AI Plate v1.0.2 — Patch Release

AI Plate v1.0.2 — Patch Release

We are pleased to announce AI Plate v1.0.2, focusing on modular plugin portability, real-time Material Symbols customization, process lifecycle reliability, and comprehensive test suite stabilization.

---

🌟 What's New & Fixed in v1.0.2

• 📦 Custom Plugin .zip Export & Distribution:

  • Export any installed or custom plugin directly from its card in the Plugins section as a portable .zip package.
  • Automatically bundles all tool schemas, parameters, code handlers, and UI extensions with a standardized root plugin.json manifest.
  • Exported archives are 100% plug-and-play and can be installed into any AI Plate instance via drag-and-drop or the Install ZIP button.

• 🎨 Google Material Symbols Live Customization & Glyphs:

  • Live Auto-Apply: Changing any individual icon glyph, font weight, fill, or size in the plugin settings applies to the UI in real time without having to click "Run / Apply".
  • Ligature Integrity: Corrected legacy font ligatures (such as replacing chat_bubble_outline with chat_bubble and question_answer with chat) to guarantee clean, ungarbled icon rendering across all platforms.
  • Expanded Glyph Coverage: Custom icons now cleanly style reasoning depth indicators (.thinking-chip, .thinking-brain-pulse), all Settings tabs (General, Skills, Models, Plugins, Connectors, Security, Config), and the Past Chats section header.
  • State Persistence & Reset: Restored complete parameter serialization in local storage and added a dedicated "🔄 Reset Default" button that restores default glyphs and synchronizes all open controls immediately.

• 🔌 Connector Companion Lifecycle & Stability:

  • Improved process termination on Windows with process-tree cleanup and explicit intentional-stop state tracking, ensuring companion processes report stopped cleanly rather than false-positive crashed.
  • Bundled modular sample connector packages (blender, file-explorer-bridge, obsidian-vault, office-bridge).

• 🛡️ Subsystem Integrity & 100% Test Coverage:

  • Automated test suite passes 100% across Plugin Manager Subsystem, Security Governance, Plug-and-Play Connectors, Companion Lifecycle, Intelligent Context Compressor, and Multi-Agent Chat Modes.

---

💻 Installation Instructions (Windows)

1. Download AI.Plate.Setup.1.0.1.exe from the Assets section below.
2. Run the installer and follow the setup wizard (creates a Desktop shortcut and Start Menu entry).
3. Launch AI Plate:
   • Add your preferred API key (Gemini, OpenRouter, Mistral, Anthropic, OpenAI, etc.) in the Settings tab, or connect to your local Ollama / LM Studio endpoint.
   • Start chatting, analyzing code, or giving autonomous tasks to your sovereign AI assistant!

---

📋 System Requirements

• OS: Windows 10 / 11 (64-bit)
• RAM: 4 GB minimum (8 GB recommended)
• Disk Space: ~250 MB for app installation

---

Open Power to All. Sovereign AI for everyone.

AI Plate v1.0.1 — Patch Release

AI Plate v1.0.1 — Patch Release

We are pleased to announce AI Plate v1.0.1, focusing on modular plugin portability, real-time Material Symbols customization, process lifecycle reliability, and comprehensive test suite stabilization.

---

🌟 What's New & Fixed in v1.0.1

• 📦 Custom Plugin .zip Export & Distribution:

  • Export any installed or custom plugin directly from its card in the Plugins section as a portable .zip package.
  • Automatically bundles all tool schemas, parameters, code handlers, and UI extensions with a standardized root plugin.json manifest.
  • Exported archives are 100% plug-and-play and can be installed into any AI Plate instance via drag-and-drop or the Install ZIP button.

• 🎨 Google Material Symbols Live Customization & Glyphs:

  • Live Auto-Apply: Changing any individual icon glyph, font weight, fill, or size in the plugin settings applies to the UI in real time without having to click "Run / Apply".
  • Ligature Integrity: Corrected legacy font ligatures (such as replacing chat_bubble_outline with chat_bubble and question_answer with chat) to guarantee clean, ungarbled icon rendering across all platforms.
  • Expanded Glyph Coverage: Custom icons now cleanly style reasoning depth indicators (.thinking-chip, .thinking-brain-pulse), all Settings tabs (General, Skills, Models, Plugins, Connectors, Security, Config), and the Past Chats section header.
  • State Persistence & Reset: Restored complete parameter serialization in local storage and added a dedicated "🔄 Reset Default" button that restores default glyphs and synchronizes all open controls immediately.

