# Bolt Guo > AOUOS is the personal website of Bolt Guo, an independent software developer known as boltguo on GitHub. This index covers his background, products, public tools, learning projects, and GitHub organizations. Bolt has a frontend development background and mostly works with TypeScript, React, and Vue. He also explores native apps, local-first software, and AI agents. He builds things he wants to use and learn from; VizLearn grew out of an interest in making complex concepts easier to understand. Outside of software, he enjoys games, anime, and music. His public work spans the boltguo account and the aouos and vizlearnapp organizations. AOUOS is his personal website and project identity; AOUO is the separate pre-alpha agent app runtime described below. These descriptions are based on public project documentation and source code reviewed on October 9, 2026. Check the linked product pages and repositories for current status and requirements. ## About and contact - [Personal website](https://aouos.com/): Personal introduction, selected products and experiments, social profiles, and contact information. The site is a compact static Astro page. - [GitHub](https://github.com/boltguo): Bolt's public repositories, including native apps, frontend tools, Cloudflare services, AI experiments, and earlier learning projects. - [Email](mailto:hello@aouos.com): Contact Bolt about his work, ideas, or potential collaboration. ## Main products - [VizLearn](https://vizlearn.app/): A browser-based learning platform for algorithms, data structures, systems, and AI. Users can change inputs, step through execution, scrub a timeline, and inspect code, visualizations, and logs that stay in sync. Written explanations cover invariants, costs, edge cases, and common mistakes. Prediction exercises ask learners to choose the next operation. Users can share a lesson with its exact input, embed an interactive player, and export images. Recent lessons, bookmarks, and saved inputs can stay in the browser. Cross-device sync and video export are available through the paid plan. - [KotoPlanet](https://kotoplanet.com/): A Japanese-learning app for native Chinese speakers, with a continuous learning path from N5 through N1 covering kana, vocabulary, grammar, kanji, listening, and reading. Spaced review sets the daily workload. Speaking scenarios, shadowing, and pronunciation feedback help learners practice actively. Core learning records are stored locally, and the iOS app bundles offline course content and audio. Signing in enables cross-device sync. An OAuth-protected MCP connection lets compatible AI clients read courses and review queues, and write completed practice back to the same learning record. The public architecture documentation describes Expo clients, a Cloudflare Worker, and separately versioned course content. - [Luti](https://luti.aouos.com/): A native macOS project runtime that gives MCP-compatible AI clients access to a project enabled by the user. With explicit permissions, it can read and edit files, run builds and tests, inspect Git, drive browsers, interact with the desktop, and export artifacts. Shared memory, sessions, and activity records retain project context across AI clients. Local clients connect through loopback MCP. Remote connections use supported tunnels and require authorization and approval on the Mac. Project context is stored on the Mac, and connected AI hosts receive the results of tool calls made through them. Its public Git tools are limited to read-only inspection. ## AI and creative projects - [AOUO](https://github.com/boltguo/aouo): A pre-alpha runtime for local-first agent apps packaged as .aouo packs. Each pack defines its own skills, persistent storage, memory, schedules, tools, permissions, and UI surface. The implementation includes manifest validation, pack-scoped SQLite databases, an agent loop, a scheduler, a Telegram adapter, and a local dashboard with per-pack chat. Sample packs explore notes, content creation, and vocabulary learning. The pack format and product design are still evolving. The [AOUO website](https://aouo.ai/) explains the runtime, pack model, and intended uses. - [Agent Skills](https://github.com/boltguo/skills): Installable workflows for AI assistants that support Agent Skills, including Codex, Claude Code, and Cursor. The collection currently includes gemini-image-gen for Gemini API image generation and human-writing-craft for Chinese and English writing. The image workflow lets users choose the model, image size, and aspect ratio, then saves generated images with their prompts and parameters. The writing workflow covers drafting, revision, translation, style matching, and prose audits while preserving facts and meaning. It also includes an optional Python text scanner. - [Colab model notebooks](https://github.com/boltguo/colab): Google Colab notebooks and ComfyUI workflows for image and video generation on an A100 80 GB runtime. The repository currently documents MiniMax H3 workflows for text-to-video, first-frame animation, first-and-last-frame generation, and two-image reference generation. Qwen-Image 2.1 workflows cover text-to-image, reference-image editing, and optional anime-consistency LoRA variants. The notebooks set up the environment, obtain model weights and workflows, launch ComfyUI, and make the results available to download. These notebooks and workflows integrate third-party models, whose licenses apply separately. - [Anya Codex Pet](https://github.com/boltguo/anya-codex-pet): An unofficial, fan-made animated Anya Forger companion for Codex Work mode. The repository contains standalone v1 and v2 pet packages with metadata and WebP sprite sheets, plus preview contact sheets and validation reports. The v1 package contains the standard animation rows; v2 adds look-direction frames. The documentation gives different import instructions for uploading a sprite sheet and installing a local package. The fan artwork features the SPY×FAMILY character. ## Utilities and developer tools - [CodePin](https://codepin.aouos.com/): An iPhone utility for generating and scanning QR codes and common barcodes, saving useful codes to history, and pinning them to the Lock Screen or Dynamic Island through Live Activities. Codes can contain links, text, Wi-Fi credentials, and vCards. Users can scan with the camera or from photos and customize colors and icons. Data is stored on-device without an app account. The Pro purchase adds private iCloud history sync, custom logos, unlimited Live Activity starts, and HD PNG and SVG export. The [App Store listing](https://apps.apple.com/us/app/codepin/id6763878476) describes the current feature set. - [PicNest](https://github.com/boltguo/PicNest): A self-hosted image host built on Cloudflare Workers, R2, and D1. Its React frontend