yushulx/dynamsoft-sdk-skill

dynamsoft-sdk

Choose and license the right Dynamsoft SDK before writing code.

Quelltext ansehen

Verwendung

  1. Installationsbefehl kopieren und das Quellpaket vor der Installation in der eigenen Agentenumgebung prüfen.
  2. Benötigte Werkzeuge, Zugangsdaten und Hosts im Original prüfen.
  3. Eingabe und gewünschtes Ergebnis beschreiben und die Ausgabe mit den Quellen abgleichen.
npx skills add https://github.com/yushulx/dynamsoft-sdk-skill --skill dynamsoft-sdk

Vor dem Start

  • Die Installation umfasst weder Modellabo noch API-Kontingent oder Dateizugriffsrechte.
  • Sterne und Installationen zeigen Aktivität, keine unabhängige Qualitäts- oder Sicherheitsprüfung.
Originales Skill-Dokument

Aus dem Quell-Repository gerendert; Überschriften, Beispiele, Code, Tabellen, Links und Bilder bleiben erhalten.

Dynamsoft SDK

Use this skill when the product, the package, or the license is still open. It contains no platform integration code; it decides what to build and how to license it, then hands off.

1. Pick the capability

Capture Vision is one SDK per platform whose modules cover four capabilities: barcode, document, MRZ and zoned text. The capability is selected by the template string, not by a different package, so "which product" and "which platform" are two separate questions.

The user needsCapabilityModule
Barcodes, QR, DataMatrix, PDF417, MaxiCodeBarcodeBarcode Reader
Document edge detection, crop, deskew, cleanupDocumentDocument Normalizer
MRZ from passports, ID cards, visasMRZMRZ model on Label Recognizer
Zoned text: price tags, VIN plates, label fieldsTextLabel Recognizer
Typed fields from a raw string (VIN, GS1 AI, AAMVA)ParsingCode Parser

Then pick the packaged product, which changes what you install, not how you call it:

SituationProduct
Only barcodes, smallest dependencyDBR — dynamsoft-barcode-reader-bundle
Barcodes plus any other capabilityDCV — dynamsoft-capture-vision-bundle
A preset document scan screen, web only, no custom UXMDS — dynamsoft-document-scanner
A preset MRZ scan screen, no custom UXMRZ Scanner — web and mobile SDKs
A ready-made multi-barcode scanning app, no codeBBS — Android/iOS app, not an SDK
Physical scanner in a browserDWT
Document viewer in a browserDDV

Two capability facts that are commonly answered wrong:

  • A custom document or MRZ experience is a Capture Vision job. The Document Scanner (MDS) and

MRZ Scanner SDKs are packaged flows; the moment the UX or workflow has to be different, build it on the Capture Vision templates (DetectAndNormalizeDocument_Default, ReadPassportAndId).

  • There is no standalone full-page OCR SDK. The OCR add-on belongs to Dynamic Web TWAIN and is

used alongside it, typically on a desktop or workstation. Short text in known regions (labels, cards, printed VIN plates) is read by the Dynamsoft Label Recognizer, a DCV component.

  • AAMVA driver licenses are read by DBR, which decodes the PDF417 and parses the fields.

Passports and ID cards with an MRZ zone are MRZ Scanner / DCV.

2. Pick the package: DBR or DCV

DBRDCV
Packagedynamsoft-barcode-reader-bundle (JS), dynamsoft-barcode-reader-bundle (pip)dynamsoft-capture-vision-bundle
ContainsBarcode engine onlyBarcode + document normalization + label/MRZ recognition
SizeSmallerLarger
Entry pointCaptureVisionRouterCaptureVisionRouter

DBR is a submodule of DCV, not a legacy product: the barcode engine is identical, and the same CaptureVisionRouter code works with either package after changing the import. Pick DBR when the task is barcode-only; pick DCV the moment a second capability is needed.

3. Get a license key

Every SDK in this family needs a key before it will run. Never hardcode a real key and never ask the user to paste one into the conversation.

With the license MCP (preferred)

Call get_trial_license with the product and the edition that matches the app:

Product argumentEdition (package)
dcv (default), dbr, mrz, mdscross-platform, web, mobile, desktop
dwt, ddvweb

The call returns licenseKey, licenseKeyMasked, expirationDate and source.

The edition must match where the code runs. The runtime enforces it: a desktop key used in a browser fails with [License] License mismatch: Browser requested. Ask for web for a browser app, mobile for a phone app, and cross-platform when the same key has to serve both.

One key covers the products. A key issued for one product licenses the others too — in testing a Capture Vision cross-platform key licensed the Barcode Reader web bundle without complaint. Reuse the key you already have instead of requesting one per product; only the edition has to be right.

Do not classify a key by its prefix. Current keys start with DLS, but older or product-specific entries in an account can look different, and the prefix alone does not tell you what the key licenses. describe_license reports what you actually hold.

  • `status: "ok"` — write licenseKey into an environment variable (see below) and show only

licenseKeyMasked in the conversation.

  • `status: "login_required"` — a browser window opened on the Dynamsoft Google sign-in. Ask the

user to finish signing in, then call get_trial_license again. The session is captured, so there is no second sign-in. Do not switch to another approach while this is pending.

