{% extends "base.html" %}

{% block main %}
<section>
    <h2>LLM AI Assistant Support</h2>
    <div>
        Sketchingpy supports <a href="https://llms-txt.io/">llms.txt</a>, an open standard for providing information to AI assistants. This guide explains how to use AI assistants effectively with Sketchingpy.
    </div>
    <section class="subsection" id="optional">
        <h3>AI Assistant Use is Optional</h3>
        <div>
            Using an AI assistant with Sketchingpy is completely optional. The library is designed to be intuitive and well-documented for direct human use. AI assistants are simply an additional tool that some developers may find helpful for learning or accelerating their creative coding workflow.
        </div>
    </section>
    <section class="subsection" id="llms-txt-support">
        <h3>llms.txt Standard Support</h3>
        <div>
            Sketchingpy implements the <a href="https://llms-txt.io/">llms.txt standard</a>, which provides a structured way for AI assistants to understand the project's documentation and capabilities. This standard helps AI systems provide more accurate and helpful responses about Sketchingpy.

            <p>The project provides two key files:</p>
            <ul>
                <li><strong>llms.txt</strong> - A navigation overview of available documentation and resources</li>
                <li><strong>llms-full.txt</strong> - Complete API documentation and installation instructions</li>
            </ul>

            <p>If the AI assistant encounters issues with the more compact llms-full.txt file, you can also direct it to the more comprehensive <strong>reference.md</strong> file available at <code>https://sketchingpy.org/reference.md</code>, which contains the complete API reference with detailed examples and explanations.</p>

        </div>
    </section>
    <section class="subsection" id="directing-ai">
        <h3>Directing AI Assistants</h3>
        <div>
            To get the best help from an AI assistant when working with Sketchingpy, direct it to read the llms.txt and/or llms-full.txt files before asking for help. This ensures the AI has current and accurate information about the library.

            <p>Here's an example prompt to get started:</p>

<pre>
Hello! I would like help with working in Sketchingpy. Please read
https://sketchingpy.org/llms-full.txt?v=20250915 and
https://sketchingpy.org/llms.txt?v=20250915 to learn more about
the Python package.
</pre>

            <p>The version number (v=20250915) is optional but recommended as it helps ensure the AI reads the most current version of the documentation.</p>

            <p>For AI assistants that do not have internet access, you can provide these documentation files as attachments:</p>
            <ul>
                <li><a href="https://sketchingpy.org/llms.txt" download>llms.txt</a> - Navigation overview of documentation and resources</li>
                <li><a href="https://sketchingpy.org/llms-full.txt" download>llms-full.txt</a> - Complete API documentation and installation instructions</li>
                <li><a href="https://sketchingpy.org/reference.md" download>reference.md</a> - Most comprehensive API reference with detailed examples</li>
            </ul>
            <p>Simply download and attach these files when working with offline AI assistants to ensure they have access to complete Sketchingpy documentation.</p>
        </div>
    </section>
    <section class="subsection" id="example-prompts">
        <h3>Example Prompts</h3>
        <div>
            Once the AI has read the documentation, you can ask for help with specific tasks. Here are some example prompts:

<pre>
Please make a simple sketch which is 500 by 500 pixels with a
background of #F0F0F0. Then, in its center, please draw an ellipse
with radius of 20 by 20 pixels at the center. This ellipse should
have a fill of #C0C0C0 and a stroke of #000000.
</pre>

            This type of prompt should result in code like:

<pre>
import sketchingpy

sketch = sketchingpy.Sketch2D(500, 500)

sketch.clear('#F0F0F0')

sketch.set_fill('#C0C0C0')
sketch.set_stroke('#000000')
sketch.set_ellipse_mode('radius')
sketch.draw_ellipse(250, 250, 20, 20)

sketch.show()
</pre>

            <p>Other helpful prompts might include:</p>
            <ul>
                <li>"Help me create an animated bouncing ball sketch"</li>
                <li>"Show me how to load and display an image in Sketchingpy"</li>
                <li>"Create a sketch that responds to mouse clicks"</li>
                <li>"Help me understand how to use layers in Sketchingpy"</li>
            </ul>
        </div>
    </section>
    <section class="subsection" id="version-numbers">
        <h3>Version Numbers</h3>
        <div>
            When referencing the llms.txt or llms-full.txt files, including a version number as a query parameter is optional but recommended. The version number helps ensure you're getting the most current documentation and can help with cache management.

            <p>Format: <code>?v=YYYYMMDD</code></p>
            <p>Example: <code>https://sketchingpy.org/llms-full.txt?v=20250915</code></p>

            <p>The version number typically corresponds to the date the documentation was last updated.</p>
        </div>
    </section>
    <section class="subsection" id="best-practices">
        <h3>Best Practices</h3>
        <div>
            <ul>
                <li>Always direct the AI to read the llms-full.txt file first for comprehensive help</li>
                <li>Be specific in your requests - describe the visual outcome you want</li>
                <li>Include details about canvas size, colors, and positioning when relevant</li>
                <li>Ask for explanations if you want to understand the code, not just get working examples</li>
                <li>Remember that AI-generated code should be reviewed and tested before use</li>
            </ul>
        </div>
    </section>
</section>
{% endblock %}