sphinx-likec4

Embed interactive LikeC4 architecture views — including dynamic view sequence diagrams — in Sphinx documentation.

The extension runs the pinned LikeC4 CLI for you (node ≥ 20 required at doc-build time), caches the build on a content hash, validates every embedded view id at build time (-W friendly), and serves the viewer with hash routing so it works from any subpath.

Agent skill

An Agent Skill documenting the directive surface ships in skills/:

npx skills add ckeller42/sphinx-likec4      # installs into your agent's skills dir

An LLM-oriented summary of the whole project lives at /llms.txt.

Quickstart

pip install sphinx-likec4

# conf.py
extensions = ["sphinx_likec4"]
likec4_source_dir = "model"

.. likec4-view:: cloud-to-amazon

Static images and PDF

Views can also be embedded as static PNG/JPG images — per directive, per output format, or by default:

.. likec4-view:: cloud-to-amazon
   :render: png          # static image on an HTML page (iframe is the default there)
   :mode: sequence       # dynamic views: sequence layout, in the iframe and in the image
   :width: 80%

LaTeX/PDF and epub builds do this automatically for every view — sphinx-build -M latexpdf just works, Chromium (via Playwright) is installed once on first use. Configure the default per output format with likec4_render = {"latex": "jpg"}.

Example: the PDF rendering of these docs — every diagram in it, including the sequence view, is a static image produced this way.

Docs: https://ckeller42.github.io/sphinx-likec4/ · License: MIT