jd-resume-tailorlisted
Install: claude install-skill xungirl/jd-resume-tailor
# jd-resume-tailor
Goal: for one job posting, produce the strongest **honest** one-page resume the user already has the facts for, and get the application submitted cleanly. Tailoring means **choosing and ordering**, not writing new claims.
## 0. Setup (read first)
Look for the user's config at `~/.config/jd-resume-tailor/config.md`. It says where their bullet bank lives, how to build, where PDFs go, and (optionally) a tracker to record applications in. If it does not exist, ask once for:
- the resume source: a bullet-bank file (this repo's `builder/` TOML format, see `builder/example/resume.toml`) **or** their own build script;
- the output folder;
then write the config so the next run skips the questions.
## 1. Get the JD
- Run `python3 <skill_dir>/fetch_jd.py "<url>"`. It follows referral/short links, detects the ATS, and prints title, location, **OPEN / NOT FOUND**, and the JD text. Exit code 2 = posting gone: tell the user before doing any work.
- If the fetcher can't read it (custom career site), ask the user to paste the JD. Never guess a JD from the job title.
## 2. Read it like a recruiter
Extract, in this order:
1. **Hard constraints** (pass/fail): graduation window, degree level, "N semesters remaining after the internship", program dates / duration, onsite city, work authorization or "no sponsorship", application limits ("max 2 applications per cycle"), "must have prior internship".
2. **Must-haves**: languages, domains, fundamentals.
3. **Nice-to-haves**: