jobhunt

Wake up to ready job applications.

A Claude Code skill that scans job portals, ranks openings against your CV, and builds tailored application packages — CV + cover letter + interview prep — for the top matches each morning.

What it does

Each morning, jobhunt:

1. Scans ~20 portals (Ashby, Greenhouse, Lever + cross-portal queries)
2. Ranks new openings against your CV
3. Builds the top 5 application packages — tailored CV, cover letter, mock-interview prep — in ~/.jobhunt/company/

You wake up to 5 folders, review them, submit the ones worth submitting.

Quickstart

There are two ways to install jobhunt:

Method 1: NPM Auto-Installer (Recommended)

You can install jobhunt via NPM. It will automatically detect your local Claude configuration and install the skill seamlessly:

$ npm install -g jobhunt-skill
$ npx jobhunt-skill

Method 2: Manual / npx skills add

If you are already familiar with the community skills manager, you can use it:

$ npx skills add github.com/Nyx-abu/jobhunt

After installation, run the setup wizard:

$ /jobhunt setup
  > Country? US
  > CV file path? ./resume.pdf
  > Target roles? Junior backend, AI engineer
  > Experience? junior
  > Salary floor? [40000] <enter>
  > Skip tiers? none
  > Cloud cron? later
  Setup complete. Run /jobhunt morning.

$ /jobhunt morning
  Scanning 47 portals... 38 openings found.
  Ranking top 5 against your CV...
  Building applications:
    ✓ langchain-fse-langsmith   8.1/10  ~$50-65K
    ✓ resend-platform-eng       7.9/10  ~$60-80K
    ✓ inngest-applied-ai        7.8/10  unknown
    ✓ pinecone-swe-rag          7.5/10  ~$45-60K
    ✓ supabase-edge-functions   7.4/10  ~$50-70K

  5 folders ready in ~/.jobhunt/company/. Open the top one.

Three commands · seven questions · one PDF · working output.

Features

• Cross-platform: macOS, Linux, Windows
• 13 locales: US, UK, EU, CA, AU, SG, IN, BR, DE, NL, IE, UAE, MX
• Hybrid mode: local-only by default; optional cloud cron for "wake up to ready folders"
• Honest about skips: /jobhunt why <slug> explains why each opening was skipped
• Token-thrifty: ~$0.40-0.60 per morning at default settings (Anthropic API)
• Zero auto-submit: produces artifacts; you review and submit

[!TIP] Cross-Agent Testing Initiative: While jobhunt is battle-tested on Claude Code, we want to ensure it works across all agents (Cursor, Gemini CLI, Antigravity, Windsurf, etc.). We need testers! Check out CONTRIBUTING.md if you can help verify or improve cross-agent compatibility.

Modes

/jobhunt setup              # 7-question wizard (first run)
/jobhunt morning            # full chain: scan + rank + build top 5
/jobhunt apply <slug-or-url># one-off package for a specific company
/jobhunt status             # today's queue + next-best action
/jobhunt why <slug>         # explain why a slug was skipped
/jobhunt scan               # scan only, no rank/build
/jobhunt diagnose           # ATS audit of cv.md

Docs

• Quickstart — local-only flow
• Cloud cron setup — power-user upgrade
• Custom portals — add your own companies
• Developing — for contributors
• Recipes — opinionated configuration examples

Requirements

• Node.js 18+
• Claude Code
• A CV in PDF or DOCX format (or a LinkedIn profile as a fallback)

Credits

jobhunt is built on top of career-ops by Santiago Fernandez, a comprehensive job-hunting Node toolchain released under the MIT license. We vendor and extend a fork of career-ops in vendor/career-ops/ — see ATTRIBUTION.md for the full credits list.

Resume-tailoring prompts are adapted from the resume-diagnoser, resume-recruiter, resume-rewriter, and resume-hiring-manager skill patterns.

The skill installer machinery is from obra/superpowers.

License

MIT — see LICENSE.