9bb02b3dd5a5ba43a6cf44af517cbee0e222d370
System Landscape Simulator
An interactive, frontend-only architecture map for modelling systems and simulated data journeys. It uses a classic ComfyUI-style LiteGraph canvas, but does not fork or modify ComfyUI and performs no live integrations.
Run locally
npm install
npm run dev
The initial landscape contains Website, API, ERP, CRM and Email Platform plus two demo journeys:
- Place order: Website → API → ERP, followed by parallel CRM and Email Platform updates.
- Register customer: Website → API → CRM → Email Platform.
- Monday demo journeys: Contact enquiry, Sample request, Commercial order, and Order retry and recovery.
- Deployment demo: Build Server → Octopus release → Sitecore CM and two CD targets, with parallel target outcomes.
- Process analysis demo: Sitecore → Sample Request Application → manual CRM entry → SAP → Customer Service → Lab / Warehouse → Sample Shipment, with customer and sales notification branches.
Features
- Drag and rearrange systems on the dark LiteGraph canvas.
- Edit or add systems, connections, scenario steps and parallel branches.
- Group systems in explicit nested environment containers such as Azure → Subscription → Resource Group. Containers can be resized, moved with their contents, collapsed to summary nodes, and persisted independently of provider metadata.
- Drill into logical system environments such as Dev, QA, UAT and Prod. Environments have editable components, instances, safe example configuration placeholders and internal topology; Sitecore Prod is included as a fictional CM/CD/MSSQL/Solr example.
- Classify systems with Application, SaaS Service, Cloud Service, Development Tool and Data Store categories, plus independent tags such as Production, Test and Customer Data.
- Use typed Data Flow, Deployment, Dependency and Source Repository relationships. Development playback includes the simulated Bitbucket commit → Azure DevOps build → website deployment journey.
- Model physical operations alongside software with Physical Location and Business Operation categories, plus Physical Goods and Human Task relationships. The Warehouse preset includes receiving, storage, picking, dispatch and returns with clearly labelled physical-goods ports and fictional shipment payloads.
- Run the mixed Physical goods receipt demo: SAP purchase order → Supplier shipment → Warehouse → Warehouse Operative decision → Warehouse Management System → SAP confirmation. Physical locations remain independent from their software systems; all transfers are simulated and inspectable.
- Model persistent database integrations separately from activity: Direct Database Access, Scheduled Import, Replication / Sync and File Transfer. Configure access mode, trigger/frequency, source tables/views/queries, destination, authoritative source and an optional simulated mapping. Live query requests and returned database rows are separate connections; scheduled imports identify downstream copies such as Sitecore items.
- Separate flow purpose from cable relationship, payload type and integration mechanism. Scenarios have editable purposes for Business Transactions, Content & Reference Data, Build & Deployment, Operations & Monitoring, Identity & Access and Physical Logistics; individual transfers can override the scenario purpose and connections can carry multiple purposes without duplicating cables.
- Filter flow purposes independently from system categories, colour active packets by purpose and use the accessible purpose legend. Deployment cables retain their relationship styling while the packet colour communicates Build & Deployment.
- The deployment demo carries fictional package, release, version, environment and target fields through a sequential package/release step and a parallel CM/CD target step. Each target has an independent simulated outcome; no build or deployment system is contacted.
- Compare the example Sitecore live-query and scheduled-import scenarios for the fictional Product / Literature / Webinar Management application and MSSQL database. Database payloads are sample rows/JSON only; no database connection or inventory/accounting engine is implemented.
- Place reusable Website, CRM, ERP, Email Platform, Analytics and API presets, selecting only the capabilities needed for each instance.
- Browse the layered preset library from the Presets tab: searchable system/product catalog entries, shared capabilities, deployment layouts and starter kits. Entries include tags, version metadata and official documentation sources; favourites are stored locally.
- Inspect labelled capability ports and fictional action payloads, expand node actions, and save customised systems as private browser-local presets.
- Build reusable custom systems with form-based action, output and endpoint definitions. Actions support button, incoming request/event and simulated schedule triggers; outputs can run in parallel, and endpoints support multiple delayed example responses that may trigger actions.
- Add restricted transformation pipelines to action definitions for middleware-style Receive, Validate, Transform, Route and Send stages. Pipelines support explicit example payloads, copy/rename/fixed/default/type-conversion/object-construction/array-item-mapping rules, reference lookup values, validation-failure paths and playback inspection without executing arbitrary JavaScript. The demo includes Website → Order Middleware → ERP.
- Model direct request/response exchanges separately from onward event transfers. Request and response payloads are both recorded in the data inspector, with compatibility warnings shown without blocking unusual integrations.
- Run, pause, advance and reset scenarios with animated packets and event log output.
- Analyse an end-to-end process over an existing scenario with a start event, intended/alternative/rejection/cancellation outcomes, required milestones, owner and operation metadata, queue/trigger information, modelled processing/waiting estimates, evidence status, correlation IDs and local record mappings.
- Compare Current State and Proposed State process versions without overwriting the existing process. The analysis highlights the intended path, keeps notification/side-effect branches separate, and reports missing onward paths, unowned handoffs, manual re-entry, untracked notifications and unknown transitions as review findings.
- Adjust playback speed and inspect the active payload.
- Persist the complete landscape, including node positions, in browser local storage.
- Export and import the landscape as JSON. Invalid imports and deleted connection references are rejected with actionable errors.
- Load a landscape JSON document from an HTTP(S) URL through the URL import control. The remote server must return JSON and allow browser CORS requests; the same version and reference validation as file imports is applied.
- Use the Monday Demo guide at the bottom of the app to walk through playback, data inspection and failure recovery.
npm test
npm run build
# Browser UI smoke tests (requires a Playwright browser install once)
npx playwright install chromium
npm run test:ui
## Docker deployment
The production image is a static Nginx container. It serves the simulator on
port 80 inside the container and uses `SYSTEM_LANDSCAPE_PORT` for the host
port (7004 by default):
```bash
docker compose up -d --build
curl http://localhost:7004/healthz
The Gitea workflow builds and deploys main to port 7004, then runs the
Playwright suite against https://syssim.dumas.ddns.net. The post-deploy
browser job fails the workflow if the deployed UI does not pass.
## Limitations
This is intentionally a local simulation prototype. It has no authentication, collaboration, server persistence, external API calls, real payload delivery, or production-grade graph editing. Custom definitions and placed instances are stored in browser local storage; playback state is transient and is not included in saved JSON. LiteGraph supplies canvas rendering and cable interaction; scenario semantics are managed by the Vue application. The minimap is diagnostic rather than a full viewport navigator, and transformation highlighting currently compares top-level JSON fields. Pipeline mappings are deliberately limited to the listed rule operations; arbitrary JavaScript, network lookups and live middleware execution are not supported. The custom builder exposes pipeline definitions as validated JSON so advanced step graphs can be edited without adding another runtime dependency. Flow categories are simulation metadata and do not enforce model behaviour; a packet can be displayed on one selected category at a time even when a connection supports multiple categories. Process findings are review prompts based on configured model data, not proof of operational failure; processing and waiting durations are modelled estimates unless explicitly supported by measured evidence. Preset documentation URLs are reference metadata only; the simulator does not inspect or synchronize vendor products.
Languages
TypeScript
56.4%
Vue
43.3%
HTML
0.2%
Dockerfile
0.1%