Open-source · Claude Code plugin

A job application that sounds like you, not a robot.

Drop in a job link. Get an ATS-ready CV and cover letter, tailored to the role, scored honestly, and written only from your real experience. Local to your machine, built from your facts.

No data leaves your computer
claude, job-application
/job-application https://careers.firm.se/jobs/4821
Reading the ad… scoring your fit…
Fit score: 82 / 100, strong, apply with confidence
Drafting cv.md, letter.md, form-answers.md…
Cleaning tone (slop-proof · nordic-style)…
Exporting → final/cv.docx, final/letter.docx
✓ Done. 1 honest gap flagged before you submit.
What it does

One command. A complete, tailored application.

It reads the ad the way an employer wrote it, then builds each piece around the parts of your history that genuinely match.

i.

Scores the fit, honestly

A clear number out of 100 with a breakdown of where you match and where you fall short. No flattery. A stretch is called a stretch.

ii.

Writes CV and letter

Tuned to this role, in your own voice, leading with your strongest matches and the ad's real keywords, only where you truly have them.

iii.

Answers the form

A researched salary range for your role and city, plus notice and work-permit fields marked clearly for you to fill in.

iv.

Cleans the tone

Two passes strip AI-sounding filler and over-the-top hype, so the writing reads human and quietly confident.

v.

Exports real documents

ATS-styled Word files land in a final/ folder, ready to upload. No Word? You get a clean printable page instead.

vi.

Keeps a quiet record

A simple board tracks every application from drafting to interview, so nothing slips through the cracks.

What's inside

Five skills, one pipeline.

The toolkit is a set of small, focused skills that hand work to each other. Hover or tap a step to see what it does. A job ad goes in one end; a finished, cleaned, exported application comes out the other.

clean

Hover a step to explore

Each skill does one job and passes the result on. Together they take a single job link all the way to ready-to-send Word documents, only ever using your real experience.

Honest by design

It never invents your experience.

The toolkit ships blank. You fill in your real facts once, and every word it writes traces back to them. If a job wants something you do not have, it tells you, rather than making it up.

  • Only uses what is in your own master file. No borrowed achievements, no invented metrics.
  • A title you have not held is never claimed as a full match. Gaps are surfaced, not hidden.
  • Runs entirely on your machine. The scripts make no network calls and collect nothing about you.
  • You always have the final read. It prepares everything and hands it to you. It never hits submit.
Why "ATS-ready" matters

Most CVs are read by software first.

Before a person sees your CV, an Applicant Tracking System reads it as plain text and looks for the ad's exact words. Beautiful layouts often confuse it, and you are filtered out before page one. This toolkit stays clean to the eye and readable to the machine.

✕ What breaks the parser

  • Two-column layouts and sidebars
  • Text boxes, icons, and photos
  • Tables used for visual layout
  • Headers and footers with key details
  • Keyword stuffing it can detect

✓ What this toolkit produces

  • A single clean column
  • Standard, parseable headings
  • Real keywords, only where you match
  • Black-only typographic hierarchy
  • Designed to read, easy to extract
Setup · about 15 minutes, once

From nothing installed to your first application.

Never used a tool like this? Follow the steps in order. Each one shows exactly what to type. A "terminal" is the text window on your computer: Terminal on Mac, PowerShell on Windows. Copy a line, paste it in, press Enter.

Install the free tools you need

These are the pieces the toolkit builds on. Pick your operating system and run the lines for it. You only do this once.

# 1. Homebrew, the installer that fetches the rest (paste, press Enter, follow prompts) /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"
# 2. Git (usually already there) + Node.js + Pandoc brew install git node pandoc
# 3. Claude Code, the app that runs the toolkit curl -fsSL https://claude.ai/install.sh | bash
# Run these in PowerShell. Winget ships with Windows 10/11. # 1. Git + Node.js + Pandoc winget install Git.Git OpenJS.NodeJS JohnMacFarlane.Pandoc
# 2. Claude Code, the app that runs the toolkit irm https://claude.ai/install.ps1 | iex
# 3. Close and reopen PowerShell so the new tools are found.
# Debian / Ubuntu. For other distros use your package manager. # 1. Git + Node.js + Pandoc sudo apt update && sudo apt install -y git nodejs npm pandoc
# 2. Claude Code, the app that runs the toolkit curl -fsSL https://claude.ai/install.sh | bash

What each is for: Git fetches the toolkit, Node.js powers some helpers, Pandoc turns your CV into a Word file, and Claude Code is the app you talk to. Pandoc is optional: without it you still get a clean printable page.

