Open-source project
rendercv/rendercv avatar
rendercv/rendercv

RenderCV's schema is generated, and its AI skill is evaluated against its own validator

Resume builder for academics and engineers

17,961 stars1,387 forksPythonMIT

At a glance

What is it?
RenderCV turns a YAML file into a typeset CV PDF and is aimed at academics and engineers who want their CV in version control. The useful detail sits in the design block, the locale block, and the build files rather than in the front page, and one place where the container and the development sync disagree is worth knowing before you rely on either.
Who is it for?
RenderCV suits someone who already thinks of a CV as text worth versioning, and who will accept a fixed set of section shapes and nine themes rather than arbitrary layout. Before adopting it, confirm you are on Python 3.12 or newer, since the install requires it, and check that the `full` extra covers whatever you need, because the container image and the development sync do not install the same set.
Can I use it commercially?
Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
Is it still maintained?
Yes. The repository last received commits 15 days ago.
What is it written in?
Mainly Python, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.

From daily snapshots

Star history

  • rendercv/rendercv17,961 ★ · +346
Star history
Repository2026-09-212026-11-15
rendercv/rendercv17.6k18k

346 Stars in the last 55 days, measured from the daily GitHub snapshot. Compare with other projects

Editorial analysis

Two verbs, one extra name, and a hard floor on Python

The whole workflow is two commands. One scaffolds a file:

bash
rendercv new "John Doe"

The other renders it to a PDF:

bash
rendercv render "John_Doe_CV.yaml"

Everything between those two points is editing YAML. The install is equally short, and the extra name is part of the instruction rather than an optional suggestion:

bash
pip install "rendercv[full]"

The floor is Python 3.12 or newer, which is the first thing to verify on an existing machine. The consequence for a reader is that the `full` extra is the documented install, so a bare package name gives you something narrower than what the guide assumes.

The design block is nine named colour roles, not a free-form style sheet

Theming is expressed as a fixed vocabulary of semantic roles rather than as arbitrary styling. The page block carries `size: us-letter`, four separate margin settings at `0.7in`, and two boolean toggles, `show_footer` and `show_top_note`. The colours block names eight roles: `body`, `name`, `headline`, `connections`, `section_titles`, `links`, `footer` and `top_note`, each set with an explicit `rgb()` triple. The shipped example sets `body` to rgb(0, 0, 0), gives five roles the same blue rgb(0, 79, 144), and gives both `footer` and `top_note` the same grey rgb(128, 128, 128). Typography takes `line_spacing: 0.6em`, `alignment: justified` and `date_and_location_column_alignment: right`, alongside a `font_family` key. The consequence is that you colour by meaning rather than by element, which is coherent, but it also means a design that needs a colour no role covers has nowhere to go.

Nine themes ship, and only one has a checked-in output directory

The examples table links nine rendered PDFs for the same John Doe content, and the theme names are ClassicTheme, EngineeringresumesTheme, Sb2novTheme, ModerncvTheme, EngineeringclassicTheme, HarvardTheme, InkTheme, OpalTheme and EmberTheme. The repository's examples/ directory backs that with a matched pair per theme, a `.yaml` input and a `.pdf` output, so you can read the exact input that produced a given layout. The asymmetry is in the last entry: `examples/output_Opal/` appears for OpalTheme and for no other theme. So Opal has an extra output directory, and for the other eight the PDF is the whole artifact. The consequence for a reader comparing themes is that you render them yourself from the sample YAML rather than expecting side-by-side intermediates, and that picking a theme is a decision made by rendering all nine against your own content.

Localisation is a data block you fill in, not a translation pass

The locale block is how a non-English CV gets built, and the shape is instructive. Under `locale` you set `language`, then supply the literal strings the renderer will print: `last_updated: Last updated in`, plus `month`, `months`, `year`, `years` and `present`, and a `month_abbreviations` list beginning with Jan, Feb and Mar. Every one of those is a string you author. The consequence is that localisation is data entry rather than translation: nothing infers the plural form of `month` versus `months` from your language setting, because you have already told it both. The same applies to the `present` marker used in date ranges and to the phrasing of the last-updated footer, which is a sentence you write, not a format flag.

