Minimal Claude Code style harness workspace, modified with Claude Code using Opus.

• Primary authoring agent: Claude Code
• Model: Opus
• Repository: az9713/claw-code-tutorial
• Focus: tutorial-oriented variant and Claude-led modifications
• Status: maintain according to this repo's current branch state

This repository documents and demonstrates a Claude/Opus-driven modification path for a Claude Code style harness workspace.

It is designed for:

• tutorial workflows,
• Claude-led coding iterations,
• learning by progressive changes.

This is one of two sibling variants:

• claw-code-tutorial: Claude/Opus-authored path
• claw-code-by-codex: Codex/gpt-5.3-codex-authored path

Use both to compare model/operator style, architecture decisions, and implementation choices.

Cross-link: https://github.com/az9713/claw-code-by-codex

1. Identity block (authoring agent/model/date).
2. Exact quickstart commands that currently work.
3. Major architecture notes for this repo only.
4. Differences versus sibling repo.
5. Validation status (tests/checks).

Replace with commands that match this repository's current implementation.

python -m src.main summary
python -m unittest discover -s tests -v

Recommended:

• add a docs/ portal similar to the Codex repo,
• include a documentation-governance.md file,
• require docs updates in each behavior-changing PR.

• This repository is an educational/tutorial reconstruction workspace.
• It is not an official Anthropic repository and is not a full proprietary runtime replacement.