README: A Self-Documenting Artifact

Welcome to the Human-AI Workflow Kernel

This repository provides a configuration kernel for building self-documenting, Human-AI collaborative workflows. Instead of complex scripts, you define content generation and artifact assembly using structured JSON files, guided by the schemas in the schemas/ directory.

This approach allows teams (including AI assistants) to reproducibly generate various artifacts like documentation, tests, or even code scaffolding.

Key Documents:

• This README: Explains its own generation as an example.
• KERNEL.md: Describes the core concepts, schemas, and how to extend the kernel.

How This README is Generated (Example Workflow)

This README.md file serves as a simple demonstration of the workflow:

1. Content Definition (configs/content/readme/readme-content.json): This file defines this introductory section and the section you are currently reading. It uses the content-schema.json schema and includes descriptions, example outputs (used for this demo), and fields for tracking Human-AI collaboration (generation_source, review_status).

2. Artifact Definition (configs/artifacts/readme-artifact.json): This file, based on artifact-schema.json, orchestrates the assembly. It points to the readme-content.json above, specifies how to retrieve the content (retrievalStrategy: latest), how to combine sections (compositionStrategy: concat), and defines the final output file (README.md). It also marks the content as requiring review (review_required: true).

3. Generation Simulation: In this repository (which focuses on the kernel definition), we simulate the generation by concatenating the example_output fields from the content config. A full implementation would involve a tool that reads the artifact config, resolves content based on the strategy, potentially uses AI based on prompt_guidance if no pre-generated content exists, performs validation based on schema rules, and writes the final README.md.