Vector OS Nano

Cross-embodiment robot OS: natural language controls everything.
No training. No fine-tuning. Just say what you want.
The agent decomposes any NL goal into a verified plan and executes long-chain tasks end-to-end.

Being developed at CMU Robotics Institute.

---

Demo

Click to watch full demo video

---

Quick Start

**Install** (requires Python 3.10+ and [uv](https://docs.astral.sh/uv/)):

git clone https://github.com/VectorRobotics/vector-os-nano.git
cd vector-os-nano
uv venv .venv
source .venv/bin/activate
uv pip install -e ".[all]"

**Configure API key** — copy the example and add your key:

cp .env.example .env
# then open .env and fill in one provider key:
#   DEEPSEEK_API_KEY=sk-...      # default (deepseek-v4-flash)
#   OPENROUTER_API_KEY=sk-or-... # multi-model fallback
#   ANTHROPIC_API_KEY=sk-ant-... # direct Anthropic

`.env` is git-ignored. One provider key is enough. DeepSeek `deepseek-v4-flash` is the default; OpenRouter is the multi-model fallback.

**Run:**

vector-cli                  # interactive AI agent REPL
vector-cli