Nanobot Setup and First Run
1. What installation actually gives you​
The upstream repo currently offers three practical install paths:
uv tool install nanobot-ai
python -m pip install nanobot-ai
git clone https://github.com/HKUDS/nanobot.git
cd nanobot
python -m pip install -e .
There is also a one-command installer in the repo's scripts/ directory for people who want the project-managed bootstrap path.
2. Verify the install​
After installation, the simplest health check is:
nanobot --version
That confirms the CLI is on your path before you start debugging provider keys or workflow config.
3. Which install path to choose​
| Path | Best when |
|---|---|
uv tool install nanobot-ai | You want the cleanest local CLI install |
pip install nanobot-ai | You already manage Python environments yourself |
| editable source install | You want to read or change the code right away |
If pip complains about an externally managed environment, the upstream guidance is to use uv, pipx, or a dedicated virtual environment instead of forcing a system install.
4. Docker and service-style setups​
The repo also ships Dockerfile, docker-compose.yml, and entrypoint scripts. That is a strong hint that the maintainers expect two operating modes:
- a normal local developer install,
- a more reproducible containerized setup for shared or hosted runs.
For evaluation work, start local first. For team rollouts or repeatable demos, move to Docker after the basic agent loop works.
5. Practical first-run advice​
A good first-run sequence is:
- install with
uv, - confirm
nanobot --version, - read the overview and architecture pages once,
- only then add providers, tools, or UI surfaces.
That order keeps "install trouble" separate from "agent behavior trouble."