miaotouy.AIOHub 0.4.2-r.1

AI platform application, providing various practical development and daily tools, as well as highly controllable LLM interaction.

A desktop hub application developed with Tauri + Vue 3 + TypeScript, offering a variety of practical development and daily-use tools. 🌟 Flagship Feature: LLM Intelligent Chat AIO Hub's chat function is a highly customizable, conversational AI workspace designed for complex tasks and deep exploration. 🌐 Open Model Ecosystem Seamless integration, flexible expansion. - Multi-provider support: Natively supports major LLM providers including OpenAI, Anthropic (Claude), Google (Gemini), Cohere, DeepSeek, and more. Easily connect and manage locally running language models via OpenAI-compatible APIs (e.g., Ollama, LM Studio, Llama.cpp). - Custom endpoints: Add custom API endpoints for all integrated service types (e.g., OpenAI, Claude, Gemini), enabling exceptional extensibility. - Visual model management: Centrally manage model metadata in settings, define capabilities (e.g., vision, tool use), and enable intelligent filtering in tools based on task requirements (e.g., "requires vision capability"). 🌳 Dual-View Conversation Management Freely switch between traditional linear lists and innovative non-linear graphs to suit different scenarios. - Traditional Linear View (Linear List View) - Ultimate performance: Virtual scrolling powered by @tanstack/vue-virtual effortlessly handles ultra-long conversations with thousands of messages while maintaining smoothness. - Familiar experience: Classic top-down message flow, intuitive and zero learning curve. - Message navigator: Equipped with quick jump (to top/bottom) and new message indicators. - Fine-grained control: Hover toolbar for messages supports copy, edit, regenerate, branch creation, context analysis, message disabling, and more—enabling deep conversation control. - Non-linear Conversation Graph - Infinite canvas: Driven by Vue Flow + D3.js, presents conversation history as an interactive tree-like network. - Dual layout engine: Offers hierarchical tree layout and dynamic force-directed physics layout with one-click switching. - Advanced node operations: Enable/disable branches, view details, copy content, delete subtrees. - Free structural reorganization: Drag and connect lines to "graft" branches, freely restructure conversation flows—all structural operations support undo/redo. - Visual assistance: Built-in minimap, zoom/pan controls, and active branch highlighting for easy navigation within complex conversation networks. 🛠️ Professional-Grade Prompt Engineering Full control over every token sent to the model. - Context Analyzer: Developer-level debugging tool. Clearly view the complete final prompt, message history, token consumption, and raw request payload sent to the LLM. - Advanced Context Injection: Introduces declarative message injection strategies, allowing precise control over where preset messages are placed in context (e.g., "second-to-last" or "before a specific anchor"), and supports importing SillyTavern character cards and presets for seamless reuse of community resources. - Macro System: Use dynamic macros like {{user}}, {{time}}, {{random::...}} in prompts to build powerful automated workflows. - Agents: At its core, an agent is a sequence of preset messages, allowing you to combine System, User, and Assistant messages (including special placeholders like "User Profile" and "Chat History") into reusable templates. By dragging to reorder or enabling/disabling specific messages, you can create highly customized conversation starters or contexts, with support for import/export and community sharing. - Dynamic Agent Switching: Unlike traditional chat apps that bind one session to one assistant, AIO Hub decouples sessions from agents. You can switch agents within the same conversation—let a "coding expert" write code, then seamlessly switch to a "documentation expert" to write comments—enabling conversations centered around the "task" rather than a fixed "assistant." - Personas: Predefine multiple user identities (e.g., "Senior Architect," "Code Reviewer") and switch anytime to help AI better understand your role and needs. 🖼️ Multi-modal Interaction - File upload and reference: Drag and drop or paste images, PDFs, TXT files into the chat window. Files are uniformly processed by the asset manager, enabling advanced functions like "summarize this PDF" or "analyze this chart." - Vision model support: Fully compatible with vision capabilities of multimodal models like Gemini, Claude, GPT-5, Qwen-VL. ✨ Core Highlights 🖼️ Free Window Management Break free from traditional layout constraints. - Component-level separation: Not just tools—chat input boxes and conversation areas can also be dragged into independent floating windows. - State synchronization: All detached windows share the same state source—changes in one window update all others in real time. - Memory function: Automatically remembers the position and size of all windows. 🎨 Ultimate Visual Experience - Native effects: Supports Windows Mica / Acrylic and macOS Vibrancy frosted glass effects. - Dynamic wallpapers: Supports video/image carousel wallpapers, combined with CSS blend modes for an immersive workspace. - Deep customization: Built-in CSS editor allows real-time modification of every visual detail. 🧩 Powerful Plugin System - JavaScript plugins: Lightweight UI extensions, write-and-use immediately. - Native plugins (Rust): High-performance backend extensions, dynamically loaded via DLL. - Sidecar plugins: Independent process plugins written in any language. 🚀 Productivity Toolkit 📊 Git Repository Analyzer High-performance analysis tool built on Rust git2-rs - No dependencies: Doesn't rely on system Git commands—directly reads .git database for lightning-fast speed. - Streaming analysis: Uses streaming technology to instantly open large repositories and render commit graphs in real time. - Multi-dimensional visualization: Contributor heatmaps, commit frequency charts, lines-of-code statistics. 👁️ Smart OCR Multi-engine text recognition solution - Multi-engine switching: Supports VLM (GPT-4o), Windows Native (offline, fast), Tesseract.js (pure frontend). - Smart slicing: Proprietary long-image slicing algorithm automatically detects blank areas and splits long screenshots, greatly improving recognition accuracy. - Batch processing: Supports concurrent recognition of multiple images and result export. 📦 Asset Manager Application-level resource center - Unified indexing: Centrally manage all images, documents, and media files generated across tools. - Automatic deduplication: Content-based hashing (SHA-256) automatically removes duplicates, saving storage space. - High performance: Fast indexing and filtering powered by Rust backend, supports infinite scrolling. 🎨 Regex Applier Dual-engine regex processing tool - Real-time preview: Frontend JS engine provides millisecond-level input feedback. - Batch processing: Backend Rust engine handles large-scale file modifications with high performance. - Rule chains: Combine multiple regex replacements into a single processing pipeline. 📝 Rich Text Renderer High-performance rendering engine built for LLM streaming output - Flicker-free streaming: Uses incremental diff algorithm and patch system to perfectly eliminate flickering during streaming, delivering a smooth typewriter-like experience. - Deep mixed typesetting: Proprietary parser supports perfect nesting of Markdown with arbitrarily deep HTML tags. - Rich interactive components: - Code blocks: Integrated with Monaco Editor, offering professional syntax highlighting, folding, and font controls. - Chain-of-thought: Native support for <think> tags, displaying LLM reasoning processes via expandable dynamic components. - Charts and formulas: Built-in Mermaid diagram rendering (with zoom/separate window support) and KaTeX math formula rendering. - MD Style Editor: - WYSIWYG: Real-time preview panels for each Markdown element (headings, paragraphs, quotes, code) with independent customization. - Full control: Fine-tune CSS properties like font, color, margins to create a unique reading experience. - Flexibility: Supports one-click toggle of custom styles or reset to system defaults. 🛠️ Additional Practical Tools - JSON Formatter: Smart JSON formatting and beautification with syntax highlighting and error hints—can be sent directly to chat for analysis with one click. - Token Calculator: Estimate text token count, supporting multiple tokenization models. - Color Extractor: Screen color picker, image palette analysis. - Text Diff Tool: Professional-grade diff tool based on Monaco Editor. - Directory Tree Generator: Generate project structure trees with .gitignore filtering support. - And more tools…