The content model

Everything you write lives under content/, in typed collections. Typed means each collection publishes a schema, the build validates every file against it, and an unknown key is an error naming the file, not a silently ignored line. Ask the CLI, never guess:

$ cherry schema posts
posts frontmatter:
  title: string (required) — Post title.
  date: date — ISO 8601 date; defaults to the date in the filename.
  slug: string — URL slug; defaults to the filename after the date.
  tags: list of string [default: []] — Tags from the shared site taxonomy (also used by the portfolio).
  draft: boolean [default: false] — Drafts are skipped unless `--drafts`.
  description: string — Meta description for SEO and feeds.

Posts

content/posts/YYYY-MM-DD-slug.md. The filename is the source of truth for the date and slug; frontmatter can override the slug. A post renders at /slug/, joins the blog index, its tags' pages, and both feeds.

The draft flow is built in. cherry gen.post "Title" scaffolds with draft: true; drafts appear in cherry serve (your writing loop) but never in cherry build unless you pass --drafts. When it is ready:

$ cherry publish my-draft

renames the file to today's date and removes the draft: line. Nothing else changes.

Pages

content/pages/*.md, freeform. about.md becomes /about/; nested directories nest the URL. A permalink: overrides the derived route, which is how a section index gets a clean URL:

---
title: Docs
permalink: /docs/
---

The portfolio

Five more collections carry your developer story, each a directory of markdown files under content/portfolio/: positions, projects, talks, oss, and education. A portfolio.yaml at the site root holds the profile (name, headline, location, links). A position looks like this:

---
title: Staff Engineer
org: Orchard Systems
start: 2020-02-01
location: Remote
tags:
  - elixir
highlights:
  - Grew the platform from seed to fruit
cv:
  include: true
  weight: 10
---
The long-form story, in markdown, for the portfolio page.

Those collections render three surfaces for free:

• /portfolio/: a dated timeline of everything, with the profile header.
• Story pages at /story/TAG/: everything sharing a tag, portfolio entries and blog posts together. Tags are one taxonomy across the whole site.
• /cv/: the employer-shaped page. The cv: block is the curation layer: include opts an entry in, weight orders it, and an optional cv.highlights list overrides the bullets with a tighter cut. Print it and the stylesheet produces a clean one-pager; cv.json ships beside it in JSON Resume format.

The portfolio guide builds all of this from scratch.

Markdown

GitHub-flavoured throughout: tables, strikethrough, task lists, footnotes, autolinks. Heading anchors are automatic. Syntax highlighting is class-based and coloured by theme tokens, so code blocks follow your palette in both renditions. GitHub-style alerts work as blockquotes (> [!NOTE]), and the richer content components cover figures, video, and callouts. Raw HTML passes through, which is how this site's landing page is built.