ForwardAgentKit

An agent harness that runs inside iOS and macOS apps, with Swift APIs, multiple AI capabilities, and built-in UI components.

Supports iOS 17+ and macOS 14+.

What it does

Apps supply their domain tools and control credentials, data and permissions. Model inference uses configured cloud providers or a local model server. Available capabilities depend on the provider, model and account.

Apps using it

Linberry

An iOS language-learning app with context-aware tutoring, translation and reply drafting. It saves learning points for flashcards and exercises, provides pronunciation and podcast lessons, and supports selected-text learning through an iOS action extension.

A real Linberry conversation explaining serendipity with bilingual definitions and examples.
Ask about a word and get bilingual explanations and examples.
The learning point sheet generated from the serendipity query, showing translation, meanings and usage notes.
Open the saved learning point for structured meanings, usage notes and pronunciation.
A real Linberry conversation with a generated illustration comparing serendipity and luck.
Use an illustrated comparison to make similar words easier to remember.
An existing Linberry podcast covering eight learning points, actively playing with progress and follow-along dialogue.
Listen to a saved eight-point Podcast, with playback progress and highlighted follow-along dialogue.

Violet

A native iOS and macOS agent app for using the component’s AI capabilities in a conversation. It supports file attachments, image and audio tools, music and video generation, web research, media playback and saved conversation artifacts.

A real Violet conversation generating spoken audio, with in-app playback and sharing controls.
Generate spoken audio in a conversation, then play or share the saved result.
Violet's AI services screen listing text, image, speech, music, video, web and other capabilities.
Configure AI capabilities independently, from text and images to audio and web tools.
Violet's text-to-speech settings with Microsoft Edge and OpenAI providers and an ordered model list.
Set model priority across providers for each capability; speech synthesis is shown here.

Both apps are in development. App workflow screenshots were captured from a physical iPhone through iPhone Mirroring, using real queries and generated results, plus an existing multi-point Podcast. Violet settings screenshots use an iOS Simulator with sample provider configurations. Swipe or use the arrows to browse; click an image to enlarge it.