  • MCP not available — call get_trial_page_url and hand the URL to the user, or use

https://www.dynamsoft.com/customer/license/trialLicense/?product=<product>&package=<package>.

Then call get_license_guide with the product and platform to get the exact initialization call for that stack, and describe_license to show a masked summary or the expiry.

Wiring the key

Always through configuration, never as a literal:

WhereVariable
Node / web buildDYNAMSOFT_LICENSE_KEY exposed by the bundler (VITE_, NEXT_PUBLIC_, …)
PythonDYNAMSOFT_LICENSE_KEY
.NET / Java / C++DYNAMSOFT_LICENSE_KEY
AndroidBuildConfig field fed from a local gradle.properties, never committed
iOS / Flutter / RNa secrets file excluded from version control

DWT takes the same license key, assigned to Dynamsoft.DWT.ProductKey, and the DWT service on the host must be configured with the same key.

4. Open the implementation skill

Skills are split by platform, not by capability, because that is how the SDK is packaged and installed: one router class, one asset tree, several templates. MRZ and document capture therefore live inside the platform skill, not in separate ones.

TargetSkillCovers
Browser: JS/TS, React, Vue, Angular, Nextcapture-vision-webbarcode, document, MRZ, label, PDF
Android, iOS, Flutter, React Native, MAUIcapture-vision-mobilebarcode, document, MRZ, label, camera
Python, .NET, Java, C++, Node.js, Electroncapture-vision-desktopbarcode, document, MRZ, label, batch/PDF
Physical scanner acquisition, including the REST serviceweb-twainTWAIN/SANE/WIA/ICA/eSCL, service, REST
In-browser document viewer and annotationdocument-viewerDDV viewer/editViewer, export
Replacing ZXing / ML Kit / Scandit, or a version upgradesdk-migrationAPI mapping, RTU → core, v9 → v11
Improving decode rate on hard imagesbarcode-template-optimizertemplate tuning, reports

If a request crosses platforms (for example "scan on the phone and review on the web"), read two skills and state which file each part belongs to.

Read only the one you need, plus its references/ when the answer depends on platform detail.

5. Version baseline

Checked 2026-11-10. Treat anything older in an existing project as a possible upgrade case, and prefer these when generating new code.

PackageVersion
dynamsoft-capture-vision-bundle (npm / pip)3.6.5000
dynamsoft-barcode-reader-bundle (npm / pip)11.6.5000
dynamsoft-barcode-reader mobile (Android/iOS/Flutter/RN/MAUI)11.6.4000
dynamsoft-capture-vision mobile3.6.4000
dwt19.5.0
dynamsoft-document-viewer5.0.1
dynamsoft-mrz-scanner4.1.0
dynamsoft-document-scanner1.5.0
dynamsoft-capture-vision-for-node3.6.4003

Rules that prevent most bad answers

  1. Use the core API, not the RTU wrappers. BarcodeScanner, BarcodeReader, DocumentScanner

and MRZScanner are ready-to-use components with a fixed UI and almost no programmability. Production code uses CaptureVisionRouter + CameraEnhancer + CameraView + CapturedResultReceiver. Mention RTU only as a prototype shortcut, with that caveat.

  1. Never invent API names. If a class or method is uncertain for the user's version, say what

needs checking and link the official sample instead.

  1. Sample first. Name the sample family the code is based on.
  2. Self-host SDK assets. Do not load wasm/models from a third-party CDN at runtime in code you

generate; copy the installed package's dist into the app's static assets.

  1. Clean up. Dispose routers, enhancers, views and workers; remove listeners; handle

React StrictMode double-mounting.

  1. Camera needs a secure context — HTTPS or http://localhost, and a user gesture on mobile

browsers.

aus demselben Repository

Weitere Skills

Alle Skills
yushulx
Community

capture-vision-desktop

Dynamsoft SDKs in server and desktop code with Python, .NET, Java, C++, Node.js and Electron. Covers barcode and QR decoding from images and PDFs, document normalization, MRZ parsing, result item typing, per-language license initialization, and the native asset and handle-release rules that break batch jobs.

Installationen / 1h
1
GitHub Stars
0
Aktualisiert
11. Nov.
yushulx
Community

capture-vision-mobile

Dynamsoft SDKs in mobile apps. Covers barcode and QR scanning, MRZ passport and ID reading, document detection and normalization on Android, iOS, Flutter, React Native, .NET MAUI and Swift Package Manager, with licensing, camera permissions, lifecycle and the CaptureVisionRouter pipeline.

Installationen / 1h
1
GitHub Stars
0
Aktualisiert
11. Nov.
yushulx
Community

capture-vision-web

Dynamsoft SDKs in the browser with JavaScript and TypeScript. Covers barcode and QR scanning, MRZ and passport reading, document edge detection and normalization, label text recognition, PDF input, React/Vue/Angular/Next asset paths, licensing, and the camera and cleanup pitfalls that break web capture.

Installationen / 1h
1
GitHub Stars
0
Aktualisiert
11. Nov.
yushulx
Community

document-viewer

View, annotate, reorder and export PDFs and images in a browser with Dynamsoft Document Viewer. Covers Core licensing and init, the viewer/editViewer split, page and annotation operations, saving back to PDF, and pairing with Web TWAIN for a scan-to-edit flow.

Installationen / 1h
1
GitHub Stars
0
Aktualisiert
11. Nov.