Case Studies Overview
This series is distinct from the Extension API reference and the Dashboard Component API reference β those cover "what the API looks like and what it can do." This series focuses purely on real engineering practice: why a design was chosen, what pitfalls were hit, and how engineering standards are applied on the ground. Every case is selected from a real NeoMind ecosystem repository and audited by hand into a reproducible engineering write-up.
The 7 Cases at a Glanceβ
| # | Case | Type | Difficulty | Size | Key Value |
|---|---|---|---|---|---|
| 1 | weather-forecast-v2 | Data extension | Beginner | ~700 LOC | First-extension template (HTTP fetch + periodic metrics + React frontend) |
| 2 | yolo-device-inference | AI inference | Intermediate | ~1950 LOC | Lazy model loading / cross-session reuse / device camera integration |
| 3 | yolo-video-v2 | Streaming extension | Intermediate | ~3900 LOC | stream session + video-frame processing + VLM dashboard interop |
| 4 | onvif-bridge | Protocol bridge | Intermediate | ~2700 LOC | IP camera / standard protocol integration (parallels NeoEyes) |
| 5 | uink-rms-bridge | Protocol bridge | Intermediate | ~2250 LOC | Production-verified bridging |
| 6 | metric_card | Dashboard component | Beginner | ~400 LOC | Component template (value card + threshold / trend / unit) |
| 7 | ne101_camera | Extension + component interop | Flagship | ~2500-3500 LOC (8 sub-pages) | End-to-end deep tutorial |
Reading Pathsβ
Pick the shortest path for your role and goal:
- Path 1 (Newcomer): 1 β 6 β 7 β Build intuition with the simplest extension and component, then tackle the end-to-end flagship case.
- Path 2 (AI Engineer): 2 β 3 β 7 β From single-frame inference to video-stream processing, then full product integration.
- Path 3 (Industrial Integrator): 4 β 5 β 7 β Two protocol-bridge cases as foundation, then a camera-grade product integration.
- Path 4 (Component Developer): 6 β 7 β any extension case β Master the component paradigm first, then learn how to interop with extensions.
Version Alignment Tableβ
Case code is aligned with the source repository's release; the audit locks to a specific commit.
| Case | Source repo version | SDK version | Last audit |
|---|---|---|---|
| 1 weather-forecast | v2.7.6 | SDK 0.6 | 2026-06-22 |
| 2 yolo-device-inference | v2.7.6 | SDK 0.6 | 2026-06-22 |
| 3 yolo-video-v2 | v2.7.6 | SDK 0.6 | 2026-06-22 |
| 4 onvif-bridge | v2.7.6 | SDK 0.6 | 2026-06-22 |
| 5 uink-rms-bridge | v2.7.6 | SDK 0.6 | 2026-06-22 |
| 6 metric_card | v1.7.0 | β | 2026-06-22 |
| 7 ne101_camera | v2.14.9 | β | 2026-06-22 |
A case audit is triggered when the source repository cuts a release (manual, not automated).
Component Source Format Noteβ
NeoMind Dashboard Components ship as hand-written IIFE JavaScript β the file is named
bundle.js, but it is not a compiled artifact; it is human-authored source.Conventions:
- Runtime dependencies are injected via
var React = window.React+var jsx = window.jsxRuntime.jsx(React / JSX runtime are provided by the host page).- Full comments and reasonable line breaks are preserved; readability is close to plain source code β you can read it directly without a source map.
- Current sizes:
metric_cardis 352 lines,ne101_camerais 1972 lines.Therefore the "key code walkthroughs" in cases 6 / 7 point directly at specific line numbers in
bundle.jsβ readers can open the source file and follow along.
Further Readingβ
- Extension API reference β API docs for extension traits, macros, capabilities, lifecycle, and ML model loading.
- Dashboard Component API reference β API docs for component schema, data-source binding, and the render pipeline.
- Shared Engineering Standards Appendix β Engineering standards shared across all cases (code style, error handling, logging, testing, version alignment, etc.).
Last updated: 2026-06-22