PyreMark

v0.2.0 — A document generation pipeline for multi-lingual, multi-profile professionals who need to produce tailored CVs and cover letters from a portable, self-owned data repository.

Define your career data once. Compose targeted outputs.

Overview

PyreMark turns a single repository of YAML data into many tailored documents. Select different sections, reorder them, switch languages, or apply different themes — all from the same source data, controlled by a TOML configuration file.

Workflow

Layer	Format	Purpose
Data	YAML	Capture experience, skills, education as independent modules
Config	TOML	Select files, set language, reorder sections, override fields
Template	Jinja2 + HTML/CSS	Rendering via headless Chromium

YAML data modules are selectively assembled via a TOML configuration file, rendered through Jinja2 HTML/CSS templates, and converted to PDF using Playwright/Chromium. Cover letters can be authored as Markdown files with YAML front-matter and automatically fed into the same pipeline.

Features

• Multi-lingual data. Every field can be a plain string or a language-indexed dictionary. Built-in fallback — no duplicate files needed.

• Modular by design. Sections are self-contained data objects that declare their own template. Compose, reorder, or omit sections per build without touching template code.

• Configuration-driven. A single TOML file controls which data to include, section order, field overrides, and style tokens. Different profiles produce entirely different documents from the same YAML repository.

• Theme system. Named themes provide templates, styles, and images. Themes can extend other themes, and users can override any template with a local theme directory. Pre/post CSS files allow fine-tuning without forking.

• QR codes. Sections can automatically generate QR code images from a URL field, embedded directly in the output.

• XMP metadata. Generated PDFs carry embedded metadata allowing for traceability and reproducibility

• Markdown workflow. Write cover letters in Markdown with PyreMark front-matter. The body becomes document content, front-matter provides metadata, and a referenced TOML config handles the build.

• Data export. Export the merged data or full build dictionary to YAML for debugging or external use.

• Pydantic validation. All YAML data is validated against a typed pydantic model hierarchy. Dates support multi-language month names via dateparser (30+ languages) — "Nov 2021 - <br> Nov 2023" and {fr: "Fev-Oct 2015", en: "Feb-Oct 2015"} are both valid. Consistency checks catch translation drift across language variants. JSON Schema generation via pyremark schema cv / pyremark schema cover-letter for LLM integration.

• Layout metadata separation. Rendering directives (template, id, classes) live in a separate layout.yaml file, keeping the data schema clean for validation and LLM consumption.

What this project does not plan to become

• Drag and drop/ WYSIWYG: This project does not intend to become a universal CV creator with canvas-based editing. Finely-tunable theming is the way to go for users to get a personalized appearance for their documents.
• Data ownership: Insisting on human-readable interfaces makes much more sense for users that want to own their data. That implies interoperability, and an open environment/ecosystem for the future.

Similar projects

1. Yaml Resume — LaTeX-based, harder to customize visually, no native multi-lingual support.

2. JSON Resume — Schema-driven with community themes, but no built-in multilingual support.

3. ModernCV / Awesome-CV — LaTeX-based, single-language. PyreMark's Jinja2 approach requires no LaTeX knowledge.

Getting Started

Prerequisites

• Python 3.12+
• Poetry

Installation

git clone https://github.com/stalhatz/PyreMark
cd PyreMark
poetry install
poetry run playwright install chromium

Quick Start

Generate a CV from a TOML configuration:

poetry run pyremark --cv testdata/cv/single_page.toml -o output.pdf

Generate a cover letter from Markdown:

poetry run pyremark -m testdata/md/cover_letter.md -o cover_letter.pdf

Preview the HTML before generating PDF:

poetry run pyremark --cv testdata/cv/single_page.toml -s html

Key CLI Flags

Flag	Description
--cv <file>	TOML configuration file
-m, --md <file>	Markdown file with front-matter
-l, --lang <code>	Output language (default: en)
-o, --output <file>	Output PDF path
-s, --show <pdf|html|None>	Preview output
--theme <name>	Named theme
--tags <tag>	Document tags for PDF metadata (repeatable)

Run poetry run pyremark --help for the full list of options.

Future Features

• MCP server for LLM-assisted CV assembly
• Web app with live preview and section management
• Obsidian plugin for Markdown-to-CV workflow

Contributing

See CONTRIBUTING.md for architecture, conventions, and development workflow.