Content reference

Every file in content/ is validated before it is used. An unknown key is a build error naming the file and the field, not a silently missing line on your site.

Shared conventions first, then one section per file.

---

Shared field types

Dates

Three precisions, written as you would in a CV:

date: 2024                  # a year
date: 2024-03               # a month
date: 2024-03-15            # a day
date: { from: 2019, to: 2023 }
date: { from: 2023, to: present }     # open-ended

Precisions mix freely and still sort correctly. 2024-13 or 2023-02-30 are errors, as is a range that ends before it starts.

Entries are sorted newest first automatically. Formatting follows the precision you wrote: 2024, Mar 2024, 2019–2023, 2023–present.

Institutions

institution: University of Graz
institution: { name: University of Graz, url: https://uni-graz.at }

The short form is expanded for you; with a url, the name is linked in both outputs.

Links

links:
  - https://example.org/a # label derived from the URL
  - { label: ORCID, url: https://orcid.org/… }

URLs

A field of type url takes an http(s) or mailto URL, or a path on your site such as /slides/talk.pdf — a file in public/:

url: https://aofa.org/2025
slides: /slides/aofa2025.pdf # public/slides/aofa2025.pdf

On the website, a path gets the deployment base path, like paths in Markdown. The PDF has no site to be relative to, so where it prints such a link it needs your site's address, url in content/profile.yml; without it, build:cv stops and says so. Links the PDF does not print, such as slides of a talk, need nothing.

Text with mathematics

Fields marked rich below accept Markdown and $…$ mathematics:

title: The shape of large random $k$-ary trees

On the website it reads: The shape of large random $k$-ary trees.

They are typeset by KaTeX for the website and passed through to LaTeX for the PDF. Invalid TeX fails the build with the file, the entry index and the field.

Visibility

Every section entry accepts:

visibility: public          # default — website and PDF
visibility: cv              # PDF only: address, phone, referees, full lists
visibility: private         # local builds only

The website loader is hard-wired to public; vitakit-check fails if text from a non-public entry appears anywhere in the generated site. See privacy.

---

content/profile.yml

The only file that is not a CV section: who you are, plus the shape of the site.

Field	Type	Required	Notes
name	text	yes	Site header, page titles and CV header
title	text	no	e.g. Postdoctoral Researcher
affiliation	institution	no
email	email	no
links	list of links	no	ORCID, GitHub, arXiv, …
summary	text	no	One or two sentences for metadata; not rendered on the page — prose belongs in content/pages/index.md
url	http(s) URL	no	Where the site is published, e.g. https://jane.github.io/site/. The PDF needs it to link paths on the site
avatar	path	no	Below public/, e.g. /img/portrait.jpg
nav	list of { label, to }	no	Navigation; explicit, so wording and order are yours
footer	text	no	Inline Markdown, shown at the bottom of every page

Navigation is not derived from your pages: adding a page does not change the menu.

content/pages/*.md

The file name is the route: research.md → /research/, index.md → /.

---
title: Research
description: Analytic combinatorics.
order: 10
---

Prose, $mathematics$, tables and images.

Front matter	Type	Required	Notes
title	rich	yes
description	text	no	Meta description and page lead
route	path	no	Override the file-name route, e.g. /about/me/
order	integer	no	Sort order; defaults to 100

Body Markdown supports GFM (tables, strikethrough, task lists), $inline$ and $$display$$ mathematics, and images from public/. Write paths as if the site were served from /: root-relative links and assets such as [CV](/cv/) or /img/tree.svg are rewritten for the deployment base URL, so they work both locally and under user.github.io/repo/. That covers Markdown links and images as well as raw HTML — href, src, srcset, poster, <object data> and form actions. Relative paths (img/tree.svg) and full URLs are left alone.

content/education.yml

Field	Type	Required
date	date	yes
degree	text	yes
institution	institution	yes
thesis.title	rich	—
thesis.type	text	no
thesis.supervisor	text or link	no
thesis.url	url	no
note	text	no

