What AI Job Search Is
AI Job Search is a free, open-source job application framework built on Claude Code. You fork it, fill in your profile, and let Claude run your job search: it searches job portals, evaluates each posting against your actual background, drafts a tailored CV and cover letter for the good matches, and prepares you for interviews.
The pitch is simple and the repo states it plainly: the job search that runs on your machine. Nothing is sent to a third-party job platform. Your profile, your applications and your tracking live in your own fork, and the AI does the repetitive parts of applying while you stay in control of what actually gets sent.
It has passed 62,000 stars on GitHub under the MIT license, and it is currently the number one trending repository on the platform. That kind of attention is rare for a career tool, and it makes sense once you see the story behind it.
Why It Matters: It Actually Got Its Creator Hired
Most job search tools promise to save you time. This one has receipts. The author is a geophysicist by training whose position was cut in late 2025. He built this framework to run his own job search, using the same workflow in the repo week after week on his own career.
The numbers from his run: 69 tailored applications, 20 first interviews, and one signed contract. He started as an AI engineer in June 2026. He was upfront with every employer about using AI in his applications, and instead of counting against him, it usually sparked a genuine technical conversation.
That is the strongest possible endorsement for a tool like this. It was not built as a demo or a side project for stars. It was built by someone who needed a job, and it worked.
Key Features
- Fit scoring before you apply: Every posting is evaluated against your profile and scored, so you spend effort on roles you can actually land instead of spraying applications everywhere.
- Tailored CV and cover letter drafting: For each match, Claude drafts a CV and cover letter tailored to that specific posting, compiled to PDF with LaTeX for a clean, professional result.
- Drafter-reviewer pipeline: A second agent reviews and critiques every draft before you see it, then revises. Nothing goes out without a second pair of eyes.
- Interview preparation: The /interview command builds a stage-specific prep pack from the exact posting, the CV and cover letter the interviewer actually read, and your earlier rounds, then maps likely questions to your STAR examples and offers a mock interview.
- Application tracking: The /outcome command records interviews, offers, rejections and silence, archives everything per application, and surfaces applications that have gone quiet so you can follow up.
- Posting integrity checks: Every posting passes a deterministic gate first that looks for hidden text, invisible characters and instruction-shaped lines, since postings are treated as untrusted input.
- Language- and country-agnostic core: The self-profiling, fit evaluation and drafter-reviewer pipeline work anywhere. The portal search skills are built for the Danish market, but the pattern is designed to be swapped for your local job boards.
- Gmail and Notion sync: Optional commands read your Gmail for status signals on open applications and publish a read-only pipeline view into Notion, so the tracker stays current without manual data entry.
How It Works
You start by forking the repo and cloning it, then installing the job portal CLI tools. Inside Claude Code you run /setup, which builds your profile three ways: it can read a documents folder with your CV, LinkedIn export, diplomas and reference letters, import a single CV pasted in chat, or walk you through an interview.
Then /scrape searches multiple job portals for positions matching your profile, deduplicates the results and presents them sorted by fit. If a scrape returns more jobs than you want to eyeball, /rank batch-scores them all against the fit framework and hands you a ranked shortlist.
When you pick a match, /apply runs the full workflow on it: evaluate fit, draft the CV and cover letter, run the reviewer agent over both, revise, and present the final output. If a posting URL cannot be fetched, you can paste the job description directly instead.
From there, /interview preps you for each scheduled interview, and /outcome records what happened and keeps the tracker honest. It is a pipeline, not a one-click apply button, and that is the point.
Who It's For
AI Job Search is for anyone actively job hunting who is tired of the copy-paste application grind: developers, engineers, and career switchers who want every application tailored without spending a Sunday per cover letter. It is especially strong for people who already use Claude Code, since the whole workflow lives inside it.
It is also for skeptics. The author's own funnel is public, the workflow never auto-submits anything, and every draft passes a reviewer agent before you see it. You stay the human in the loop at every step that matters.
If you are casually browsing and not ready to set up a profile, it is probably more machinery than you need. But if you are sending real applications this month, it pays for itself in the first week.
Limitations
First, the cost of entry: Claude Code has no free tier, so you need a Claude Pro, Max or Team subscription, or Anthropic API credits. The framework itself is free, but the engine it runs on is not.
Second, the portal search skills are built for Denmark (Jobindex, Jobnet, Akademikernes Jobbank and similar). The core workflow is portable, but if you are outside Denmark you will need to adapt the search skills to your local boards, or paste postings in manually.
Third, there is a real privacy gotcha the README flags in bold: a fork of a public repo is always public on GitHub, and /setup writes your personal data into tracked files. If the copy is for your own job search rather than contributing back, use a private repository with this repo as upstream instead. Do not put your CV and salary expectations in a public fork.
Finally, the CV pipeline needs a LaTeX distribution with lualatex and xelatex, plus Python 3.10 and Bun. It is a developer tool with developer setup, not a consumer app.
Verdict
AI Job Search earns its trending spot the honest way: it solved a real problem for its creator, then he gave the solution away. The 69-application funnel that ended in a signed contract is the kind of proof most career tools never offer, and the workflow itself is thoughtful, with fit scoring, a reviewer agent, integrity checks and no auto-submit.
At 62,000 stars, free under MIT, and built on a tool many developers already pay for, it is an easy recommendation for anyone running a serious job search right now. Just set it up in a private repo, not a public fork.