Start Claude Code and log in

Open your terminal and check it installed, then start it. The first time, it opens your browser to sign in with a Claude account (Pro/Max) or an Anthropic API key. You only log in once.

claude --version # should print a version number
claude # starts it, opens the browser to log in

Install the toolkit

With Claude Code running, type these two lines, pressing Enter after each. The first tells it where the toolkit lives, the second installs it. If the new commands do not appear, type /reload-plugins.

/plugin marketplace add mcmespinaa/job-application-toolkit
/plugin install job-application@job-application-toolkit

Create your workspace

This makes the folder where your applications will live. Use any path you like. It is safe to run again and never overwrites anything you have filled in.

/job-application:job-setup ~/Documents/Job Applications

Fill in your details, once

Open the new folder in any text editor. In 00_master/master-cv.md replace every [FILL IN] with your real information, and set your tone in voice.md. To teach it your style, drop old CVs and letters into _archive/ and run the mine command. If the archive is empty, just skip it. It invents nothing.

/job-application:job-setup --mine-archive # optional: learn from your past applications

Apply to your first job

Copy a job ad's link and run the command below. If the page is behind a login, it will ask you to paste the text or a screenshot. When it finishes, you get a folder with your CV, letter, form answers, and ready-to-upload Word files, plus a short "before you submit" checklist. Read it, fill the blanks, and send.

/job-application:job-application https://link-to-the-job-ad

After setup, every future application is just this one line.

Optional extras

Two nice-to-haves, only if you want them.

Everything works without these. Add them later if a cloud copy or a visual board would help. Nothing here is required, and nothing changes where your local files land.

Save copies to Google Drive

Have each CV and cover letter also saved as a Google Doc in a tidy Job Applications/<Company>/ folder on your Drive. Google Drive is a Claude account connector, so you sign in once and it works everywhere you use that account, including VS Code and the terminal.

  1. Open claude.ai (or the Claude desktop app). Go to the Connectors settings, choose Google Drive, and sign in once.
  2. That connection now carries into Claude Code in VS Code and the terminal automatically, because they use the same Claude account. Nothing to install or add per app.
  3. To confirm it is live from VS Code or any terminal, run the command below. You want to see Google Drive marked Connected.
claude mcp list

Once connected, the toolkit syncs automatically on each run. Skip it for one run with --no-drive. Your documents always land in your local final/ folder regardless. This only adds a cloud copy in your own Drive.

View it as a board in Obsidian

Your workspace is plain text, so it opens directly in Obsidian. With the Kanban community plugin, Applications Board.md becomes a drag-and-drop board from drafting to interview.

brew install --cask obsidian
winget install Obsidian.Obsidian
# download the AppImage / .deb from obsidian.md, or: flatpak install flathub md.obsidian.Obsidian
  1. Open Obsidian and choose Open folder as vault, then pick your Job Applications folder.
  2. In Settings → Community plugins, install and enable Kanban, then open Applications Board.md.

The vault is local and yours. Nothing syncs anywhere unless you set up Obsidian Sync yourself.

Questions

Good to know.

Is my personal data uploaded anywhere?
No. Your CV and applications live in a local folder on your computer. The scripts make no network calls. The only optional exception is Google Drive sync, which you turn on yourself and which goes to your own Drive.
Will it lie to make me look better?
No, that is the core design rule. It uses only what you put in your master file, scores you honestly, and flags gaps rather than papering over them. Better to get the interview and pass than to over-claim and fail the case study.
Do I need to know how to code?
No. You copy and paste a few commands, then fill in a CV template in a text editor. That is the whole technical bar.
Does it submit applications for me?
No. It prepares everything and hands it to you with a checklist. You do the final review and submit wherever the ad asks.
What if I do not install Pandoc?
You still get clean, self-contained HTML files you can open in a browser and print to PDF, or paste into Word. Install Pandoc later and re-export for native Word documents.
Can my whole family or team use it?
Yes. Each person installs it and runs the setup to make their own workspace with their own facts. The toolkit ships blank and contains no one's personal data.
Learn more

Where to go deeper.

This page is the quick start. For the complete reference, the method behind it, or a hand when something does not work, here is where to look.

Read

Full documentation

The complete README: every command, every option, the troubleshooting table, and the file-by-file walkthrough. The best place to start after this page.

Open the README ↗
Understand

The ICM method

Why the workspace is shaped the way it is. The Integrated Context Method is what keeps the toolkit both tailored to each job and honest to your real history.

Read the method ↗
Ask

Questions and help

Hit a snag, found a bug, or want a feature? Open an issue on GitHub. Contributions and ideas are welcome, it is open-source and built in the open.

Open an issue ↗
Free and open-source

If it helped you land an interview…

Hi, I'm Maria. I built this toolkit, and it is free, with no email and no sign-up, and it always will be. If it saved you time or got you a callback, you can say thanks with a coffee. It genuinely means a lot.

Buy Maria a coffee