Resumx

Resumx renders resumes from Markdown to PDF, HTML, PNG, and DOCX. It auto-fits content to a target page count, supports tags and views for tailored output from a single source file, and uses style options for styling.

Resources

This skill includes reference documents for specific workflows. Read them when applicable:

Resource When to use

json-resume-to-markdown.md Converting between Resumx Markdown and JSON Resume format (either direction)

writing-resume.md Interactive resume creation, guiding users step-by-step to collect info and generate a resume

tagging-resume.md Systematically tagging a resume for tailored output, including hierarchical tag taxonomy

Markdown Syntax

Standard Markdown with extensions for inline columns, bracketed spans, and fenced divs.

Structure

Element Syntax

Name

Full Name

Contact email@example.com | github.com/user | linkedin…

Section

Section Name

Entry

Title || Date

Subtitle Role or Degree || Location

Bullets

• Achievement with
  tech tags

Skills Definition list (Term

• : values )

Inline Columns

|| splits a line into columns, pushing them to opposite sides. Simplest way to right-align a date or location:

Google || Jan 2020 - Present

Senior Software Engineer || San Francisco, CA

More than two columns: A || B || C . Escape with || .

Inline Formatting

Syntax Result

Bold

Bold

Italic

Italic

Code

Tech tag/badge

==Highlight==

Highlighted text

H2O

Subscript

E = mc^2^

Superscript

-- / ---

En-dash / Em-dash

"text"

Smart quotes

Definition Lists

Languages : JavaScript, TypeScript, Python

Frameworks : React, Node.js, Express

Also used as inline metadata below entries:

Google || June 2022 - Present

Senior Software Engineer : Infrastructure Platform Team : San Francisco, CA

Tables

Horizontal Rule

renders as a standard thematic break or section divider.

Comments

HTML comments (<!-- --> ) are stripped from output.

Classes, IDs & Fenced Divs

Bracketed Spans

[text]{.class} wraps text in a <span> with classes/IDs/attributes:

Google [2022 - Present]{.right}

Senior Software Engineer [San Francisco, CA]{.right}

Element Attributes

{...} at end of a block element applies to the whole element:

• Built interactive dashboards {.@frontend}

Fenced Divs

::: applies attributes to block content. Single child: attributes go directly on it (no wrapper div). Multiple children: auto-wraps in a <div> .

::: {.grid .grid-cols-3}

• JavaScript
• TypeScript
• Python :::

Prefix a tag name for a specific HTML element: ::: footer {.text-center} .

Frontmatter

YAML (--- ) or TOML (+++ ). CLI flags always override frontmatter.

Render Fields

Field Type Default Description

css

string | string[]

None CSS file path(s) or inline CSS string(s). Ends with .css = file, otherwise = inline

output

string

Input filename stem Output path (name, directory with / , or template with {view} /{lang} /{format} )

pages

positive integer

No clamping Target page count

sections

{ hide?: string[], pin?: string[] }

All in source order Section visibility and ordering

bullet-order

none | tag

none

Bullet ordering strategy

style

Record<string, string>

No overrides Style option overrides

tags

Record<string, string[] | TagConfig>

No composed tags Tag composition and tag view configuration

vars

Record<string, string>

No variables Template variables for {{ }} placeholders

icons

Record<string, string>

No custom icons Custom icon definitions (SVG, URL, or base64)

extra

Record<string, unknown>

No custom data Arbitrary user-defined data

Validate Fields

Under a validate key:

Field Type Default

validate.extends

string

recommended

validate.rules

Record<string, Severity | 'off'>

Per-preset defaults

Presets: recommended , minimal , strict , none .

Available rules: missing-name , missing-contact , no-sections , no-entries , empty-bullet , long-bullet , single-bullet-section , unknown-fenced-div-tag , non-pt-font-size .

Unknown Fields

Unknown top-level keys error. Use extra for custom data.

Full Example

pages: 1 output: ./out/Jane_Smith-{view} bullet-order: tag style: link-color: '#0ea5e9' tags: fullstack: [frontend, backend] leadership: false vars: tagline: 'Full-stack engineer with 8 years of experience' validate: extends: recommended rules: long-bullet: warning single-bullet-section: off extra: name: Jane Smith target-role: Senior SWE

CLI

resumx <file> # Render (defaults to resume.md, PDF) resumx init [filename] # Create template resume

Sandbox Requirement

Resumx uses Playwright with Chromium for PDF rendering. Chromium cannot launch inside Cursor's default sandbox because the sandbox blocks syscalls Chromium needs to initialize. Always run resumx commands with required_permissions: ["all"] to disable the sandbox. Without this, rendering will fail with "Chromium not found" even though Chromium is installed.

