Gemma documentation

The Google DeepMind Gemma documentation, ported to Zensical.

Local development

The project uses uv and pins the documentation toolchain in uv.lock.

uv sync --frozen
uv run --frozen python scripts/agent_docs.py
uv run --frozen zensical serve

The preview is available at http://localhost:8000/. Zensical's preview server may download raw .md routes because of its local MIME mapping; GitHub Pages serves those static files with their Markdown content type.

Verification

Run these checks before committing documentation or styling changes:

uv run --frozen python -m unittest discover -s tests -p "test_*.py"
uv run --frozen python scripts/lint_docs.py --baseline
uv run --frozen python scripts/build_docs.py

build_docs.py runs the normal Zensical build and then publishes raw Markdown routes such as /get-started/index.md. The committed docs/llms.txt is regenerated from page frontmatter and verified by the generated lint check.

See PLAN.md for the migration roadmap and docs/contributing/components.md for the CSS tokens and reusable components.