Recurring automations
  • Task management
  • Knowledge bases
  • 13 vector database integrations (core + non-core)
  • Hybrid BM25/vector search with cross-encoder reranking
  • 8 extraction engine options
  • Agentic retrieval
  • Full-context document injection
  • Model presets, dynamic variables, bound tools and per-model access
  • Notes with AI editing and agentic note access
  • Shared Channels with model tagging and access control
  • Open Terminal integration
  • Python tools/functions
  • Pipelines
  • MCP Streamable HTTP
  • OpenAPI tool discovery
  • Skills and versioned prompts
  • RBAC, groups and resource permissions
  • SSO/OIDC/LDAP
  • SCIM 2.0
  • API keys
  • Usage analytics
  • Model arena/A-B/ELO evaluation
  • System banners/default UI settings
  • Webhooks
  • S3/GCS/Azure Blob storage options
  • OpenTelemetry
  • Redis-backed horizontal scaling
Used by
Private AI WorkspaceKnowledge Hub & RAGSecure Coding AgentVoice & Multimodal ExperienceAI Platform-as-a-Service
Upstream documentation
AI Gateway

LiteLLM

RUNNING
Optional

Implemented in AICORTEX

  • Unified model gateway
  • Routing/keys/spend tracking per census
Available upstream (15)
  • OpenAI-style input/output across 100+ models/providers
  • Chat Completions and Responses patterns
  • Embeddings, images, audio, batches and other translated endpoints
  • Consistent output/error formats
  • Streaming
  • Retries/fallbacks
  • Application load balancing
  • Authentication/authorization
  • Multi-tenant spend tracking
  • Budgets per project/user
  • Virtual keys
  • Rate limits
  • Per-project logging, guardrails and caching
  • Admin dashboard
  • Observability callbacks
Used by
Private AI WorkspaceUniversal AI GatewayVoice & Multimodal ExperienceAI Platform-as-a-Service
Upstream documentation
Research/Search

SearXNG

RUNNING
Optional

Implemented in AICORTEX

  • Private metasearch backend for Open WebUI
Available upstream (10)
  • Self-hosted metasearch
  • No user tracking/profiling design
  • Aggregation across hundreds of search services
  • Categories such as general, images, videos, news, maps, music, IT, science, files and social media
  • GET/POST Search API
  • JSON, CSV and RSS output when enabled
  • Search engine/category/language syntax
  • Optional Tor/proxy use
  • Private-instance request privacy protections
  • Custom engine/plugin development
Used by
Private AI WorkspaceResearch Intelligence
Upstream documentation
Voice

Kokoro 82M

RUNNING
Optional

Implemented in AICORTEX

  • Local TTS
Available upstream (6)
  • 82M-parameter open-weight TTS
  • Apache-2.0 weights
  • Local/offline deployment
  • American/British English voices
  • Japanese, Mandarin Chinese, Spanish, French, Hindi, Italian and Brazilian Portuguese voice sets
  • Lightweight inference relative to larger TTS models
Used by
Private AI WorkspaceVoice & Multimodal Experience
Upstream documentation
Voice

Gemini TTS

RUNNING through proxy
Optional

Implemented in AICORTEX

  • Cloud speech synthesis adapter
Available upstream (6)
  • Gemini 3.1 Flash TTS Preview, 2.5 Flash TTS and 2.5 Pro TTS options
  • Single-speaker and multi-speaker synthesis
  • Prompt-directed audio profile, scene and director notes
  • Style/accent/pace control via prompting
  • Premium studio-quality Pro TTS positioning
  • Batch API support for Gemini 2.5 Pro TTS
Used by
Private AI WorkspaceVoice & Multimodal Experience
Upstream documentation
Data

PostgreSQL

RUNNING - three separate DB instances in census
Required

Implemented in AICORTEX

  • Enterprise DB schemas sessions, recall, openwebui, trading, gateway
  • Dedicated LiteLLM DB
  • Dedicated Authentik DB
Available upstream (6)
  • Relational ACID database capabilities
  • Rich SQL and indexing
  • Full-text search with ranking
  • JSON/JSONB
  • Transactions, constraints, views and extensions
  • Replication/backup ecosystem
Used by
AICORTEX Core PlatformUniversal AI GatewayKnowledge Hub & RAGPersistent AI MemoryZero-Trust Edge & IdentityAI Platform-as-a-Service
Upstream documentation
Knowledge/Data

pgvector

RUNNING in enterprise DB image
Optional

Implemented in AICORTEX

  • Vector storage and recall/OpenWebUI usage
Available upstream (7)
  • Exact nearest-neighbor search
  • HNSW approximate indexes
  • IVFFlat approximate indexes
  • L2, inner product, cosine and other supported distances
  • Metadata filtering with SQL
  • Hybrid full-text + vector search patterns
  • Vector columns in ordinary PostgreSQL tables
Used by
Private AI WorkspaceKnowledge Hub & RAGPersistent AI MemoryResearch Intelligence
Upstream documentation
Knowledge/Inference

