Native Cross-Platform AI Operating Layer and Desktop Companion Core.
Sentinel is a modular, native AI operating layer and desktop companion built from the ground up in C++20 and Qt 6 / QML. Engineered for absolute privacy, offline autonomy, and low resource overhead, Sentinel executes local streaming LLMs, manages isolated workspaces, indexes local knowledge bases via RAG, and orchestrates user-approved autonomous agent workflows — with zero telemetry, zero hidden cloud calls, and no unapproved background execution.
- Local AI Execution & Streaming: Real-time streaming LLM chat powered by local loopback engines (Ollama, LM Studio, llama.cpp) and user-configured explicit cloud providers.
- Isolated Workspaces & Memory Architecture: Strict context partitioning across projects (Personal, Engineering, Academic, Research) backed by dedicated local SQLite stores (
memory.sqlite3,chat_history.sqlite3,local_rag.sqlite3). - Explainable Local RAG: Offline document and vector retrieval for PDF, Markdown, TXT, CSV, JSON, and source code files, featuring dynamic context budget allocation and source attribution.
- Controlled Foreground Agent Workflows: Safe, gated tool execution with interactive plan step editing, skip/retry actions, sandbox verification, and explicit human approval gates.
- Native Liquid Glass Desktop Shell: Hardware-accelerated translucent matte interface in Qt 6 / QML, tailored for Linux (Fedora KDE Plasma / Wayland), macOS, and Windows.
- Multi-Client Architecture & CLI Dispatcher: Unified companion suite offering the graphical desktop shell (
sentinel-desktop), headless scriptable CLI (sentinel-cli), and background daemon (sentinel-daemon).
When launched without arguments, sentinel-desktop opens the graphical Liquid Glass companion interface. For headless administration, terminal interaction, and scripting, use the CLI commands:
| Command / Shortcut | Description |
|---|---|
sentinel-desktop | Launches the native Liquid Glass desktop companion shell |
sentinel-cli chat "<prompt>" | Executes a one-shot or interactive terminal chat query |
sentinel-cli status [--json] | Displays health status of local AI providers, stores, and active workspace |
sentinel-cli models list | Enumerates available local and configured LLM models |
sentinel-cli workspace switch <id> | Switches active workspace context and memory scope |
sentinel-cli rag query "<text>" | Performs semantic search across the local document knowledge base |
sentinel-daemon | Runs the headless background companion and IPC coordination service |
Ctrl+K / Cmd+K | Opens the universal Command Palette within the desktop shell |
Ctrl+N / Cmd+N | Creates a new conversation thread |
Ctrl+, / Cmd+, | Opens Workspace & Model Settings |
Ctrl+P / Cmd+P | Toggles pinned status for the active conversation |
Ctrl+B / Cmd+B | Toggles conversation sidebar visibility |
Ctrl+Shift+T / Cmd+Shift+T | Toggles Dark / Light Liquid Glass theme |
Comprehensive guides for administrators, maintainers, and developers:
- 📦 Installation & Deployment Guide: System requirements, pre-built package deployment (RPM, DEB, AppImage, DMG, MSI), and local AI provider setup (Ollama / LM Studio).
- 🛠️ Build & Development Guide: Toolchain prerequisites, compiling from source (CMake / Ninja / Qt Creator), running CTest suites, and packaging workflows.
- 🏗️ System Architecture & Specification: Subsystem architecture, C++ core interfaces, SQLite persistence, view models, and security boundaries.
- 🤝 Contributing Guidelines: Development workflow, pull request guidelines, and code standards.
- 🔒 Security Policy: Vulnerability reporting, explicit consent model, and sandbox authority rules.
Sentinel is open-source software licensed under the GNU General Public License v3.0 or later (GPL-3.0-or-later). See LICENSE for details.