• 🔌 Connector Companion Lifecycle & Stability:

  • Improved process termination on Windows with process-tree cleanup and explicit intentional-stop state tracking, ensuring companion processes report stopped cleanly rather than false-positive crashed.
  • Bundled modular sample connector packages (blender, file-explorer-bridge, obsidian-vault, office-bridge).

• 🛡️ Subsystem Integrity & 100% Test Coverage:

  • Automated test suite passes 100% across Plugin Manager Subsystem, Security Governance, Plug-and-Play Connectors, Companion Lifecycle, Intelligent Context Compressor, and Multi-Agent Chat Modes.

---

💻 Installation Instructions (Windows)

1. Download AI.Plate.Setup.1.0.1.exe from the Assets section below.
2. Run the installer and follow the setup wizard (creates a Desktop shortcut and Start Menu entry).
3. Launch AI Plate:
   • Add your preferred API key (Gemini, OpenRouter, Mistral, Anthropic, OpenAI, etc.) in the Settings tab, or connect to your local Ollama / LM Studio endpoint.
   • Start chatting, analyzing code, or giving autonomous tasks to your sovereign AI assistant!

---

📋 System Requirements

• OS: Windows 10 / 11 (64-bit)
• RAM: 4 GB minimum (8 GB recommended)
• Disk Space: ~250 MB for app installation

---

Open Power to All. Sovereign AI for everyone.

AI Plate v1.0.0 — Official Release

AI Plate v1.0.0 — Official Release

We are thrilled to announce the official v1.0.0 release of AI Plate — Open Power: the universal, sovereign AI agent harness, knowledge intelligence engine, and autonomous butler built to live directly on your local machine with complete data sovereignty and zero vendor lock-in.

---

🌟 What's New in v1.0.0

• 🕸️ Open Knowledge Format (OKF) & Hybrid Knowledge Router: Dual-tier hybrid knowledge architecture combining Tier 1 deterministic ground-truth specifications with Tier 2 Float32 vector RAG search.
• 🎨 Interactive Visual Force Graph (🕸️ Force Graph): Real-time canvas force-directed graph with anti-collision badge layout, dynamic zoom/pan, and live node inspectors.
• ⚡ Auto-Distillation Engine: Ingest raw PDFs, manuals, and unstructured text and automatically extract invariants, rules, and code links into structured OKF nodes in $< 30\text{ ms}$.
• 🔍 Code Grounding & Invariant Drift Verifier: Automated scanning of workspace code implementations linked to OKF specs to detect missing files or broken implementations.
• 🌪️ Mistral AI Provider: Native integration with Mistral models (e.g., mistral-large-latest, mistral-embed) with automated API key resolution.
• 🤖 Unified Multi-Agent Modes: Seamless Normal, Plan, and Code execution modes with strict capability boundaries.
• 📦 Artifacts & Deliverables Sorting: Sort deliverables by Date (Newest/Oldest), Name (A-Z/Z-A), and File Size (Largest/Smallest).
• 🎬 Video & Media Streaming: Full HTTP 206 byte-range streaming for large video artifacts and external system player integration.
• 🛡️ API Message Sanitization: Strict message sequencing when dispatching multi-turn tool calls to OpenAI-compatible endpoints.
• 🐍 Bundled Python & Voice Engine: Standalone embedded Python 3.11 runtime pre-configured with Kokoro TTS and Moonshine STT libraries for zero-friction local voice without requiring system Python. Model weights remain lightweight and downloadable on-demand in Settings.

---

💻 Installation Instructions (Windows)

1. Download AI.Plate.Setup.1.0.0.exe from the Assets section below.
2. Run the installer and follow the setup wizard (creates a Desktop shortcut and Start Menu entry).
3. Launch AI Plate:
   • Add your preferred API key (Gemini, OpenRouter, Mistral, Anthropic, OpenAI, etc.) in the Settings tab, or connect to your local Ollama / LM Studio endpoint.
   • Start chatting, analyzing code, or giving autonomous tasks to your butler!

---

📋 System Requirements

• OS: Windows 10 / 11 (64-bit)
• RAM: 4 GB minimum (8 GB recommended)
• Disk Space: ~250 MB for app installation

---

Open Power to All. Sovereign AI for everyone.

