Nothing earned is lost.
A recruiter told me last month she almost messaged a candidate to ask for a cleaner resume PDF. She didn’t, because she could tell the application came from an AI apply tool. The candidate stopped being a candidate.
Now flip the camera. Somewhere, in a kitchen, refreshing LinkedIn for a notification that isn’t coming, sits the candidate. They paid for the tool that sent that application. They paid for the next forty it sent this week. They have no idea any of those applications got filtered before a human read them. No system tells you a recruiter clocked your file as AI slop and quietly moved on.
Here’s the problem. The AI apply tools we’re talking about, the ones that submit on your behalf with no human in the loop and charge by the application, make more money the more you apply. A jobseeker who actually lands a job wants to apply less, to the right roles, with better materials. The tool’s incentive and the jobseeker’s goal point in opposite directions, and the candidate at the kitchen table is paying for the side that loses.
Huntr’s Q3 2025 report tracked 1.61 million applications: tailored ones convert to interview or offer 5.95% of the time, untailored ones 2.9%. But it takes forever to do manually, auto-apply tools do it poorly and the LLM you already use is just as capable with the right framework.
So we built it. It’s called job-hunt-skills, it’s free, open-source, and it lives at github.com/Remotivated/job-hunt-skills. It’s not a tool. It’s a set of skills, prompts, and workflows for using the AI you already pay for to do the thing that actually works.
The workflow is simple. You build a source document that captures your real experience in depth. Then for each role that’s actually worth your time, you research the company, tailor your materials against that one posting, and check every claim against your source files. Every verified bullet, story, or proof point you confirm along the way flows back into your source documents, so each application starts from richer material than the last.
It runs three ways: as a Claude Code plugin in the terminal, as a Cowork plugin in Claude Desktop, or as a copy-paste prompt library for whatever LLM you already pay for.
I’ve put roughly 90 hours into building and testing this over the past few weeks. It’s free, and it stays free. There’s no hosted service, no email capture, no premium tier waiting in the wings. The ask is small: even if you’re not job-hunting today, take a few seconds to star the repo on GitHub or share it with one person who is. That’s the entire mechanism by which a free resource finds the people it was built for.
