resume-as-code is driven by an AI agent (Claude Code is the running example). You install a binary, install the agent skill, and from then on you just talk to your agent — it runs the CLI and leads your first-time setup instead of waiting for commands.
You don’t need to memorize any commands. The steps below show what happens under the hood, but in practice you install once and let your agent drive.

​
1. Install the binary

macOS (Intel or Apple Silicon) or Linux (x64 or arm64). One command: 
curl -fsSL https://github.com/yevgetman/resume-as-code-dist/releases/latest/download/install.sh | sh
This installs resume to ~/.local/bin/resume. Verify: 
resume --version
There’s no native Windows build yet. Install WSL2 (a Linux environment for Windows) and run the same command from your WSL shell — it installs the Linux build.
If the installer says ~/.local/bin isn’t on your PATH, add the line it prints to your ~/.zshrc or ~/.bashrc, then open a new terminal. On macOS the installer also clears the Gatekeeper quarantine flag, so the binary just runs.
The binaries target glibc. Lightweight musl-based distros (Alpine, some minimal Docker images) aren’t supported — use Debian, Ubuntu, Fedora, or Arch.

​
2. Point your agent at it

​
3. Start your resume — two ways

Your resume lives in a folder you own (a git repository). Pick whichever fits:

Make a folder, open your agent in it

mkdir ~/my-resume && cd ~/my-resume
# then open your agent here
The agent sees it isn’t a resume repo yet and initializes it for you.

Just ask your agent

Start your agent anywhere and say:
“Use resume-as-code to start my resume.”
It asks where the resume should live and initializes it.

​
4. Let your agent lead the setup

The first time, your agent leads — it doesn’t wait to be asked. Behind the scenes it runs a read-only readiness check and walks you through:
1

Check readiness

The agent runs resume bootstrap to see whether you’re inside a resume repo, whether git is initialized, whether this is a first run, and which optional tools are present.
RESUME_JSON=1 resume bootstrap
2

Initialize the repo (if needed)

If you aren’t in a resume repo yet, the agent runs resume init — scaffolding resume/, a git repo, and a CLAUDE.md / AGENTS.md bootstrap so any agent that opens the folder knows what it is and how to help.
3

Import an existing resume — if you have one

The agent asks: “Do you have a current resume — a PDF, Word doc, LinkedIn export, or JSON Resume file? Point me at it and I’ll import it.”
resume import ~/Downloads/resume.pdf
LinkedIn exports and JSON Resume files import deterministically; PDFs and Word docs are read and structured into entries for you to review.
4

Or seed from scratch

No resume yet? The agent fills in the essentials by conversation — your name, headline, contact, summary, and your most recent role.
5

Preview and go

The agent offers a live preview (resume preview) so you see it render, then hands off: from here, just say what to add, edit, tailor, or export.
This guided first run is built into the agent skill — you don’t trigger it manually. Install, open your agent, and answer its questions.

​
Dependencies

Everything core works with the binary alone. A few capabilities use optional tools, which resume bootstrap reports on:
CapabilityNeedsIf missing
PDF importnothingInstalling pdftotext (Poppler) improves layout fidelity; otherwise the bundled unpdf fallback handles it.
PDF exportChrome for TestingDownloaded automatically (~170 MB) on the first export, or pre-fetch with resume bootstrap --ensure-chrome.
DOCX exportpandocInstall pandoc; the readiness report prints the command for your OS.
The CLI never installs system tools for you and never prompts — it surfaces what’s missing, and your agent helps you install it.

​
What’s next

Command reference

Browse every resume command by category.

In-binary cheatsheet

Run resume cheatsheet any time for the exhaustive, always-current reference your agent reads.