The schema is a generated file, and the recipe to regenerate it is checked in

Strict validation is the project's headline claim, phrased as no surprises, where an error tells you exactly what and where. A `schema.json` sits at the repository root, and that file is what lets an editor fill out the YAML interactively with autocompletion and inline documentation. It is not hand-maintained. The justfile carries an `update-schema` recipe that runs `scripts/update_schema.py`, alongside `update-examples` running `scripts/update_examples.py` and `update-entry-figures` running `scripts/update_entry_figures.py`. The consequence is a real maintenance edge: the schema, the example CVs and the documentation figures are all generated artifacts, so after changing the data model you have three regeneration steps to remember, and an editor's autocompletion can go stale relative to your installed version if you skip them.

The agent skill is generated from the source and graded on validity, not on quality

RenderCV ships an agent skill, installed with:

bash
npx skills add rendercv/rendercv-skill

It works with any agent supporting the skills standard, and the provenance is unusually explicit. The skill is auto-generated from RenderCV's own source code by `scripts/rendercv_skill/generate.py`, and it is evaluated with promptfoo under `scripts/rendercv_skill/evals` against RenderCV's own Pydantic validation pipeline. The consequence is worth stating plainly, because it bounds what the evaluation proves. Grading an agent against the project's own validator establishes that the YAML the agent writes is accepted by RenderCV. It does not establish that the CV is accurate, complete or well ordered, and no such check is described. The second consequence is staleness: generation is a manual recipe in the justfile, so a skill generated before a schema change teaches an agent a schema that no longer validates.

The container installs one extra, and the development sync installs all of them

The Dockerfile and the justfile disagree about what gets installed, and the difference is easy to miss. The container builds from `ghcr.io/astral-sh/uv:python3.12-bookworm-slim`, then runs `uv sync --frozen --no-install-project --no-editable --extra full --no-default-groups`, and the final stage is `python:3.12-slim-bookworm` with a non-root `rendercv` user at uid and gid 999, `ENTRYPOINT ["rendercv"]` and `CMD ["--help"]`. The justfile's own `sync` recipe is `uv sync --frozen --all-extras`. So the image carries the `full` extra alone while a developer sync pulls every extra, and a capability you verify locally can be absent in the container. Two smaller details from the same files. Both use `--frozen`, so the checked-in `uv.lock` governs. And the build backend is pinned to `uv_build>=0.10.3,<0.11.0`, a narrow range that must be satisfied to build from source. The last release is v2.8 on 2026-03-21, and the last push is dated 2026-10-31.

Editorial conclusion

RenderCV suits someone who already thinks of a CV as text worth versioning, and who will accept a fixed set of section shapes and nine themes rather than arbitrary layout. Before adopting it, confirm you are on Python 3.12 or newer, since the install requires it, and check that the `full` extra covers whatever you need, because the container image and the development sync do not install the same set. If you plan to let an agent write the YAML, note that the skill is generated from the source and evaluated against RenderCV's own Pydantic validation pipeline.

Frequently asked questions

how to use rendercv

Install with `pip install "rendercv[full]"`, which requires Python 3.12 or newer. Create a file using `rendercv new "John Doe"`, edit the YAML, then generate the PDF with `rendercv render "John_Doe_CV.yaml"`. The hosted version is deployed at rendercv.com.

what is rendercv

A resume builder for academics and engineers. It reads a CV written in a YAML file and generates a PDF with professional typography, and the intent is that you version-control the CV because it is just text. The package version is 2.8 and the license is MIT.

is rendercv free

The repository is MIT licensed and the package installs from PyPI with `pip install "rendercv[full]"`, so the software carries no stated purchase cost. A hosted version is deployed at rendercv.com, and the repository text does not state any pricing for it.

is rendercv ats friendly

The repository text does not discuss applicant tracking systems at all. What it claims for the output is consistent alignment and spacing handled for you, and the typography is set through the design block, including `alignment: justified` and `date_and_location_column_alignment: right`.

rendercv vs latex

No comparison with LaTeX appears in the repository text. The documented workflow is a YAML file rendered through the `rendercv` CLI into a PDF, with layout controlled by a design block holding theme, page, colors and typography settings.

rendercv alternatives

The repository names no alternative tool. It points to its own surfaces instead: a hosted version at rendercv.com, documentation at docs.rendercv.com, nine bundled themes, and a root-level schema.json that gives editors autocompletion and inline documentation for the YAML.

Official sources

  1. License: MIT
  2. Project website
  3. README
  4. Releases
  5. rendercv/rendercv on GitHub
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/rendercv-rendercv.svg)](https://hysenlabs.com/projects/rendercv-rendercv)

Star history chart for your README

markdown
[![Star history](https://hysenlabs.com/chart/rendercv-rendercv.svg)](https://hysenlabs.com/projects/rendercv-rendercv)

Was this useful?