"All of time and space; everything that ever happened or ever will - where do you want to start?"
Tardis OS is an experimental Operating System designed with Artificial Intelligence at its core. Unlike traditional OSs where AI is an application layer, Tardis embeds Large Language Models (LLMs) and Vector Databases deep into the system services, enabling semantic understanding of data, temporal queries, and natural language interaction.
Tardis is built as a Rust workspace containing several specialized crates:
| Crate | Role | Description |
|---|---|---|
kernel | 🧠 Core | UEFI bootloader and no_std kernel (requires Nightly). |
vortex | 🌪️ Inference | LLM inference engine using Candle. |
gallifrey | 🕰️ Memory | Temporal knowledge store with bi-temporal queries. |
chronos | ⚡ RAG | Orchestration layer bridging Vortex and Gallifrey. |
shell | 🐚 UI | Natural language shell with intent routing. |
telemetry | 📡 Observability | Full-stack tracing from Kernel to User space. |
common | 🔧 Utils | Shared types and error handling. |
Tardis uses feature flags to enable experimental capabilities:
nova: Enables experimental modules ("Nova" features) such as:
Enable it by running:
cargo run --features nova --bin tardis
qemu-system-x86_64 (for kernel testing).Build Userspace Components:
cargo build
Run the AI Shell:
cargo run --bin tardis
Run Tests:
cargo test
Build Documentation:
cargo doc --open
We welcome contributions!
For Developers & AI Agents: Please read CLAUDE.md first. It contains critical information about the workspace structure, coding standards, and the "worktree workflow" we use.
git checkout -b feature/amazing-feature).MIT OR Apache-2.0
156 commits
5 commits
Rust
98.7%
"All of time and space; everything that ever happened or ever will - where do you want to start?"
Tardis OS is an experimental Operating System designed with Artificial Intelligence at its core. Unlike traditional OSs where AI is an application layer, Tardis embeds Large Language Models (LLMs) and Vector Databases deep into the system services, enabling semantic understanding of data, temporal queries, and natural language interaction.
Tardis is built as a Rust workspace containing several specialized crates:
| Crate | Role | Description |
|---|---|---|
kernel | 🧠 Core | UEFI bootloader and no_std kernel (requires Nightly). |
vortex | 🌪️ Inference | LLM inference engine using Candle. |
gallifrey | 🕰️ Memory | Temporal knowledge store with bi-temporal queries. |
chronos | ⚡ RAG | Orchestration layer bridging Vortex and Gallifrey. |
shell | 🐚 UI | Natural language shell with intent routing. |
telemetry | 📡 Observability | Full-stack tracing from Kernel to User space. |
common | 🔧 Utils | Shared types and error handling. |
Tardis uses feature flags to enable experimental capabilities:
nova: Enables experimental modules ("Nova" features) such as:
Enable it by running:
cargo run --features nova --bin tardis
qemu-system-x86_64 (for kernel testing).Build Userspace Components:
cargo build
Run the AI Shell:
cargo run --bin tardis
Run Tests:
cargo test
Build Documentation:
cargo doc --open
We welcome contributions!
For Developers & AI Agents: Please read CLAUDE.md first. It contains critical information about the workspace structure, coding standards, and the "worktree workflow" we use.
git checkout -b feature/amazing-feature).MIT OR Apache-2.0
156 commits
5 commits
Rust
98.7%