Nucleus Framework is a production-tested platform for Java 21 / Spring Boot microservices. One reference architecture gives every service built-in orchestration, observability, security, messaging and integration. Below is each module and what it does — follow any card into the module reference for the full detail.
Build multi-step processes in a drag-and-drop designer as a JSON DAG — 14 step kinds including branches, loops, dynamic scatter/gather, sub-workflows, human approvals, GraalVM scripting and service calls. Durable waits survive restarts; simulate a flow before it touches anything real.
Workflow details →The deployable engine host: a durable run store and a control plane to start, suspend, resume, replay and cancel runs. Triggers from REST, a Kafka topic, a cron schedule or a file arrival; idempotent steps, saga compensation and SLA timers built in.
Reference →Add the dependency, enable the monitoring profile, and a service self-registers with live health, CPU, memory, disk, threads and GC — streamed over Kafka into a fleet registry with history charts, red status when an instance is in trouble, and SSE-pushed alerts.
Observability details →Schedule a diagnostic capture window ahead of time (tonight's batch, next release) or record on the spot. Logs, GC pauses, thread dumps and CPU/heap samples replay on one synchronized timeline, then archive to object storage.
How replay works →Every HTTP request carries a session ID from the browser through every backend service into each log line, so support can trace one user's whole session across microservices from a single ID.
Reference →An OAuth2 / OIDC authorization server with JWT and rotating keys, client-credentials for service-to-service calls, and a password-reset pipeline with anti-enumeration and anti-replay.
Reference →A reactive WebFlux edge gateway that routes from the live service registry, with an off-heap blacklist (revoked tokens, blocked users, blocked IPs) and marker-token verification so backends reject anything that bypassed the gateway.
Reference →Add @AuditAction to any method and Nucleus records who, what, when and with what data — one audit format across every service, searchable, exportable and retention-aware.
Reference →Identity, roles, groups and route/authority rules in the database, managed from the operations console — grant or revoke access at runtime without a redeploy.
Reference →Per-client, per-field AES/GCM encryption for sensitive values (descriptions, memos) — encrypted at rest, decrypted in-memory only when processing needs it.
Reference →A logging framework that keeps personal data out of log output by construction, so diagnostics never leak what they shouldn't.
Reference →One @NucleusListener / @NucleusPublish annotation works across 13 messaging brokers — Kafka, RabbitMQ, SQS/SNS/Kinesis, Google Pub/Sub, Azure Service Bus and JMS (ActiveMQ, Artemis, Solace, IBM MQ). Switch brokers with a config change, not a refactor.
Connectors details →A typed, Vault-backed catalog of 25 connection types — databases, REST/SOAP endpoints, SSH hosts, object stores, file/FTP servers, raw sockets, SNMP and more. Register once, run cataloged operations, with the native client as an escape hatch.
Connectors details →A provider-agnostic LLM client: build an AiRequest, get a strongly-typed AiResult. Anthropic and OpenAI built in (multimodal, JSON responses), with classified errors and retries — add another provider with three beans.
Reference →A MinIO / S3-compatible object-storage abstraction — upload, download and delete on standardized paths, with a pluggable backend you can swap from MinIO to AWS S3.
Reference →The server-sent-events channel that pushes live updates — health, alerts, run progress — from the platform to the console in real time.
Reference →A Kafka-backed mail pipeline for transactional email — templated, asynchronous, and decoupled from the services that trigger it.
Reference →The reference architecture foundation — JWT security plumbing, MDC diagnostics, the in-process step model, low-latency primitives (off-heap caches, bounded executors, AFTER_COMMIT side effects) and the conventions every module builds on.
Reference →A Spring Cloud Config server backed by MySQL, so configuration is materialized, versioned and queryable instead of scattered across files.
Reference →The gateway for runtime properties — including a service's own per-instance settings — written through one audited endpoint, never by reaching into another service's tables.
Reference →A managed store and browser for shared lookup data (buckets, codes, enumerations) used consistently across services.
Reference →A reusable address model and components, so every product handles postal addresses the same way.
Reference →The @nucleus/* Angular libraries the whole console is built from — page shells, list and edit pages, design tokens, the workflow designer and more.
Reference →You may see Nucleus described as working with Kafka "across 13 brokers." To be precise: one messaging API — @NucleusListener and @NucleusPublish — runs unchanged across 13 different messaging systems. Kafka is the reference adapter; the same code path also targets RabbitMQ, Amazon SQS/SNS/Kinesis, Google Pub/Sub, Azure Service Bus, and JMS brokers (ActiveMQ, Artemis, Solace, IBM MQ). Your listener never knows which one is on the other end of the wire.
About the name. This is Nucleus Framework at nucleusframework.ca — a Java/Spring microservices platform. It is unrelated to other products that share the word "Nucleus." When you search, use Nucleus Framework with the .ca domain.