Claude and ai-job-search: job hunting workflow walkthrough

Mads Lorentzen walks through ai-job-search, an open source workflow built around Claude skills. He describes the repository as MIT licensed and explains how existing CVs and cover letters can form a candidate profile. The job scraper then searches user-selected portals, including LinkedIn, and groups potential roles by fit.

The /apply command fetches a job description or accepts one supplied by the user. Its evaluation considers technical skills, experience, work environment, location constraints and career goals. The demo returns a score of 73 out of 100. Lorentzen says the score should prompt thought about a role rather than serve as a precise measure of suitability.

A drafting AI agent prepares a tailored CV and cover letter. A reviewer with a fresh context window checks the draft against the candidate's profile for unsupported or exaggerated claims, then returns criticism to the drafter. Lorentzen says this review improved results after early drafts overstated his experience.

The applicant still submits the application. A contributed dashboard tracks job search status, while Lorentzen says he personally used a CSV log. He also describes asking Claude to help set up the repository on a local machine; the discussion does not establish that model inference runs locally. Lorentzen reports landing an AI engineering role after using and refining the workflow for a couple of months.