AI Plate v1.0.0-rc.3 — Mistral AI & Unified Agent Mode

AI Plate v1.0.0-rc.3

🚀 What's New

• Mistral AI Provider: Native integration with Mistral models (e.g., mistral-large-latest, mistral-embed) with automated API key resolution.
• Unified Agent Mode: Streamlined agent experience defaulting to full-access Agent mode with clean mode gating.
• Artifacts & Deliverables Sorting: Sort deliverables by Date (Newest/Oldest), Name (A-Z/Z-A), and File Size (Largest/Smallest).
• Video & Media Streaming: Full HTTP 206 byte-range streaming for large video artifacts and external system player integration.
• API Message Sanitization: Resolved 400 error (Unexpected role 'tool' after role 'user') by ensuring strict message sequencing when dispatching to OpenAI-compatible endpoints.

---

Open Power to All. Sovereign AI for everyone.

AI Plate v1.0.0-rc.2 — Hotfix for Connector Installations

AI Plate v1.0.0-rc.2 — Hotfix Release

This maintenance release resolves a Windows-specific file locking and permission denial (EPERM) issue encountered when installing or updating modular App Connector packages (such as Blender 3D, OBS Studio, and custom companions).

---

🐛 Fixes & Improvements

• Companion Process Lifecycle Management: Ensures active background companion processes (e.g., blender_bridge.py, Python HTTP/WebSocket bridges) are cleanly stopped before attempting package installation or directory modification.
• Synchronous Process Tree Termination: Switched Windows process cleanup to synchronous execution (taskkill /F /T) to guarantee all file handles are fully released by the operating system before file operations commence.
• Resilient Directory Cleanup: Added retry backoff logic (maxRetries: 5, retryDelay: 100) and graceful in-place overwrite fallback for directory extraction on Windows NTFS filesystems.
• Version Bump: Updated to v1.0.0-rc.2.

---

Open Power to All. Sovereign AI for everyone.

AI Plate v1.0.0-rc.1 — Open Power

AI Plate v1.0.0-rc.1 — Open Power

We are thrilled to announce the first release candidate of AI Plate — Open Power: the universal, sovereign AI agent harness and autonomous butler built to live directly on your local machine with zero vendor lock-in.

---

🌟 What's New in v1.0.0-rc.1

• 🧠 Universal Multi-Provider Support: Seamless out-of-the-box reasoning with Google Gemini 3.5, OpenRouter, Anthropic Claude 3.7, OpenAI GPT-4o / o3-mini, DeepSeek R1, Groq, and offline local engines (Ollama, LM Studio, vLLM).
• ⚡ High-Speed Vector RAG: Native SQLite-backed vector store powered by better-sqlite3 with IEEE 754 Float32 binary BLOB embeddings, automatic TTL pruning, and instant contextual document recall. Tested with Liquid / LFM 2.5 free embeddings.
• 🗣️ Local Hands-Free Voice Engine: Fully offline speech processing featuring Moonshine AI STT and ultra-natural Kokoro ONNX TTS with LaTeX algebraic voice synthesis.
• ⚙️ Sandboxed Code Execution: Bundled Python 3.11 runtime for isolated mathematical analysis, data visualization, and script execution without contaminating your host OS.
• 👤 Sovereign Personal Dossier: Automatic, local observation of your preferences, ongoing projects, and communication demeanor into a transparent USER_PROFILE.md file.
• 🛡️ Human-in-the-Loop Governance: High-risk actions (file modifications, shell commands, web queries) require explicit user approval with interactive visual diffs.
• 🎛️ Cognitive Thinking Knobs: Universal reasoning budget controls (off, low, medium, high, max) for deep chain-of-thought exploration across all models.

---

💻 Installation Instructions (Windows)

1. Download AI Plate Setup 1.0.0-rc.1.exe from the Assets section below.
2. Run the installer and follow the setup wizard (creates a Desktop shortcut and Start Menu entry).
3. Launch AI Plate:
   • Add your preferred API key (Gemini, OpenRouter, OpenAI, etc.) in the Settings tab, or connect to your local Ollama / LM Studio endpoint.
   • Start chatting or give autonomous tasks to your butler!

---

📋 System Requirements

• OS: Windows 10 / 11 (64-bit)
• RAM: 4 GB minimum (8 GB recommended)
• Disk Space: ~250 MB for app installation

---

Open Power to All. Sovereign AI for everyone.