Render Options

Flag Description

--css <path>

Path to custom CSS file, repeatable, comma-separated

-o, --output <value>

Output path (name, directory, or template)

-f, --format <name>

pdf , html , docx , png , repeatable, comma-separated

-s, --style <name=value>

Override style property, repeatable

--for <name-or-glob>

Tag view name, custom view name, glob pattern, or default for the default view

-v, --var <key=value>

Override a template variable, repeatable

--hide <list>

Hide sections (comma-separated data-section values)

--pin <list>

Pin sections to top in order (comma-separated data-section values)

--bullet-order <value>

Bullet ordering: none (default) or tag

-l, --lang <tag>

Language filter (BCP 47), repeatable, comma-separated

-p, --pages <number>

Target page count

-w, --watch

Auto-rebuild on changes

--check

Validate only, no render

--no-check

Skip validation

--strict

Fail on any validation error

--min-severity <level>

Filter validation output

Stdin

cat resume.md | resumx git show HEAD~3:resume.md | resumx -o old

Output Naming

Scenario Output

No view, no langs resume.pdf

With tag/view resume-frontend.pdf

With langs resume-en.pdf

Tag/view + langs frontend/resume-en.pdf

Template variables: {view} , {lang} , {format} .

Style Options

Override via frontmatter style: or CLI --style .

Typography: font-family , title-font-family , content-font-family , font-size (default 11pt ), line-height (default 1.4 ).

Colors: text-color , link-color , background-color .

Headings: name-size , name-caps (small-caps , all-small-caps , petite-caps , unicase , normal ), name-weight , name-italic , name-color , section-title-size , section-title-caps , section-title-weight , section-title-italic , section-title-color , section-title-border , header-align , section-title-align , entry-title-size , entry-title-weight , entry-title-italic .

Links: link-underline (underline , none ).

Spacing: gap (unitless scale factor for all vertical gaps), page-margin-x , page-margin-y , section-gap , entry-gap , row-gap , col-gap , list-indent .

Lists: bullet-style (disc , circle , square , none ).

Features: auto-icons (inline , none ).

Custom CSS

Your CSS cascades on top of the default stylesheet, so you only write overrides:

:root { --font-family: 'Inter', sans-serif; --section-title-color: #2563eb; }

h2 { letter-spacing: 0.05em; }

Reference by path: css: my-styles.css or --css my-styles.css . For small overrides, inline CSS avoids creating a file:

css: | h2 { letter-spacing: 0.05em; }

Mix file paths and inline CSS in an array:

css:

• base.css
• | h2::after { content: ''; flex: 1; border-bottom: var(--section-title-border); }

For building a stylesheet from scratch, @import bundled modules: common/base.css (reset, typography, layout), common/icons.css (icon sizing), common/utilities.css (.small-caps , .sr-only , .max-N ).

Fit to Page

Set pages: N to auto-fit. Shrinks in order of visual impact (least noticeable first):

• Gaps (row-gap, entry-gap, section-gap)

• Line height

• Font size

• Margins (page-margin-x, page-margin-y, last resort)

For pages: 1 , gaps also expand to fill remaining space.

Minimums: font-size 9pt, line-height 1.15, section-gap 4px, entry-gap 1px, page-margin-y 0.3in, page-margin-x 0.35in.

When pages: is set, style: values are starting points that may be reduced. Without pages: , they apply as-is.

Icons

Syntax

:icon-name: for built-in icons, :set/name: for Iconify (200k+ icons), standard emoji shortcodes as fallback.

Auto-Icons

Links to recognized domains get icons automatically:

mailto: (Email), tel: (Phone), linkedin.com , github.com , gitlab.com , bitbucket.org , stackoverflow.com , x.com /twitter.com , youtube.com /youtu.be , dribbble.com , behance.net , medium.com , dev.to , codepen.io , marketplace.visualstudio.com .

Disable with style: { auto-icons: none } .

Custom Icons (Frontmatter)

icons: mycompany: '<svg>...</svg>' partner: 'https://example.com/logo.svg'

Resolver order: Frontmatter > Built-in > Iconify > Emoji.

