job-hunt

A Claude Code skill that runs a USA-only, tech-only job search end to end:

1. Pulls live postings from legal, official ATS JSON APIs — Greenhouse, Lever, Ashby, SmartRecruiters, Recruitee. (No LinkedIn/Indeed scraping.)
2. Filters to USA tech roles in your chosen sub-sectors/titles (drops non-US, non-tech, and clearance-required roles).
3. Scores each role 0-100 for relevance to what you asked to search for.
4. Groups roles by company and renders a browsable HTML web app.

You review and apply to everything yourself. This skill never auto-applies, and it does not write resumes or cover letters.

Invoke it

In Claude Code, just say:

run my job hunt

On first use (no config/config.yaml), Claude walks you through setup — choosing sub-sectors, target titles, and companies — then runs the search. To reconfigure later: "set up my job hunt".

Install

This is a Claude Code skill. Clone it into your Claude skills directory — the folder must be named job-hunt:

# macOS / Linux
git clone https://github.com/canmenzo/jobscope.git ~/.claude/skills/job-hunt

# Windows (PowerShell)
git clone https://github.com/canmenzo/jobscope.git "$env:USERPROFILE\.claude\skills\job-hunt"

Install the two Python dependencies (Python 3.11+):

cd ~/.claude/skills/job-hunt          # Windows: %USERPROFILE%\.claude\skills\job-hunt
pip install -r requirements.txt

Restart Claude Code so it picks up the skill, then say "run my job hunt". First run walks you through setup; your answers are saved to config/config.yaml (git-ignored, never shared).

Setup (what onboarding asks)

• Sub-sectors (tech only): Cybersecurity, Software Engineering, Data/ML/AI, DevOps/SRE/Cloud, IT/Systems, Product/Design, QA/Test.
• Target titles: pick from the titles under your sub-sectors, or add your own.
• Companies: search all catalog companies (default) or a chosen subset.

Your answers are written to config/config.yaml (see config/config.example.yaml). The title/keyword taxonomy lives in config/taxonomy.yaml; the company list in config/companies_catalog.yaml.

Run manually

python scripts/job_hunt.py                 # full run
python scripts/job_hunt.py --limit 10      # cap companies (test)
python scripts/job_hunt.py --pick "greenhouse:stripe,ashby:workos"
python scripts/job_hunt.py --min-score 70  # raise the relevance bar
python scripts/job_hunt.py --new-only      # only roles unseen on prior runs

python scripts/build_dashboard.py          # build + open the web app
python scripts/build_dashboard.py 2026-06-24   # a specific date

The dashboard (how you review results)

Every run writes runs/<DATE>/index.html and opens it. It shows:

• a stats header — roles, new, companies with matches, searched OK, failed;
• each company as a collapsible branch, its roles underneath with a relevance score badge, an Apply link, a NEW badge, and matched-keyword chips;
• a live search box and a New only toggle;
• companies that searched OK but matched nothing (collapsed list);
• failed boards at the bottom, each with the failure reason and a direct careers-page link so dead slugs are easy to spot and prune.

Relevance score (0-100)

• 78-100 — a selected title phrase appears in the job title (strongest).
• 55-77 — a sub-sector keyword appears in the job title.
• 38-54 — match only in the description (noisier; below the default cutoff).

Default min_score is 55, so you see title matches by default. Lower it in config/config.yaml to include description-only matches.

Add or change companies

Edit config/companies_catalog.yaml — slugs grouped by ATS source. The slug is the identifier in that ATS's public URL:

• Greenhouse: job-boards.greenhouse.io/<slug>
• Lever: jobs.lever.co/<slug>
• Ashby: jobs.ashbyhq.com/<slug>
• SmartRecruiters: jobs.smartrecruiters.com/<slug>
• Recruitee: <slug>.recruitee.com

A failing slug never crashes the run — it's listed in the dashboard's failed section with a careers link. Companies on Workday/iCIMS/Google careers aren't reachable by these public APIs.

Layout

job-hunt/
  SKILL.md                  workflow Claude follows (setup + run)
  README.md                 this file
  requirements.txt
  config/
    config.yaml             your choices (written by setup)
    config.example.yaml     schema reference
    taxonomy.yaml           sub-sectors -> titles -> keywords
    companies_catalog.yaml  companies by ATS source
  scripts/
    fetch.py                pull the ATS APIs (+ per-company status)
    filter.py               USA + selected-sector gate
    score.py                0-100 relevance
    job_hunt.py             orchestrator (run this)
    build_dashboard.py      build + open the web app
  runs/<DATE>/_run.json     grouped results
  runs/<DATE>/index.html    the dashboard
  seen_jobs.json            new-vs-seen cache across runs