Scraib.app intercepts text through a system-level keyboard shortcut on macOS. When you press Control plus R in any application, it captures the selected text, sends it through your chosen AI provider, and replaces it with an improved version. The whole process happens without leaving your current app or opening a separate interface.
The data pipeline works through macOS accessibility APIs. Scraib monitors for the keyboard shortcut trigger, reads the selected text from the active application, passes it to the AI model with your predefined rules, then injects the rewritten text back into the same field. This approach means Scraib functions identically whether you're writing in VS Code, Figma, Slack, or any other Mac application. The technical architecture treats every app the same.
Custom style rules let you define tone and writing preferences either globally or per application. You can set one ruleset for Slack messages and a different one for code comments in VS Code. The app stores these rules locally and appends them to each rewriting request. This gives you consistent output without manually typing instructions each time.
Scraib supports multiple AI providers through their standard APIs. You can route requests through various commercial providers or use Grok and OpenRouter. There's also support for Ollama, which means you can run models entirely on your Mac without any external API calls. LiteLLM integration gives you access to additional providers through a unified interface. With the one-time purchase option, you bring your own API keys and pay the AI provider directly for usage.
The privacy model keeps data on your device. When using Ollama, nothing leaves your Mac. When using external providers, requests go directly from your machine to that provider's API. Scraib doesn't route through its own servers or log your text. This matters for anyone working with sensitive code, client communications, or proprietary content.
The app was built by Andrei Alikimovich, a non-native English speaker in Seattle who got tired of copying text into AI chat interfaces for rewriting. He wanted the workflow to disappear. Scraib spent two months in refinement before public release after he initially built it for personal use.
Technical limitations center on platform constraints. This only works on Mac. No Windows, no Linux. The accessibility API approach that makes the universal app integration possible is macOS-specific. Prices vary by country because the app uses App Store pricing tiers, which adjust based on local currency and regional pricing policies.
The integration list includes VS Code, Figma, and Slack explicitly, but the architecture supports literally any Mac application that accepts text input. If you can type in it, you can rewrite in it. That's the technical advantage of hooking into system-level APIs rather than building individual app integrations.
Pricing includes a seven-day free trial with full feature access. After that, you can subscribe for three dollars and ninety-nine cents per month, which includes AI usage through Scraib's infrastructure. Alternatively, a ten-dollar one-time purchase gives you lifetime access but requires your own API keys for the AI providers.
The methodology is straightforward prompt engineering. Your custom rules get prepended to the rewriting request along with the original text. The AI model processes it and returns improved text. No fine-tuning, no custom models. Just API calls with structured prompts.