A collection of open-source tools, browser utilities, and AI applications I've designed and built.
A high-speed, dual-process autonomous browser agent harness combining sub-150ms Jev reflex decisions with Stagehand stealth automation and Gemini/Claude System 2 reasoning. Features a Next.js 16 operator dashboard with live CDP viewport screencasting.
A free, open-source macOS window manager for AI agents. Tell your agent in plain English to arrange your windows—like "put IDE on the left and stack browser with Spotify on the right"—and it handles the rest on any screen resolution.
An AI-powered diagram studio and interactive visual editor that converts natural language into live Mermaid.js system architecture diagrams, featuring self-healing syntax auto-fix, Monaco editor, and pan/zoom SVG viewport canvas.
An open-source AI course generator that generates complete, structured curriculums (modules, lesson content, and interactive quiz checkpoints) from a single prompt. Features a dual-provider model for cloud/local inference (Gemini & Ollama) and an offline-first data layer with IndexedDB caching and auto-sync queue to MongoDB.
A production-grade, IDE/agent-harness agnostic, and database-agnostic Python library that translates natural language queries into safe SQL statements and executes them against a database, with a built-in strict safety layer and automatic Azure Active Directory (Entra ID) token authentication support.
An agent-native, AI-powered personalized learning companion. Features RAG knowledge hubs for document Q&A, adaptive custom quizzes, mind-mapping tools, and Manim-powered visual math animations.
A utility to automatically remove digital watermarks from videos generated by Google Gemini, restoring them to a clean, watermark-free state.
A client-side browser extension that dynamically blocks subscription prompts, banner ads, and marketing overlays on DataCamp, providing an uninterrupted focus mode for students.