FlowMaster is a microservices-based business process automation platform for designing, executing, and monitoring workflows with AI integration. As of Feb 8 2026, 29 services are deployed and running on a K3S Kubernetes cluster (dev-01 server 65.21.153.235), covering all 79 requirements (R01-R79). Code is deployed but ALL services are UNTESTED — no integration, e2e, or health check testing has been performed.

Source Control: GitLab flow-master group (primary), mirrored to GitHub HCB-Consulting-ME. Code locations: ~/projects/documentation (stale copies), ~/projects/flowmaster/ (working dir, not a git repo).

Deployed Services (29 Running Pods — K3S)

Core Business Logic (13 services)

Service	Port	Tech	Status
Process Design	9003	Python/FastAPI	Running, UNTESTED
Execution Engine	9005	Python/FastAPI	Running, UNTESTED
Human Task	9006	Python/FastAPI	Running, UNTESTED
AI Agent Orchestration	9006	Python/FastAPI	Running, UNTESTED
Document Intelligence	9002	Python/FastAPI	Running, UNTESTED
Authentication	8002	Python/FastAPI	Running, UNTESTED
API Gateway	9000	Python/FastAPI	Running, UNTESTED
Event Bus	9013	Python/FastAPI+Kafka	Running, UNTESTED
WebSocket Gateway	9010	Node.js/TypeScript	Running (4 restarts), UNTESTED
Notification	9009	Python/FastAPI	Running, UNTESTED
Scheduling	9008	Python/FastAPI	Running, UNTESTED
Service Registry	8001	Python/FastAPI	Running, UNTESTED
SSO	—	Integrated w/ Auth	Working on demo

New Services (6, built Feb 2026)

Service	Port	Requirements	Status
Process Analytics	9014	R48-R50	Running, UNTESTED
External Integration	9014	R51-R53	Running, UNTESTED
MCP Server	9000	R54-R56	Running, UNTESTED
Legal Entity	8014	R66-R68	Running, UNTESTED
Business Rules Engine	8018	R69-R72	Running, UNTESTED
BAC Marketplace	9015	R73-R76	Running, UNTESTED

Rebuilt/Revived Services (3)

Service	Port	Requirements	Status
Agent Service	9016	R36-R47	Running, UNTESTED
Prompt Engineering	8012	R39-R41	Running, UNTESTED
Knowledge Hub	8009	R42-R46	Running, UNTESTED

Frontend & Companion Apps (7)

Service	Port	Requirements	Status
Frontend (Next.js)	3000	—	Running, UNTESTED
Engage App	3001	R25-R29, R45	Running, UNTESTED
Manager App	3001	R30-R33	Running, UNTESTED
Process Designer	3002	R77-R79	Running, UNTESTED
DXG Service	9011	R34-R35	Running, UNTESTED
Process Views	8019	R57-R59	Running, UNTESTED
Process Versioning	8020	R60-R62	Running, UNTESTED
Process Linking	8021	R63-R65	Running, UNTESTED

Data Intelligence

Scaled to 0 replicas (replaced by Process Analytics)

Architecture & Data Patterns

Deployment Model (CURRENT)

K3S Kubernetes on dev-01 server (65.21.153.235):

29 pods in flowmaster namespace
Local Docker registry: localhost:30500
Images deployed via docker save/gzip/scp/docker load (NOT via CI/CD pipelines)
Code NOT pushed to GitLab repos yet

Network Architecture (Dev-01 Server)

Internet → Nginx (port 80) → K3S ClusterIPs
  / → frontend (10.43.185.219:3000)
  /api/ → api-gateway (10.43.193.134:9000)
  /ws/ → websocket-gateway (10.43.43.253:9010)

WARNING: ClusterIPs are hardcoded in nginx config, will change if services recreated.

Synchronous Communication

HTTP REST via API Gateway (port 9000):

Frontend → API Gateway → Services
Service-to-service calls (sync where needed)
JWT validation at gateway + service level

Asynchronous Communication

Kafka Event Bus:

Services publish domain events (e.g., task.created, execution.completed)
Event Bus routes via subscriptions (webhook delivery)

Real-Time Streaming

WebSocket (Socket.io) + Redis:

WebSocket Gateway (port 9010)
Services publish to Redis channels
Gateway broadcasts to connected clients

Databases

ArangoDB: Process definitions, task graphs (shared container, port 8529, 3 months uptime)
PostgreSQL: Auth, executions, scheduling, notifications
Redis: Caching, pub/sub, session management

Key Concepts

Process Explorer vs. Process Designer

Process Explorer: Full experience - browse processes, view analytics, manage data
Process Designer: Visual drag-and-drop editor (R77) that sits INSIDE the Explorer

Architectural Decisions (D1-D18)

D1: SSO KEEP (working on dev-01 server)
D3: Notification + Communication MERGE
D4: AI Agent Service = Agent Orchestration
D5: Learning Management → Agent Service (BUILT)
D6: Prompt Engineering KEEP SEPARATE (BUILT)
D7: Internal Data Hub = Knowledge Hub (REVIVED)
D8: Observability SKIP FOR NOW
D9: Permissions enhancement SKIP FOR NOW
D10: SDX is MOST CRITICAL integration
D11: DXG + Engage REPLACE human task execution UI
D12: Process Explorer ≠ Process Designer
D13: SDX REST-only internally; MCP via unified FlowMaster MCP Server
D14: Agent learning from Engage feedback → Knowledge Hub → Agent Service loop
D15: Process Analytics replaces empty Data Intelligence
D16: BAC = process marketplace
D17: Business Rules = first-class DMN-style objects
D18: Process Designer = Visio-quality drag-and-drop with swimlanes + inline AI

Deployment Servers

Server	IP	URL	Status (Feb 8)
Dev-01	65.21.153.235	—	UP, K3S, 29 pods + SDX (Docker)
Dev-02	65.21.52.58	—	High-perf dev, agent pods
Playground	89.167.2.145	—	Pilots, classic Docker
Production	91.99.237.14	app.flow-master.ai	UP (HTTP 200), Docker Compose
Staging/Dev (old)	91.98.159.56	staging.flow-master.ai, dev.flow-master.ai	DECOMMISSIONED

Note: Dev-01 server has ONE FlowMaster environment (not separate dev/test/staging). SDX is part of FlowMaster (currently standalone Docker, should be in K3S). Plane CE is a separate project management tool, not part of the product.

Tech Stack Summary

Backend: 11/13 core services Python/FastAPI, 2 Node.js/TypeScript Frontend: Next.js, React, TypeScript, Radix UI (shadcn/ui), TailwindCSS Source Control: GitLab (primary), GitHub (mirror) Container Orchestration: K3S (demo), Docker Compose (prod/staging/dev) CI/CD: GitLab CI, runner flowmaster-demo-runner v18.8.0

Requirements Coverage (79 total)

All 79 requirements (R01-R79) have code written and deployed as Docker images. NONE are integration-tested. See deployment-snapshot-20260208.md for full mapping.

When to Use This Skill

System Architecture: Understanding service topology, communication patterns
Service Dependencies: Which services interact and how
Technology Decisions: Tech stack, databases, messaging
Integration Planning: Designing APIs or workflows
Deployment Status: What's running where
Stakeholder Communication: High-level system explanations

Related Skills

flowmaster-backend — API endpoints, service contracts
flowmaster-server — Infrastructure, deployment, CI/CD
flowmaster-frontend — UI components, patterns
flowmaster-database — ArangoDB schema, collections
flowmaster-environment — Service env vars, ports, config
flowmaster-tools — MCP tools, integrations, SDX