EmbeddingGemma + llama.cpp

RUNNING - low-utilization local embedder
Optional

Implemented in AICORTEX

  • Local embedding endpoint
Available upstream (4)
  • Local embedding generation
  • OpenAI-compatible server patterns through llama.cpp server
  • CPU/GPU local inference depending build/hardware
  • Model swap capability while retaining downstream vector store
Used by
Knowledge Hub & RAGPersistent AI Memory
Upstream documentation
Identity

Authentik

RUNNING - server + worker + DB + Redis
Optional

Implemented in AICORTEX

  • OIDC/SSO layer
  • Forward-auth endpoint for Traefik
  • Dedicated worker and PostgreSQL
  • Redis broker/cache
Available upstream (14)
  • OAuth2/OIDC provider and source roles
  • Authorization code, client credentials, implicit, hybrid, device code, refresh token and token exchange
  • PKCE and scope mappings
  • SAML2 provider with metadata import/export and signed requests
  • LDAP/LDAPS provider and directory
  • Code-based MFA support for LDAP-compatible authenticators
  • Proxy provider
  • Forward auth - single application
  • Forward auth - domain level
  • Header authentication patterns
  • SCIM 2.0 provider to provision users/groups into applications
  • SCIM source to provision users/groups into Authentik
  • Policies and flows
  • Blueprints for identity configuration as code
Used by
Private AI WorkspaceZero-Trust Edge & IdentityAI Platform-as-a-ServiceReal-Time & Mobile Experiences
Upstream documentation
Data/Identity Support

Redis

RUNNING - Authentik cache/task broker
Optional

Implemented in AICORTEX

  • Authentik-only cache/task broker
Available upstream (10)
  • In-memory data structures
  • Strings, hashes, lists, sets, sorted sets, JSON and more
  • Caching
  • Pub/Sub fan-out
  • Streams with consumer groups/replay/retention
  • RDB snapshots
  • AOF persistence
  • RDB + AOF combinations
  • Replication options
  • Vector/search and AI-memory use cases in modern Redis
Used by
Zero-Trust Edge & Identity
Upstream documentation
Edge

Traefik

RUNNING - only ingress path
Required

Implemented in AICORTEX

  • Central hostname routing
  • Pod IP routing
  • Authentik/CrowdSec middleware integration
Available upstream (10)
  • HTTP/TCP/UDP routing
  • Reverse proxy
  • TLS termination
  • Middleware chains
  • Load balancing
  • Weighted services
  • Sticky sessions
  • HTTP/gRPC health checks
  • Forward-auth integration
  • Dynamic configuration/providers
Used by
AICORTEX Core PlatformZero-Trust Edge & IdentityThreat Detection & Runtime SecurityAI Platform-as-a-ServiceReal-Time & Mobile Experiences
Upstream documentation
Edge

Cloudflare Tunnel

RUNNING
Required

Implemented in AICORTEX

  • Outbound tunnel; no inbound internet ports open per census
Available upstream (10)
  • Secure connection without publicly routable origin IP
  • Outbound-only cloudflared connections
  • Block-all-ingress/allow-egress firewall pattern
  • HTTP, SSH, RDP and other protocol access patterns
  • Four outbound connections per connector for availability
  • Multiple cloudflared replicas
  • Automatic failover
  • TLS 1.3 and post-quantum tunnel encryption documented in 2026 configuration guide
  • Tunnel tokens/permissions
  • Health metrics/alerts
Used by
AICORTEX Core PlatformZero-Trust Edge & IdentityAI Platform-as-a-Service
Upstream documentation
Security

CrowdSec

RUNNING
Optional

Implemented in AICORTEX

  • Behavioral intrusion prevention
  • Traefik bouncer
  • Threat-map decision feed
Available upstream (9)
  • Behavior-based Security Engine
  • Decision/remediation model
  • Layer 3/4 infrastructure remediation
  • Layer 7 application remediation
  • Traefik/Nginx/HAProxy/OpenResty bouncers
  • WAF-capable AppSec remediation patterns
  • Virtual patching
  • Defense against application-layer attacks when AppSec is configured
  • Community scenarios/blocklists ecosystem
Used by
Threat Detection & Runtime Security
Upstream documentation
Security

Trivy

SCHEDULED - not resident
Optional

Implemented in AICORTEX

  • Scheduled vulnerability-scanning pattern per census
Available upstream (9)
  • Container image scanning
  • Filesystem and repository scanning
  • Vulnerability detection
  • Secret scanning
  • IaC/misconfiguration scanning for Docker/Kubernetes/Terraform/CloudFormation and more
  • License scanning
  • Custom secret rules
  • SBOM scanning including CycloneDX and SPDX inputs
  • Severity filtering and policy automation
Used by
Threat Detection & Runtime Security
Upstream documentation
Security

