AI Job Search

AI Job Search is an open-source job application framework built on Claude Code, from geophysicist-turned-AI-engineer Mads Lorentzen. The premise: your job search should run on your machine as a repeatable agent workflow, not as a pile of tabs and browser bookmarks. Fork it, fill in your profile, and let Claude scrape job portals, score how well each posting fits you, tailor your CV and cover letter, and rehearse interview questions. Lorentzen used the exact same workflow in his own search and credits it with 69 tailored applications, 20 first interviews, and a signed offer.

Compare Claude Code for the underlying agent, Codex if you prefer an OpenAI harness, or Cursor for an editor-centric workflow.

Key Features

• Structured agent workflow: /setup, /scrape, /apply, and /interview turn a vulnerable, hand-waving process into runnable steps.
• Fit evaluation and scoring: the framework encodes career-guidance best practices into explicit criteria, so you are not deciding fit from a gut feeling at 1am.
• Drafter-reviewer pipeline: a drafts iteration, then a reviewer agent critiques it and forces a revision before you send anything out.
• LaTeX CV and cover letter output: compiles tailored, polished documents with lualatex and xelatex rather than a generic template.
• Agent-agnostic core: the profiling, fit, and drafter-reviewer workflow is language- and country-agnostic, and the portal skills can be swapped for your local job boards.

Use Cases

Who Should Use This Tool?

• Job seekers between roles: run the whole search as a system instead of a grind.
• Career switchers: the framework encodes fit criteria and comparison logic that are easy to misuse solo.
• Anyone who already uses a coding CLI agent: if you run Claude Code or Codex, this is a natural extension.

Problems It Solves

1. Context switching: no more balancing 20 tabs, a spreadsheet, and three cover-letter drafts.
2. Cover-letter paralysis: the reviewer agent gets you a first, critiqueable draft rather than a blank page.
3. Forgetting follow-ups: the workflow keeps applications, scores, and CV variants organized in one place.

Pricing

The project is MIT licensed, free, and self-hosted. The author funds it through a Ko-fi page and contributions; there is no affiliated cryptocurrency, token, or paid sponsorship program. If something ever asks you to pay for "official pro access", treat it as a scam.

Getting Started

1. Clone and fork: git clone https://github.com/MadsLorentzen/ai-job-search and fork it.
2. Install prerequisites: Claude Code, Python 3.10+, Bun, and a LaTeX distribution (MacTeX, TeX Live, TinyTeX, or MiKTeX).
3. Run /setup: fill in your profile so the framework can score postings against your real constraints.
4. Run /scrape: search job portals (portal skills ship for the Danish market, but the pattern is portable).
5. Run /apply <url>: generate a fit score, tailor the CV, and draft a cover letter, then let the reviewer critique it.

If you use Codex, Antigravity, or Gemini CLI instead of Claude Code, start at AGENTS.md; the portal search skills work there as-is.

Frequently Asked Questions

Does it actually work?

The author used it on his own job search after a layoff in late 2025 and reports the full funnel from applications to a signed offer, starting as an AI engineer in June 2026.

Is it affiliated with Anthropic?

No. It is an independent open-source project that references Claude Code to describe the toolchain.

Do I need a paid Claude plan?

Claude Code is a paid product. The framework itself is free; you pay for whatever Claude usage you consume.

Alternatives

• Career Ops: a larger sibling in the same space, runnable in Claude Code, Codex, OpenCode, and Antigravity.
• Claude Code: the underlying terminal agent if you want to build your own workflow.
• OpenCode: an open-source terminal agent alternative.

Tips & Best Practices

1. Add your own job boards: the Danish portals are defaults; the swappable pattern is the real feature.
2. Use the reviewer agent: never send the first draft; the critique-and-revise loop is where the quality comes from.
3. Be honest with employers: the author was upfront about using the workflow and says it started genuine conversations rather than hurting him.

Conclusion

AI Job Search turns a high-stakes, high-anxiety process into a runnable, reviewable system built on Claude Code. It is free, open source, and proven on its own author's career, which is as good a credibility signal as you will find.