AI Job Search
An AI job-application framework built on Claude Code: evaluate postings, tailor CVs, write cover letters, prep interviews.
About
A geophysicist built this after his position was cut in late 2025, and ran his own job search through it: 69 tailored applications, 20 first interviews, one signed contract, and a start as an AI engineer in June 2026. The framework turns Claude Code into a full-stack application assistant with slash commands for setup, scraping, applying, and interview prep.
Applications go through a drafter-reviewer pipeline where one agent writes the CV and cover letter (compiled from LaTeX) and a second critiques and revises them. The portal search skills ship tuned for the Danish market, but the core pattern — profile, fit evaluation, draft, review — is language- and country-agnostic, and community forks adapt it to other regions.
What it can do
- Posting fit evaluation with structured scoring criteria
- Tailored CV generation compiled from LaTeX
- Cover letter drafting with a reviewer-agent revision loop
- Interview preparation from your profile and the posting
- ATS parseability check on generated PDFs
Setup
- Fork and clone the repo with
gh repo fork MadsLorentzen/ai-job-search --clone - Install Claude Code CLI plus a paid subscription or API credits
- Install Python 3.10+, Bun, and a LaTeX distribution (TeX Live, MacTeX, or TinyTeX)
- Run
/setupinside Claude Code to fill in your profile - Run
/scrapeto find postings, then/apply <url>on matches
Prerequisites
Claude Code CLI (no free tier — needs Claude Pro/Max/Team subscription or API credits); Python 3.10+; Bun; LaTeX distribution with lualatex and xelatex. Portal search skills are built for Danish job boards by default.
Use cases
Works with
Trust score — 5/10
Scored 8/10 (METRICS tier, metrics-v1). Evidence: ~48.5k stars with 16.6k forks on GitHub daily trending, CI running, pushed Oct 15 2026; the author's own hiring story is public and verifiable via his LinkedIn. Docked for: (1) requires a paid Claude Code subscription on top of everything else, so the real cost is nonzero; (2) README flags scam copycats and a no-crypto notice, which speaks to popularity but also to brand confusion risk. License: MIT.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 9.0/10
- Maintenance · recency, releases, issue load
- 7.0/10
- Openness · license + docs
- 9.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · 48,548 GitHub stars; 16,639 forks; GitHub daily trending
- · pushed 2026-10-15; CI running; author's own hiring story publicly verifiable
- · manual review: paid Claude Code subscription is a real cost gate; README scam-copycat warnings signal brand-confusion risk
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.