Tetragon

DISABLED BY CHOICE
Currently disabled

Implemented in AICORTEX

  • Planned observe/enforce redesign
Available upstream (10)
  • eBPF runtime observability
  • Process, file, network and security monitoring
  • TracingPolicy framework
  • kprobes, tracepoints, uprobes, LSM hooks and USDT hooks
  • In-kernel filtering
  • Dynamic policy load/unload
  • Actions on matches
  • Signal-based enforcement
  • Return-value override enforcement where supported
  • Non-Kubernetes policy use via CLI/daemon flags
Used by
Threat Detection & Runtime Security
Upstream documentation
Observability

VictoriaMetrics

RUNNING - v1.115.0 in census
Optional

Implemented in AICORTEX

  • Primary metrics database
  • 20-year retention configured
Available upstream (6)
  • Prometheus remote-write compatibility
  • Prometheus-style query ecosystem
  • Efficient time-series storage
  • Single-node and cluster deployment options
  • Native integrations with monitoring ecosystem
  • Long retention configuration
Used by
AI Operations CenterExecutive Intelligence & Reporting
Upstream documentation
Observability

vmalert

RUNNING
Optional

Implemented in AICORTEX

  • 25 alert rules
  • Telegram delivery path restored
Available upstream (6)
  • Prometheus-compatible alerting rules
  • Recording rules
  • VictoriaMetrics/Prometheus/Graphite and other datasource integrations
  • Alertmanager notifier integration
  • Alert state persistence across restarts
  • Remote-write of recording-rule results
Used by
AI Operations Center
Upstream documentation
Observability

Grafana

RUNNING - 13.1.0 in census
Optional

Implemented in AICORTEX

  • Dashboards
  • Grafana alerting
Available upstream (10)
  • Dashboards and panels
  • Explore
  • Built-in Prometheus/Loki/Tempo/PostgreSQL and other data sources
  • Data-source plugins
  • Grafana-managed alerting across metrics/logs and multiple data sources
  • PostgreSQL alerting
  • Loki alerting
  • Trace/log/metric correlation through Tempo data source
  • Service Graph visualization
  • User/team/role features depending edition
Used by
AI Operations CenterExecutive Intelligence & Reporting
Upstream documentation
Observability

Loki

RUNNING - 14-day corpus policy
Optional

Implemented in AICORTEX

  • Central log store on custom port 3200
Available upstream (7)
  • Horizontally scalable multi-tenant logging
  • Label-only indexing with compressed chunks
  • Filesystem or object-store backing
  • LogQL queries
  • Metrics derived from logs
  • Grafana integration
  • Grafana-managed and data-source-managed alerting
Used by
AI Operations Center
Upstream documentation
Observability

Grafana Alloy

RUNNING - v1.12.2 in census
Optional

Implemented in AICORTEX

  • Container/file log shipping including Traefik access logs
Available upstream (9)
  • Unified collection of metrics, logs, traces and profiles
  • OpenTelemetry Collector distribution
  • Prometheus pipelines
  • Loki/Pyroscope integrations
  • OTLP receivers/exporters
  • Batching and processing pipelines
  • Service discovery
  • Component-based programmable pipelines
  • Edge, gateway and hybrid deployment patterns
Used by
AI Operations Center
Upstream documentation
Observability

Tempo

RUNNING - 2.9.0 in census
Optional

Implemented in AICORTEX

  • OTLP trace backend
Available upstream (8)
  • Distributed tracing backend
  • OTLP, Jaeger and Zipkin ingestion
  • TraceQL search
  • Trace-to-logs
  • Trace-to-metrics
  • Service Graphs and RED metrics via Grafana
  • Metrics generation from spans
  • Multi-tenancy and scalable deployment patterns
Used by
AI Operations Center
Upstream documentation
Knowledge Graph

Neo4j

RUNNING - 5-community in census
Optional

Implemented in AICORTEX

  • Service-map graph
  • ~280k session-related nodes per census
Available upstream (8)
  • Property graph model
  • Cypher query language
  • Range/text/point/token indexes
  • Full-text semantic indexes
  • Vector indexes
  • Vector search on nodes/relationships
  • Hybrid full-text + vector patterns
  • Drivers/APIs and GenAI ecosystem integrations
Used by
Knowledge Graph & Institutional IntelligenceSecure Coding AgentResearch IntelligenceExecutive Intelligence & Reporting
Upstream documentation
Agent Integration

Model Context Protocol (MCP)

