Heliokit Documentation
======================

.. raw:: html

   <section class="hk-hero">
     <div class="hk-hero__content">
       <p class="hk-eyebrow">Space-weather data, models, and desktop workflows</p>
       <h1>Heliokit</h1>
       <p class="hk-lead">
         A Python toolkit for operational heliophysics workflows: component discovery,
         data readers, transformation filters, forecast models, pipelines, and a
         protected desktop distribution.
       </p>
       <div class="hk-actions">
         <a class="hk-button hk-button--primary" href="get-started.html">Get started</a>
         <a class="hk-button" href="desktop.html">Desktop package</a>
       </div>
     </div>
   </section>

.. raw:: html

   <div class="hk-card-grid">
     <a class="hk-card" href="get-started.html">
       <span class="hk-card__label">Install</span>
       <strong>Use the host package</strong>
       <p>Install Heliokit, inspect available components, and run data or model commands.</p>
     </a>
     <a class="hk-card" href="components.html">
       <span class="hk-card__label">Components</span>
       <strong>Compose workflows</strong>
       <p>Work with data readers, filters, models, tools, and pipelines without changing the desktop codebase.</p>
     </a>
     <a class="hk-card" href="api.html">
       <span class="hk-card__label">Python</span>
       <strong>Automate from scripts</strong>
       <p>Use the Heliokit API for discovery, package installation, and component execution.</p>
     </a>
     <a class="hk-card" href="reference.html">
       <span class="hk-card__label">Reference</span>
       <strong>Know every input</strong>
       <p>Check required arguments, accepted formats, defaults, and runtime constraints before a run.</p>
     </a>
   </div>

.. toctree::
   :maxdepth: 2
   :caption: Documentation

   get-started
   components
   demo-data
   reference
   api
   desktop
   developer-guide
   license

Why Heliokit
------------

Heliokit is designed as a Python-native distribution layer around existing
space-weather capabilities. The host package discovers installed components
through entry points, exposes a small Python API, and keeps command-line
workflows consistent across data, filters, models, tools, and pipelines.

The desktop package is delivered separately. It is intended for controlled
CPython 3.10 x64 environments where heavyweight scientific dependencies already
exist, so the wheel can stay compact while still importing VTK, Qt, TensorFlow,
Torch, and related runtime libraries from the environment.