Installation¶

IINTS-AF supports Python 3.10 through 3.14 on Windows, macOS, and Linux. Use a virtual environment so the SDK and its dependencies remain isolated from the system Python.

Choose An Install¶

Need	Install	Recommended for
Simulation, reports, and MDMP	`iints-sdk-python35[full,mdmp]`	most users
AI model training and interactive research plots	add `[research]`	research workstations and Jetson
Serial hardware interfaces	add `[edge]`	Raspberry Pi, Pico, UNO Q, FPGA bridges
Python/PySide desktop development	add `[desktop]`	desktop contributors
Complete desktop/research workbench	`[desktop-all]`	app users who want every supported Python-side engine
Latest source code	editable Git clone	contributors only

The packaged desktop application has its own installer. See Install The Desktop App.

Standard Installation¶

macOS And Linux¶

python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install "iints-sdk-python35[full,mdmp]"

Windows PowerShell¶

py -m venv .venv
.venv\Scripts\Activate.ps1
python -m pip install --upgrade pip
python -m pip install "iints-sdk-python35[full,mdmp]"

Verify the installation:

iints --version
iints doctor --smoke-run

Optional Profiles¶

Install only what the workflow needs.

Research And Glucose AI¶

python -m pip install "iints-sdk-python35[full,mdmp,research]"

This adds packages for model training, ONNX work, Parquet/HDF5 data, and interactive Plotly outputs. It does not download private diabetes datasets or pretrained models.

Edge Hardware¶

python -m pip install "iints-sdk-python35[edge,mdmp]"
iints edge doctor

Continue with the Hardware Hub and the guide for the selected device.

Local Ollama AI¶

The Python install does not install or start Ollama automatically. The desktop app can guide the setup; CLI users should follow AI Assistant.

Verify the backend only after Ollama and a model are available:

iints ai local-check

Install From Source¶

Use this route only when changing the SDK itself.

git clone https://github.com/python35/IINTS-SDK.git
cd IINTS-SDK
python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -e ".[dev,full,mdmp]"

Run the quick contributor checks:

tools/dev/sdk_check.sh quick

Important folder rule:

run repository scripts from the repository root
run scaffolded project commands from that project directory
do not run pip install -e . from your home folder or another directory without pyproject.toml

Tauri Desktop Development¶

The native Tauri shell requires Node.js and Rust/Cargo in addition to Python.

On macOS with Homebrew:

brew install node rust

Then:

cd apps/iints-tauri
npm install
npm run check
cargo check --manifest-path src-tauri/Cargo.toml

See Tauri Desktop Shell for architecture, permissions, and packaging.

Update An Existing Install¶

iints update --dry-run
iints update
iints doctor --smoke-run

The desktop app exposes the same update flow. Read Update The SDK before changing an installation used for a reproducible study.

Remove The SDK¶

Preview destructive cleanup first:

iints delete --dry-run

Only confirm deletion after reviewing every path. Research data and run bundles may be irreplaceable; archive them separately before removing software or local state.

Common Problems¶

`iints: command not found`¶

Confirm the virtual environment is active and inspect:

python -m pip show iints-sdk-python35
python -m pip --version

Use python -m iints.cli.cli only for debugging source imports; normal users should use the installed iints command.

Wrong Python Version¶

python --version

Use Python 3.10 through 3.14. Recreate the virtual environment after changing Python versions.

Missing Optional Feature¶

Install the relevant extra in the active environment. For example:

python -m pip install --upgrade "iints-sdk-python35[research]"

Desktop App Does Not Start¶

Do not install PySide6 separately into a packaged .app, .dmg, .exe, or Linux bundle. Runtime dependencies are included in the artifact. Follow Desktop App Installation and collect the launch log if it still fails.

Next¶

Run the First Run.
Read Core Concepts.
Use Troubleshooting for platform-specific failures.