Meet the First Agent Skills for Barcode Scanning

When it comes to barcode scanning, right now your friendly neighborhood AI coding agent is like a toddler trying to calibrate an espresso machine. It writes code, sure, but the result is low-quality, defaults are wrong, performance isn't tuned, and edge cases break it.

So today, we’re launching Scandit Agent Skills, the first instructions that teach AI coding agents to integrate barcode scanning properly. If you use Cursor, Claude Code, GitHub Copilot, OpenAI Codex, Gemini, OpenCode, or any agent that supports the Agent Skills Open Standard, a single line of copy-paste code installs the bundle.

To be fair to the robots: most human developers don't write effective barcode scanning code either. Neither agents nor humans show up with the domain knowledge our team has accumulated over 15 years, delivering barcode scanning for everyone from 1.3 million Walmart associates to SMEs with just a handful of devices.

80 billion scans a year teaches you things.

Scandit's Agent Skills port all that knowledge into your coding agent. Describe what you want, and the skills write the integration into your codebase: product selection, SDK setup, license activation, UI wiring, performance defaults, edge cases.

One prompt. 15 years of scanning expertise at your fingertips. And because skills track every SDK release, the generated code always targets current, validated APIs.

Available skills

Each skill targets a specific Scandit product. The first batch is roughly forty skills covering barcode and label scanning, spanning native iOS, native Android, web/JavaScript, Cordova, React Native, Flutter, Capacitor, and .NET. More are on the way, ID scanning included.

Install now

Run this in a terminal in your project directory, then follow the instructions to select your coding agent.

npx skills add https://github.com/scandit/skills

Claude Code users can install directly as a plugin. Paste these commands directly in Claude Code, one at a time.

/plugin marketplace add scandit/skills
/plugin install scandit-sdk@scandit-plugins

Cursor users can do a one-click install of the Scandit plugin from the Cursor marketplace.
→ Install in Cursor

What Scandit Agent Skills teach

The skills cover documentation lookup, product selection, integration, configuration, customization, debugging, SDK version migrations, and switching from another SDK to Scandit.

Docs and product knowledge

The principle is simple: verify against the current documentation before you write any Scandit code. Scandit moves fast. Config options change, documentation gets rewritten, new APIs are released.

An agent leaning on training data alone is navigating with a photograph of the city. Possibly a beautiful photograph, but one taken months or years ago. Streets may have been renamed since.

That's why Scandit skills tell agents to look up the current docs before writing anything Scandit-specific, and to start from our curated sample code rather than reconstructing from memory.

Product selection

The developer describes the use case in plain language inside their AI coding tool ("scan barcodes on a warehouse picking app on Android," say). Photos of your use case are even better. If anything's missing, the agent asks.

Our product-picker skill (data-capture-sdk) recommends the right Scandit product and framework, grounded in fifteen years of real customer integrations.

Over to you (and your agent)

The matching skill can then write the integration directly into the codebase: SDK setup, license activation, UI wiring, performance defaults, edge cases. All of it.

After that, it’s down to your specific circumstances. Want to customize our flagship single scanning product SparkScan to match your brand colours? Just ask your agent.

Want to see if batch scanning would work better for your use case? Tell your agent in natural language to try using MatrixScan Batch instead of Barcode Capture.

Need to evaluate the performance of Scandit against your current open-source barcode scanning? Ask your coding assistant to replace the open-source library with Scandit to get a usable test app with a single prompt.

How we test Scandit Skills

We follow Anthropic’s methodology for measuring whether skills actually help developers. This means running the same model on the same task, once with the skill and once without, then comparing.

Right now we have around 500 evaluation cases across 44 skills, drawn from the typical use cases, snags, and dead ends we've seen across thousands of customer integrations. We'll keep adding to them as models change, and their behaviour with them.

For example, the 18 evaluation cases for the sparkscan-ios skill cover greenfield integrations (UIKit and SwiftUI frameworks), version migrations, and swap-ins from other libraries. The cases check scenarios such as hardware trigger support for warehouse workers in gloves, rejecting specific barcodes with custom error feedback, or continuous scanning for high-volume warehouse apps. Every test also verifies that pre-existing code stays untouched.

Get started

Run this in a terminal in your project directory, then follow the instructions to select your coding agent.

npx skills add https://github.com/scandit/skills

Claude Code users can install directly as a plugin. Paste these commands directly in Claude Code, one at a time.

/plugin marketplace add scandit/skills
/plugin install scandit-sdk@scandit-plugins

Cursor users can do a one-click install of the Scandit plugin from the Cursor marketplace.
→ Install in Cursor

• Scandit Agent Skills docs
• GitHub repo

Found an issue? Report bugs, outdated SDK patterns, or incorrect guidance in the issue tracker. If a Scandit product, framework, or workflow you need isn't covered, open a feature request.