openova/CLAUDE.md
e3mrah 9dd2ae7a76 docs: update CLAUDE.md — clarify public repo boundary
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-04 06:34:09 +01:00

7.6 KiB

OpenOva Platform (Public Repo)

This is the public, open-source repo. No proprietary content (website, deployment configs, infra secrets). Website and deployment live in openova-private.

Project Memory

IMPORTANT: Read .claude/project-memory.md for full strategic context about OpenOva positioning, architecture decisions, and product strategy.

Purpose

OpenOva is an AI-native infrastructure platform. 52 open-source components on Kubernetes — every one designed to be AI-manageable. Cloud-native is the foundation. AI-native is the differentiator.

  • AI-powered operations built in — Specter has pre-built semantic knowledge of every CRD, integration dependency, and failure mode across all 52 components
  • Converged blueprint ecosystem with operational guarantees — turnkey, production-grade, instant
  • Comprehensive migration — full legacy assessment, AI modernization roadmap, and structured migration (Exodus)
  • Both consultancy AND productized platform from day 1

Product Family (Locked 2026-02-26)

Product Name Description
Core Platform OpenOva 52 component AI-native K8s ecosystem — every component AI-manageable by Specter
Bootstrap+Lifecycle+IDP OpenOva Catalyst Bootstrap wizard, Day-2 manager, IDP, Workflow Explorer
AI Hub OpenOva Cortex LLM serving, RAG, AI safety, LLM observability
SaaS LLM Gateway OpenOva Axon Hosted inference gateway (neural link to Cortex)
Open Banking OpenOva Fingate PSD2/FAPI fintech sandbox
AIOps SOC/NOC OpenOva Specter AI brain with pre-built semantic knowledge of all 52 components. Token-efficient AI-powered SOAR and self-healing.
Data & Integration OpenOva Fabric Event-driven integration + data lakehouse
Communication OpenOva Relay Email, video, chat, WebRTC
Migration OpenOva Exodus Comprehensive legacy assessment + AI modernization roadmap + structured migration

Business Model

  • Blueprints are FREE and open source (always)
  • Revenue: per-vCPU-core platform support subscription (all software is free, only charge for support)
  • Pricing: per-vCPU-core under management (ELA with true-up or PAYG)
  • Target market: banks first (2 prospects), then regulated verticals, then broader

Monorepo Structure

openova/
├── core/                    # Bootstrap + Lifecycle Manager application
├── platform/                # All 52 component blueprints (flat structure)
├── products/                # Bundled vertical solutions
│   ├── cortex/              # OpenOva Cortex - Enterprise AI Hub
│   ├── fingate/             # OpenOva Fingate - Open Banking (+ 6 services)
│   ├── fabric/              # OpenOva Fabric - Data & Integration
│   ├── relay/               # OpenOva Relay - Communication
│   └── axon/                # OpenOva Axon - SaaS LLM Gateway
└── docs/                    # Platform documentation

Core Application

The core/ directory contains a single Go application with two deployment modes:

Mode Location Purpose IaC Tool
Bootstrap Outside cluster Initial provisioning OpenTofu
Manager Inside cluster Day-2 operations Crossplane

See core/README.md for detailed architecture.

Platform Components (52)

All components are flat under platform/:

anthropic-adapter, bge, cert-manager, cilium, clickhouse, cnpg, coraza, crossplane, debezium, external-dns, external-secrets, failover-controller, falco, ferretdb, flink, flux, gitea, grafana, harbor, iceberg, k8gb, keda, keycloak, knative, kserve, kyverno, langfuse, librechat, litmus, livekit, llm-gateway, matrix, milvus, minio, nemo-guardrails, neo4j, openbao, openmeter, opensearch, opentofu, reloader, sigstore, stalwart, strimzi, stunner, syft-grype, temporal, trivy, valkey, velero, vllm, vpa

Products

Products bundle platform components with custom services for specific verticals:

  • cortex (OpenOva Cortex - AI Hub): Uses kserve, knative, vllm, milvus, neo4j, librechat, bge, llm-gateway, anthropic-adapter, nemo-guardrails, langfuse
  • fingate (OpenOva Fingate - Open Banking): Uses keycloak, openmeter + 6 custom services (accounts-api, consents-api, ext-authz, payments-api, sandbox-data, tpp-management)
  • fabric (OpenOva Fabric - Data & Integration): Uses strimzi, flink, temporal, debezium, iceberg, clickhouse, minio
  • relay (OpenOva Relay - Communication): Uses stalwart, livekit, stunner, matrix
  • axon (OpenOva Axon - SaaS LLM Gateway): SaaS service, references Cortex infrastructure

Key Principles

  • Bootstrap wizard EXITS after provisioning (must be safe to delete)
  • Lifecycle Manager continues inside cluster for day-2 operations
  • Catalyst IDP is for developers; Lifecycle Manager is for platform operators
  • OpenOva stays in picture via blueprints, not runtime components
  • Zero external dependencies for core (no CNPG, Valkey, Strimzi for itself)
  • Every component is AI-manageable — structured CRDs, unified OTel telemetry, standardized health endpoints, declarative GitOps
  • Specter is built-in, not bolted on — pre-built semantic knowledge of the ecosystem is an architectural advantage, not a feature add-on

AI-Native Architecture

What makes OpenOva AI-native (not AI-cosmetic):

Property What It Means Why It Matters for AI
Structured CRDs Every component exposes typed Kubernetes Custom Resource Definitions Specter reads CRDs, not config files — machine-parseable by design
Unified OTel telemetry All 52 components emit metrics, logs, traces via OpenTelemetry Specter correlates signals across the entire stack in one query
Standardized health Consistent liveness, readiness, startup probes Specter knows the health vocabulary of every component
Kyverno policy-as-code Security and operational policies are declarative and machine-readable Specter reasons about compliance programmatically
Declarative GitOps All state in Git via Flux Specter diffs actual vs desired, detects drift, proposes reconciliation
Pre-built semantic models Specter has knowledge of every CRD schema, integration graph, failure mode, upgrade path, compliance mapping Surgical context = 10x fewer tokens, 10x faster, 10x more accurate than dumping raw logs

Token efficiency is the economic moat. Competitors bolt AI onto unstructured platforms and dump massive context into LLM prompts. Specter sends surgical, structured context. This is an architectural advantage that cannot be retrofitted.

Documentation

Conventions

  • All manifests are Kustomize-based
  • Secrets via External-Secrets Operator (never commit plaintext)
  • Git commits: conventional commits (feat:, fix:, docs:, infra:)

Customer Sync

This monorepo syncs to customer's multi-repo Gitea:

GitHub (monorepo)                    Customer Gitea (multi-repo)
─────────────────                    ──────────────────────────
openova/core/              ──sync──> openova-core/
openova/platform/cilium/   ──sync──> openova-cilium/
openova/platform/flux/     ──sync──> openova-flux/