- date: { from: 2016, to: 2020 }
  degree: PhD in Mathematics
  institution: { name: University of Klagenfurt, url: https://www.aau.at }
  thesis:
    title: Cutting down random trees
    type: Doctoral thesis
    supervisor: { label: A. Supervisor, url: https://example.org/s }

content/employment.yml

Field	Type	Required
date	date	yes
position	text	yes
institution	institution	yes
project.title	rich	—
project.funder, project.grant, project.url	text / text / url	no
note	text	no

content/teaching.yml

Field	Type	Required	Notes
date	date	yes
title	rich	yes
role	text	no	Lecturer, Teaching Assistant, …
institution	institution	no
term	text	no	2024W, Winter 2023/24; shown instead of the date
url	url	no
note	text	no

content/talks.yml

Field	Type	Required	Notes
date	date	yes
title	rich	yes
event	text	no
location	text	no
kind	enum	no	invited, contributed, seminar, poster, outreach
url, slides	url	no

kind is what document selection uses: where: { kind: invited }. See CV documents.

content/supervision.yml

| Field | Type | Required | | -------------- | ----------- | -------- | ---------------------------------------------------------- | | date | date | yes | | student | text | yes | | title | rich | yes | | ongoing | boolean | no | true for a thesis still in progress; defaults to false | | degree | text | no | | institution | institution | no | | coSupervisor | text | no |

content/publications.bib

A single BibTeX file, exported from Zotero, JabRef or arXiv.

@article{doe2024,
  author  = {Doe, Jane},
  title   = {The shape of large random $k$-ary trees},
  journal = {Random Structures \& Algorithms},
  year    = {2024},
  doi     = {10.5555/12345678},
  status  = {published},
}

Field	Effect
status	Grouping: published, accepted, preprint, software, other, in any capitalisation. Defaults to published; an unknown value is an error. The groups can be changed, see below
doi, arxiv, mrnumber, zbl	Become links after the reference, and arxiv becomes an eprint link in the PDF
eprint with eprinttype = {arxiv} or archivePrefix = {arXiv}	The same as arxiv, as biblatex, JabRef, Zotero and arXiv's own export write it
url	Links the title on the website. A path on the site such as /papers/doe2024.pdf (a file in public/) gets the deployment base path, and in the PDF becomes a full URL below url in profile.yml, which it then requires
year, month, day	Order within a group: newest first, by month and day where given; an entry with only a year comes after the dated ones of that year

A doi may be bare (10.1000/xyz), prefixed (doi:10.1000/xyz) or a full https://doi.org/… URL; all three link correctly in both outputs. On the website, arxiv, mrnumber and zbl may also carry the prefix exports write (arXiv:2301.00001, MR1234567, Zbl 1234.56789), and an arXiv version suffix (v2) is dropped from the link.

Mathematics in titles works in both outputs. The website formats references with a CSL style (a maths-oriented one by default); the PDF uses biblatex, so the two differ slightly in style — this is expected.

On the website, each title links to its url, else its doi, else its arxiv entry, and the fields above become badges after the reference. Those identifiers are not printed a second time as raw text, whatever the style.

Groups, links and fields

The groups, the badges after each reference and the fields passed to the website are set in schemas/publications.ts. The website, vitakit-check and the PDF all read it, so they always agree. Start from the defaults:

pnpm

pnpm vitakit eject publications   # -> schemas/publications.ts

npm

npx vitakit eject publications   # -> schemas/publications.ts

import { badgePresets, publications } from "@anja.codes/vitakit";

export default publications({
  groups: [
    { id: "published", label: "Articles" },
    { id: "preprint", label: "Preprints" },
    { id: "talk-slides", label: "Talk slides" }, // status = {talk-slides}
  ],
  defaultGroup: "published", // for entries without status
  badges: [
    { ...badgePresets.doi, className: "vita-badge" },
    {
      field: "slides",
      label: "Slides",
      url: "/slides/$1.pdf",
      className: "vita-badge",
    },
  ],
  fields: ["project"], // entry.fields.project
});

Every setting is optional; what you leave out keeps its default. Group ids are lowercase (talk-slides); the order of groups is the order on the page and in the PDF. A cv-tex/<document>.yml selects among them with publications.groups, and a misspelt group there is an error.

fields are BibTeX fields handed to the website as entry.fields, for a copy of VitaPublication that shows them (see customising).

Citation style

To change how references look on the website, put a CSL file at content/publications.csl. It is picked up without configuration — by the dev server (live), the build and vitakit-check alike. Start from the bundled style:

pnpm

pnpm vitakit eject publications.csl   # -> content/publications.csl

npm

npx vitakit eject publications.csl   # -> content/publications.csl

or save any style from the Zotero style repository under that name. Delete the file to return to the default. The PDF is not affected: it is typeset by biblatex, styled in cv-tex/.

---

Adding your own section

content/<name>.yml plus schemas/<name>.ts. See customising.