Firefox WebExtension and embeddable library that runs local MLC/WebLLM models on WebGPU, with a priority scheduler, multi-step decoding, and models from HuggingFace, your own URL, or injected straight from disk.

Core orchestration for local AI inference (WebLLM + Transformers.js + MSW)

Browser-native IICP node (consume + serve): discovery-mesh client with mandatory E2E encryption (IICP-CX) + WebLLM provider. Zero-config, ESM.

Browser-native local AI inference library with WebLLM and Ollama support

Noviq AI worker client — native (Ollama) and browser (WebLLM) inference

The browser-platform implementation of in-browser, on-device model engines. It registers two local-model factories on an `@statewalker/ai-agent` `ModelManager` — **transformers.js** (ONNX models via `@browser-ai/transformers-js`) and **WebLLM** (MLC/WebGP

OpenAI-compatible middleware for running WebLLM models locally with offline support

In-browser AI assist for Fillament forms — WebLLM-powered, schema-aware, privacy-safe.

Shared browser-side inference runtime: WebGPU detection, WebLLM engine wrapper, and Kokoro TTS — used by @vantra-design/ask-design-system and @vantra-design/screenreader-empathy.

AI-focused UI components for building LLM-powered interfaces with WebLLM

Provider-agnostic text/copy generation seam — one TextGenerator interface over an in-browser WebLLM engine or a remote OpenAI-compatible API. The text seam of the generation abstraction.

In-Browser LLM inference

On-device LLM sessions for any browser — Chrome's Prompt API, WebLLM, or wllama, picked by feature detection, smallest model first.

**Bringing the power of Large Language Models (LLMs) directly to your Angular applications with WebLLM.**

In-browser LLM integration for PivotHead — natural-language control of pivot tables via WebLLM (no server required)

Unified LLM service (Ollama + WebLLM) for Formstr apps

High-performance in-browser LLM inference for Vue.js using WebLLM and WebGPU

React bindings for @dudko.dev/agent-web: a headless useAgent hook, an AgentProvider context, and optional pre-styled components (chat panel, plan/step view, BYOK key form, WebLLM load bar) that connect the in-browser LLM agent to any React site. UI you ca

A customizable, Fluent UI–based AI chat widget for React applications. Supports light/dark themes, persistent chat history, maintenance mode, draggable UI, and pluggable AI engines (WebLLM, OpenAI, etc.).

On-device LLM inference for Schift Lawyers Ultra Secure Mode. WebGPU (MLC WebLLM) + WASM (transformers.js) fallback. Zero network egress.