accepts clipboard and drag-and-drop uploads and includes folders, sorting, a preview grid, and a lightbox. One Worker serves the app, API, image files, and share pages; R2 stores image bytes and D1 stores metadata. Content hashes deduplicate identical uploads. It offers permanent public direct links and revocable share links with optional passwords and expiry. Management requires a single-password login. The repository supports one-click Cloudflare deployment. Complete backups require both R2 objects and D1 data. - [React Vite CLI](https://github.com/boltguo/react-vite): A command-line tool for creating and extending React + Vite projects. TypeScript, Tailwind CSS, React Router, ESLint and Prettier, Zustand, and shadcn/ui are selectable features. The CLI can create a project or initialize an existing directory, then generate pages, components, stores, hooks, and layouts as the project grows. It detects the selected features and language to produce matching files. Project health commands report dependency drift and repair pinned dependency versions, while dry-run options preview supported changes before writing files. - [SBM](https://github.com/boltguo/sbm): A Go web panel and CLI manager for one sing-box server on a personal VPS. It manages VLESS inbounds with Vision and REALITY, along with Hysteria2 inbounds. It combines enabled nodes into a master subscription and supports optional WireGuard egress gateways. The bilingual panel shows host health, protocol configuration, subscription links and QR codes, traffic plans, and daily and monthly history. vnStat tracks traffic on public interfaces. Configured entry quotas can stop and resume sing-box. Configuration validation and rollback, updates, backups, restore, and repair are available through the panel or CLI. - [Reminder Bot](https://github.com/boltguo/reminder-bot): A single-user Telegram daily reminder bot implemented in TypeScript on Cloudflare Workers. Cron triggers process a D1 queue of timezone-aware reminders. A configured chat ID determines the recipient. Commands start and stop reminders, show the schedule and status, and report completion statistics. Done and skip buttons record feedback, while scene-based sticker mappings add illustrated reminders. D1 stores jobs, state, feedback, streaks, and sticker mappings. The delivery code claims jobs, retries sends, and records failed or expired jobs. Schedules and messages are customized in source. - [Watermark Remover](https://github.com/boltguo/wmremover): A Vue browser utility for removing supported Gemini and Nano Banana watermark patterns. It uses inverse alpha blending with known watermark alpha maps to recover image pixels. Processing happens locally in the browser, and images stay on the user's device. The interface supports desktop and mobile layouts and several languages. It is intended for recovering pixels affected by those patterns, rather than for general image editing. ## Learning and UI experiments - [React project-based learning](https://github.com/aouos/react_learn): A Chinese-language React 19 course built around projects, published by the aouos organization at react.aouos.com. Lessons progress through a Todo app, a task-management system, a Next.js ecommerce teaching project, and framework internals. Topics include hooks, routing, state management, data fetching, authentication, testing, deployment, and React design patterns. The repository contains the Markdown tutorials and their documentation site. Readers build the example applications by following the lessons. The course documents its own version baseline and distinguishes building the documentation from validating complete application integrations. - [Vue project-based learning](https://github.com/aouos/vue_learn): A Chinese-language Vue 3 course built around projects, published by aouos at vue.aouos.com. Lessons build from a Todo app to a task-management system, then introduce an ecommerce teaching example and Vue internals. The core lessons use the Composition API and script setup and cover routing, Pinia, reusable composables, testing, and engineering topics. Supplementary articles discuss framework comparisons and design choices, and separate experiments cover SSR and Vapor. The repository contains Markdown lessons and a VitePress documentation site. Readers build the learning applications by following the lessons. - [AOUI](https://github.com/boltguo/aoui): A work-in-progress React component catalog and design reference. It collects buttons, cards, form controls, selection controls, sliders, tables, feedback components, and tooltips in a navigable demo site. CSS-variable tokens define colors, spacing, typography, radii, and transitions, while individual component styles use CSS Modules. A theme page displays the token system. The project serves as a small UI kit and a reference for building a consistent visual language. - [React template](https://github.com/boltguo/react-template): A minimal React + Vite + TypeScript starter with Tailwind CSS, Lucide icons, and ESLint already connected. It includes an example app, strict TypeScript configuration, Tailwind's Vite integration, and scripts for development, type-checking, production builds, preview, and linting. This ready-to-use repository template has a fixed setup; React Vite CLI offers the configurable scaffolding and generators described above. - [Algo-V](https://github.com/boltguo/algov): An earlier React-based experiment in data-structure and algorithm visualization. Its source includes Big-O demonstrations, examples of bubble, insertion, and selection sort, and components for stacks and queues. It uses ECharts and TDesign with JavaScript and Vite. The repository shows examples of Bolt's earlier visualization work. ## Organizations, community, and brand resources - [aouos](https://github.com/aouos): Bolt's programming-learning organization. Its current public repositories host the React and Vue project-based tutorial sites described above. - [vizlearnapp](https://github.com/vizlearnapp): The GitHub organization associated with VizLearn. Its current public repositories provide a feedback hub and logo assets. - [VizLearn feedback](https://github.com/vizlearnapp/feedback): A public issue hub for website bugs, lesson corrections, improvement suggestions, and new lesson requests. Issue templates separate these categories. Reports can be written in English, Chinese, or Japanese. Account, billing, and other private support requests go to the product's support email. - [VizLearn logo](https://github.com/vizlearnapp/logo): The public logo-design and brand-asset repository for VizLearn. - [AOUOS logo](https://github.com/boltguo/logo): A geometric logo-design experiment based on transforming the A, O, U, and O letters into simple shapes, with SVG artwork and a design preview.