kurt-chan/ai-job-scraper — explained in plain English
Analysis updated 2026-05-18
Automatically search job boards and Reddit for roles matching your resume.
Score job postings for fit and get an apply or skip verdict on each.
Generate a tailored cover letter for every job worth applying to.
Review scored jobs and cover letters in a local web dashboard.
| kurt-chan/ai-job-scraper | internrobotics/sim1 | jbarach2012/authprobe | |
|---|---|---|---|
| Stars | 141 | 141 | 141 |
| Language | Python | Python | Python |
| Setup difficulty | moderate | hard | easy |
| Complexity | 3/5 | 5/5 | 3/5 |
| Audience | general | researcher | developer |
Figures from each repo's GitHub metadata at analysis time.
Needs a Claude Code CLI subscription plus a Firecrawl API key to run.
AI Job Hunt Agent is a tool that automates the job search process from start to finish. It reads a resume file you provide, searches the web for matching remote roles, uses Claude to score each job posting against your actual background, and writes a tailored cover letter for every job it decides is worth applying to. All of the results show up in a local web dashboard you can review in a browser. The pipeline runs in four steps. First, Claude reads your resume.md file and pulls out your target roles, key skills, and search queries. Second, a service called Firecrawl runs those search queries and scrapes the matching job board pages. Third, Claude scores every posting from 0 to 100 based on things like how well your skills match, seniority level, whether the role is genuinely remote, how fresh the listing is, and any warning signs, then gives an apply or skip verdict. Fourth, for every job marked apply, Claude drafts a short cover letter with a suggested angle. Nothing about your profession is hardcoded, so it can work for developers, designers, virtual assistants, writers, accountants, or marketers, since everything is derived from your own resume. Under the hood it is a Python and FastAPI backend with a single HTML file dashboard built with plain JavaScript and Tailwind, so there is no separate frontend build step. Setup requires Python 3.10 or newer, the Claude Code command line tool installed and signed in, and a Firecrawl API key. You install the requirements, copy the example environment file and add your Firecrawl key, then drop your own resume.md into the project folder, which stays on your machine and is never committed to the repository. You can run it either through the web dashboard, which shows live progress, or headlessly from the command line. Results land in an output folder as a scored jobs file, a file with every raw scraped listing, and a folder of cover letters. A config file lets you add or change which job boards and subreddits it searches, since the defaults cover general sites like LinkedIn and Indeed alongside a set of profession neutral subreddits.
An automated pipeline that reads your resume, finds matching remote jobs, scores each one with Claude, and drafts tailored cover letters for the best matches.
Mainly Python. The stack also includes Python, FastAPI, Claude Code.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly general.
This repo across BitVibe Labs
double-check against the repo, no cap.