Skip to content

EXTENDING.md

Latest commit

 

History

History
169 lines (126 loc) · 4.99 KB

EXTENDING.md

File metadata and controls

169 lines (126 loc) · 4.99 KB

Extending this CV

Quick recipes for common changes. For an agent-focused guide, see CLAUDE.md.

Add a section by typing it into the YAML (no code needed)

Switch the editor to YAML view and add a new top-level key:

awards:
  - title: Best Paper Award
    date: 2024
    desc: ICML 2024 Honourable Mention
  - title: Germany Scholarship
    date: 2021
    desc: Deutschlandstipendium

Hit ⌘S. Three things happen:

  1. The YAML is auto-linted — formatting normalised, header comment preserved.
  2. The sidebar grows an "Awards" entry (with a green progress dot).
  3. The Form view picks up the section. The PDF preview re-renders with the new heading.

The shape (compact / experience / publication / etc.) is inferred from the first item's keys:

Item shape Inferred
{title, date, desc} compact (one-liner with right-aligned date)
{role, company, start, end, bullets} experience
{degree, school, start, end} education
{title, authors, venue, date} publication
{label, items} skills (definition-list)

The label is title-cased from the key (speaking_engagements → "Speaking Engagements").

When you'd reach for the registry instead

Editing cv.yaml is the right move for one-off sections you'd put on your own CV. If you want to ship a section as a default for everyone who clones the repo (with a stable order, custom eyebrow, importer aliases, etc.) — register it in code:

Add a "Publications" section in 30 seconds (registered)

Open engine/render/sections.py, append to DEFAULT_SECTIONS:

SectionDef(
    key="publications",
    label="Publications",
    eyebrow="Section 08 · Papers",
    singular="paper",
    shape="publication",
    required_fields=("title",),
    rendercv_aliases=("publications", "papers"),
    text_header_pattern=r"^\s*(papers|publications|talks)\s*$",
),

Restart the editor. The form now shows a Publications section. The PDF renders it. The Claude AI extractor knows about it. Your YAML file accepts:

publications:
  - title: "Some Paper"
    authors: "A. Author, B. Author, C. Hartman"
    venue: "ICLR 2024"
    date: "04/2024"
    doi: "10.5555/abc"
    url: "https://..."

Available shape values:

Shape Best for Required fields
experience jobs role, company, start, end
education degrees degree, school, start, end
skills category lists label
compact one-liners with date title
publication papers / talks / press title

To add a new shape entirely, see CLAUDE.md.

Switch to a different theme

The editor's Theme picker (palette icon in the topbar) lists all JSON files in themes/. Click one to apply.

To add a new one:

// themes/midnight.json
{
  "name": "Midnight",
  "accent": "#7C3AED",
  "font": "serif",
  "density": "normal"
}

Refresh the editor; the new theme appears in the picker.

Themes set:

  • accent — bullet dots, section rules, brand-icon colour, save button
  • font — serif (Source Serif 4) | sans (Inter) | mono
  • density — tight | normal | airy

Fork it and own your CV pipeline

The whole thing is a small Python package + a Flask server + a few hundred lines of CSS. Fork the repo, change the accent colour or the density numbers in engine/render/css_base.py, add a section in sections.py, add a theme in themes/. Everything load-bearing — the one-page guarantee, the form-first flow, the live preview — survives those changes.

To redistribute a custom theme: paste your themes/<name>.json into a GitHub gist and tell people to drop it into their themes/ folder.

Import from anywhere

Three paths in the editor's Import modal:

  1. AI extract — drop a PDF / DOCX / TXT / MD file (or paste text). Claude maps it into the schema. Set ANTHROPIC_API_KEY to enable; without a key, a regex fallback handles the basics.
  2. rendercv YAML — paste a rendercv YAML for a deterministic conversion.
  3. Plain text — paste any resume text; heuristic regex parser tries its best.

For LinkedIn: open your profile → More → Save to PDF → drop the file into the AI extract tab.

Common questions

My CV overflows. What do I do?

  1. Toggle density to tight (Density · Tight in the topbar)
  2. Trim adjectives — favour numbers (50%+, 100+ hours/week)
  3. Drop the oldest entry from a long-tail section
  4. Last resort: drop the stack: line on older roles

Can I change the section order? Yes — reorder entries in DEFAULT_SECTIONS in engine/render/sections.py. The order on disk is the order on the page.

Can I have two pages? This tool intentionally enforces one page. If you really need two, fork it, edit engine/build.py, and remove the page-count guard. You'll lose the live overflow indicator too — they're paired by design.

Where's my file? content/cv.yaml. The form view is just a UI on top.