Tailwind CSS

Resumx compiles Tailwind CSS v4 on-the-fly. Apply via {.class} syntax:

[React]{.bg-blue-100 .text-blue-800 .px-2 .rounded}

Works with bracketed spans, element attributes, and fenced divs. Supports arbitrary values: .text-[#ff6600] .

Built-in utilities: .small-caps , .sr-only , .max-1 – .max-16 (hide children beyond the Nth).

Tags

Tags filter content. Add {.@name} to any element to mark it for a specific audience. Untagged content always passes through. Tagged content only appears when rendering for a matching tag.

• Shared bullet
• Frontend-only bullet {.@frontend}
• Backend-only bullet {.@backend}

Multiple tags: {.@backend .@frontend} .

Render with --for frontend or --for backend .

Hierarchical Tags

Use / to nest tags when a domain spans multiple ecosystems:

• Designed REST APIs with OpenAPI documentation {.@backend}
• Built microservices with Express {.@backend/node}
• Migrated Spring Boot monolith {.@backend/jvm}

Inheritance: a view includes its entire lineage (ancestors + self + descendants) and untagged content. Siblings are excluded.

• --for backend/node → @backend (ancestor) + @backend/node (self) + untagged. Excludes @backend/jvm .

• --for backend → @backend (self) + @backend/node

• @backend/jvm (descendants) + untagged.

Depth is unlimited: {.@data/ml/nlp} nests three levels.

Tag Composition

Define composed tags in frontmatter as unions of constituents:

tags: fullstack: [frontend, backend] node-fullstack: [frontend, backend/node] tech-lead: [backend, leadership] startup-cto: [fullstack, leadership, architecture]

Hierarchical tags work as constituents. Lineage expands per constituent: node-fullstack includes @frontend (+ descendants), @backend (ancestor of backend/node ), @backend/node , and untagged. Sibling @backend/jvm is excluded.

Compositions expand recursively (startup-cto includes frontend and backend via fullstack ). Every constituent must exist as a content tag or another composed tag; typos produce an error with a suggestion.

Views

Tags filter content (what to show). Views configure rendering (how to show it). Every render uses a view.

Four Kinds of View

Kind Where Nature

Default view Frontmatter render fields Base config for all renders

Tag view Frontmatter tags: expanded form Per-tag overrides, implicit for every tag

Custom view .view.yaml files Per-application config

Ephemeral view CLI flags One-off, not persisted

Tag Views

Every tag implicitly generates a tag view. Configure with the expanded form:

tags: frontend: sections: hide: [publications] pin: [skills, projects] pages: 1

fullstack: extends: [frontend, backend] sections: pin: [work, skills] pages: 2

The shorthand fullstack: [frontend, backend] is sugar for fullstack: { extends: [frontend, backend] } .

Custom Views

Custom views live in .view.yaml files, auto-discovered recursively relative to the resume:

stripe.view.yaml

stripe-swe: selects: [backend, distributed-systems, leadership] sections: hide: [publications] pin: [skills, work] vars: tagline: 'Stream Processing, Event-Driven Architecture, Go, Kafka'

Render with --for stripe-swe . Batch with --for '' or --for 'stripe-' . Use --for default to target the default view (no tag filtering); combine with named views, e.g. --for default,frontend , to render both. Do not name a view default (reserved).

Custom view fields: selects (content tags to include), sections , pages , bullet-order , vars , style , css , format , output .

A view without selects applies no content filter (all content renders). An explicit selects: [] means only untagged content.

Ephemeral Views

CLI flags create an ephemeral view inline without persisting:

resumx resume.md --for backend -v tagline="Stream Processing, Go" --pin skills,work -o stripe.pdf

Cascade Order

Built-in defaults → Default view (frontmatter render fields) → Tag view OR Custom view (whichever --for resolves) → Ephemeral view (CLI flags)

Template Variables

Inject per-application text via {{ name }} placeholders:

Jane Doe

jane@example.com | github.com/jane

{{ tagline }}

Define in frontmatter vars: , in a custom view, or via CLI -v :

vars: tagline: 'Full-stack engineer with 8 years of experience'

When undefined or empty, the placeholder produces nothing (line removed). Variable values can contain markdown formatting. Defining a variable with no matching placeholder is an error.

