para.site

A pandoc-first static site generator

para.site builds websites from documents pandoc can read: Markdown, Jupyter notebooks, and anything else pandoc has a reader for. Its documents, metadata and templates are pandoc’s own. It adds what a site needs around them: listings, categories and tags, an archive, feeds, a sitemap, search, a graph of how the pages link, and builds that redo only what changed.

The Python package is para (on PyPI as pandoc-para for now), the site generator is its module para.site, and the command is parasite:

from para import site

Why

  • pandoc first. No syntax of its own: no shortcodes, no extensions to Markdown. A source is a pandoc document; its options are pandoc’s options.
  • Configured as data. A site is described by one file, parasite.toml. Code runs only as filters you named, from packages you installed.
  • One source, several outputs. A talk can be a page and its slides, from one file (see output profiles).
  • Incremental. A build reads and renders only what changed; a rebuild with nothing changed takes a fraction of a second (see how it builds).
  • Valid XHTML. Every page is well-formed.

A site in a minute

A folder with a parasite.toml:

[site]
url = "https://example.org/"
title = "My site"
author = "Me"

and Markdown files under posts/ (dated, listed) and pages/ (not listed):

---
title: Hello
date: 2026-10-11
tags: [hello]
---

The first post.

Then, in that folder:

parasite

writes the site to output/. parasite --help lists what the command line can override.

These pages are themselves built with para.site: see docs/ in its repository.