MULTIPLE RUNNING SERVERS
Optional

Implemented in AICORTEX

  • Neo4j MCP
  • Session recall MCP
  • Scrapy MCP
  • Google Docs MCP (OAuth dead)
  • Grafana reports MCP
Available upstream (10)
  • JSON-RPC base protocol
  • Lifecycle/capability negotiation
  • Tools
  • Resources
  • Prompts
  • stdio transport
  • Streamable HTTP transport
  • Authorization framework for HTTP
  • Resource subscriptions/list-changed notifications
  • Sampling, elicitation, tasks, logging and completions in current SDK ecosystem
Used by
Persistent AI MemoryKnowledge Graph & Institutional IntelligenceMCP Agent HubSecure Coding AgentResearch Intelligence
Upstream documentation
Research/Crawling

Scrapy

RUNNING through scrapy-mcp
Optional

Implemented in AICORTEX

  • MCP-triggered crawler
  • Container-per-crawl isolation
  • Pinned MCP SDK compatibility fix
Available upstream (12)
  • Concurrent/fault-tolerant crawling
  • Spiders and generic spider classes
  • CSS/XPath/regex extraction
  • Scheduler
  • Downloader and spider middleware
  • Cookies/auth/cache/compression/robots/depth features
  • AutoThrottle and concurrency controls
  • Item pipelines for cleaning/validation/persistence
  • JSON/JSONL/CSV/XML feed exports
  • FTP/S3/local and extensible storage backends
  • Signals/extensions and custom components
  • Media download pipelines
Used by
Research Intelligence
Upstream documentation
Container Platform

Podman

RUNNING - 61 containers/15 pods in census
Required

Implemented in AICORTEX

  • Pods with infra containers owning shared network namespace/ports
  • Standalone edge services
  • Restart policies and boot units added for restored services
Available upstream (11)
  • Daemonless OCI containers
  • Pods with shared ipc/net/uts and optional cgroup/pid namespaces
  • Pod-level restart policies
  • Container health checks
  • Startup health checks
  • Health-on-failure actions
  • systemd/Quadlet integration
  • Auto-update policies for systemd-managed workloads
  • Dry-run update checks
  • Rootless and rootful modes
  • Kubernetes YAML interoperability patterns
Used by
AICORTEX Core Platform
Upstream documentation
Automation

Ansible

RUNNING in deployment container
Required

Implemented in AICORTEX

  • all.yml + roles as deployment source of truth
  • Loopback inventory deployment
Available upstream (8)
  • Human-readable playbooks
  • Configuration management
  • Deployment and orchestration
  • Multi-tier sequencing/rolling updates
  • Roles and reusable task organization
  • Inventories and variables
  • Idempotent module patterns
  • Podman collection for container lifecycle
Used by
AICORTEX Core Platform
Upstream documentation
Web Serving

nginx

RUNNING in frontend and static sites
Required

Implemented in AICORTEX

  • Production frontend server
  • Multiple static microsites
Available upstream (11)
  • Static/index file serving
  • Reverse proxy
  • Caching
  • HTTP load balancing/fault tolerance
  • FastCGI/uwsgi/SCGI proxying
  • TLS/SNI
  • HTTP/2
  • HTTP/3
  • Virtual hosts
  • Access logging
  • TCP/UDP proxying
Used by
AICORTEX Core PlatformReal-Time & Mobile Experiences
Upstream documentation
Mobile

Expo / React Native

expo-dev DOWN BY CHOICE
Optional

Implemented in AICORTEX

  • Tailnet-only Metro/Expo dev environment
Available upstream (10)
  • Universal Android/iOS/web app development from one codebase
  • Expo Router/file-based routing
  • Native SDK modules
  • EAS Build for iOS/Android
  • EAS Submit
  • EAS Update
  • EAS Hosting
  • EAS Workflows/CI-CD
  • Internal distribution
  • Cloud or local/self-hosted build options
Used by
Real-Time & Mobile Experiences
Upstream documentation

AICORTEX-native value

The integration layer that makes a platform.

Workflow IP and reference services connect open components into one governed, observable deployment rather than a stack of unrelated installs.

RESTORED/RUNNING

telegram-coder

Coder-mode Telegram gateway with approve/deny gates on writes and shell commands.

Secure Coding Agent
RESTORED/RUNNING

owui-coder

OpenWebUI CC Sessions integration for resuming Claude Code sessions from chat.

Secure Coding Agent
RUNNING

ops-dashboard

Operations UI for infrastructure status and live Claude session ingest.

AI Operations Center
RUNNING/HARDENING GAP

worldview-dev

3D globe/geospatial application; currently Vite dev server standing in for production nginx and should be hardened before blueprint packaging.

Real-Time & Mobile Experiences
OAUTH DEAD

google-docs-mcp

Google Docs/Drive MCP connector; process is running but OAuth is currently nonfunctional.

MCP Agent Hub
Build with these services