DearAI - AutoKitteh AI Instructions

This repository contains tools and data for training and running an AI assistant that helps users with AutoKitteh integration development.

Usage

$ make

This produces the data, dearai, and digests directories.

Directory Structure

dearai/

This directory is generated by running make dearai.

The processed and cleaned dataset ready for AI training/inference:

• integrations/ - Processed integration documentation (Markdown files)
• kittehub/ - Cleaned integration examples with metadata only
• pyak/ - AutoKitteh Python SDK with function bodies abbreviated to ...
• Other processed documentation files

This directory contains:

• Abbreviated AutoKitteh Python SDK code (function signatures with ... bodies)
• Integration metadata extracted from the data directory
• Cleaned documentation optimized for AI consumption

digests/

This directory is generated by running make digests.

Contains various combinations of the dearai/ directory in single files for easily including in chatbots.

data/

This directory is populated by running make data.

Contains raw source data downloaded from AutoKitteh repositories:

• kittehub-main/ - Downloaded integration examples and templates from the KitteHub repository
• autokitteh-main/ - Core AutoKitteh codebase and Python SDK
• docs.autokitteh.com/ - Official AutoKitteh documentation
• integrations.txt - List of allowed integrations extracted from metadata
• requirements.txt - Python dependencies for the SDK

src/

Contains source templates and configuration files used to generate the final dearai/ dataset:

• Template files - Jinja2 templates that are processed to generate documentation and configuration files
• pyak/ - Source version of the AutoKitteh Python SDK before abbreviation processing
• Integration templates - Source templates for integration documentation that get processed into the final format

Files in this directory are processed by uvx jinja2 during the build process (excluding files starting with _ or in _* directories) to generate the corresponding files in dearai/. This allows for dynamic content generation and templating of the final AI training dataset.

The src/ directory serves as the source of truth for content that needs to be templated or processed before being included in the final dearai/ output.