Onboarding

General fast way to understand the architecture and workflows is to follow this docs workflow:

ADR -> tools -> workflow directories

Step 1. Setup Your Environment¶

Run this scripts in your GNU/Linux terminal:

bash ./tools/scripts/configure_repo.sh

(in development) Configure JupyterLab for working with the files in this repo.

Step 2. Read Most Important Materials¶

Arthicture¶

1. “A Multi-Layered AI System Architecture” to understand the repo structure.

2. Series on how we work with Git in /tools/docs/git/.

3. ADRs in /architecture/adr to understand our choices and our stack.

Working with the Repository¶

1. Setup Jupytext and git hooks

2. “Production Git Workflow Standards”

Document Metadata Conventions¶

Every content article must include a YAML frontmatter block with MyST-native field names (title, author/authors, date) so that ownership and freshness are rendered directly on the static site. Custom fields (version, birth) go under options.* for RAG/tooling use.

1. “YAML Frontmatter for AI-Enabled Engineering” — the frontmatter schema and why it matters for RAG

2. ADR-26023 — the MyST-aligned frontmatter standard

Your AI Tools¶

1. “Multi-Phase AI Pipeline” - the Research-Apply pipeline for AI-assisted engineering.

2. “General Purpose (Abstract Synthesis) vs Agentic (Instruction Adherence) Models” - what models to use for different types of tasks.

3. Terminal AI Agents Workflow (to be created — hybrid strategy and Ollama setup)

Keep Your Work Clean¶

This repository contains the very different types of files - docs, configs, code. Before commit and push to the repository all you have done must be tested and validated for not breaking what is already built, like cross-links, scripts for internal use, etc.

The most of validation is done for you by automatic checks (or will be done for you as soon as possible) via pre-commit hooks and GitHub actions. But some validation should be done by you manually - it concerns publishing new .ipynb docs in the first hand.

Before commiting, run the MyST server locally from your terminal

uv run myst start

and:

• analyze the stdout messages for any errors,

• take a look at your new articles or modified files by opening localhost:3000 in your web browser.

Fix what you can fix or ask for help from the peers before Pull Request.