Sections

Control which sections appear and their order:

sections: hide: [publications, volunteer] pin: [skills, work]

hide removes sections. pin moves them to the top in the specified order. Non-pinned sections follow in source order. Values are data-section types: work , education , skills , projects , awards , certificates , publications , volunteer , languages , interests , references , basics .

CLI: --hide publications --pin skills,work .

Multi-Language Output

Tag content with {lang=xx} (BCP 47). Untagged content appears in all languages.

[Experience]{lang=en} [Expérience]{lang=fr}

Google

• [Reduced API latency by 60%]{lang=en} [Réduction de la latence API de 60%]{lang=fr}

Combines with tags: {lang=en .@backend} . Filter with --lang en or --lang en,fr .

Dimensions multiply: 2 langs × 2 tags = 4 PDFs.

Semantic Selectors

Resumx auto-generates semantic HTML attributes for CSS targeting:

Header: [data-field='name'] , [data-field='email'] , [data-field='phone'] , [data-field='profiles'] , [data-network='github'] , [data-field='location'] , [data-field='url'] .

Sections: section[data-section='work'] , section[data-section='education'] , section[data-section='skills'] , section[data-section='projects'] , section[data-section='awards'] , section[data-section='certificates'] , section[data-section='publications'] , section[data-section='volunteer'] , section[data-section='languages'] , section[data-section='interests'] , section[data-section='references'] , section[data-section='basics'] .

Headings are classified by fuzzy keyword matching.

Entries: .entries (container), .entry (individual <article> ).

Dates: <time> with ISO 8601 datetime . .date-range wraps start/end <time> tags.

Git Integration

If the user hasn't set up the git resumx alias yet, run this first:

git config alias.resumx '!f() { spec="$1"; shift; case "$spec" in :) ;; ) spec="$spec:resume.md";; esac; tag="${spec%%:}"; header=$(git tag -l --format="%(refname:short)" "$tag" 2>/dev/null); subject=$(git tag -l --format="%(contents:subject)" "$tag" 2>/dev/null); [ -n "$header" ] && printf "\033[2m%s\033[0m\n\033[1m%s\033[0m\n\n" "$header" "$subject"; git show "$spec" | resumx "$@"; }; f'

Tag with a note when submitting an application:

git tag -a sent/stripe-2026-02 -m "Tailored for L5 infra, emphasized Kafka + distributed systems"

When the ref is an annotated tag, the alias prints the tag name and note before rendering:

sent/stripe-2026-02 Tailored for L5 infra, emphasized Kafka + distributed systems

Then render from git history:

git resumx sent/stripe-2026-02:resume.md # render from tag git resumx HEAD~3:resume.md --css my-styles.css -o stripe # past commit git show :resume.md | resumx -o staged # staged changes

Pre-commit hook: resumx --check for validation. Post-commit hook: auto-render on every commit.

Using AI

Install agent skills: npx skills add resumx/resumx .

Tailoring to Job Postings

• Give the agent resume.md and the job posting URL

• Agent reads the JD, maps each requirement to existing bullets (covered, weak, missing)

• Agent decides what's durable vs ephemeral: will this change make the next 10 applications better, or just this one?

• Durable improvements (better phrasing, new bullets, new tags) → edit resume.md

• Ephemeral adjustments (keywords, section order, tagline) → create a view or use CLI vars

• Render: resumx resume.md --for stripe-swe -o out/stripe.pdf

With pages: 1 , layout auto-adjusts after every edit.

Zero-File-Modification Rendering

For maximum speed and zero git diff, render entirely from CLI flags:

resumx resume.md --for backend -v tagline="Stream Processing, Go, Kafka" --pin skills,work -o stripe.pdf

Resume Template

pages: 1

Full Name

email@example.com | linkedin.com/in/user | github.com/user

{{ tagline }}

Education

University Name || Sept 2019 - June 2024

Degree Name

• GPA: 3.85

Work Experience

Company Name || Start - End

Job Title

• Achievement with quantified impact using Technology

Projects

Project Name (Individual/Group)

• Description of what was built

Technical Skills

Languages : Java, Python, TypeScript

Frameworks : React, Node.js, FastAPI

Writing Best Practices

• Start bullets with strong action verbs (Led, Developed, Engineered, Increased)

• Quantify results (20% improvement, 500+ users)

• Wrap technologies in backticks

• Use consistent date formats (Jan 2020 - Present )

• Use [Date] for literal brackets in dates: [[Jun 2024]]{.right}