job-hunter-kit 0.12

Job search automation with LLM API pipeline mode and AI agent mode.

Job Hunter

Job hunting is repetitive work. Job Hunter automates the parts that don't need you: scraping listings across major job boards, scoring each one against your profile, tailoring your resume per application, and drafting cover letters. You handle the conversations.

Works interactively inside Claude Code or Codex (VS Code extensions), or runs fully autonomous via LLM API for unattended pipelines and GitHub Actions.

What It Does

• Discovers jobs across LinkedIn, Indeed, Glassdoor, Himalayas, Remotive, and more — filtered by your titles, regions, and exclusions
• Scores each listing against your career context so you know what to prioritize
• Tailors your resume per job and generates a cover letter, ready for PDF export
• Tracks applications with a dashboard and analytics so nothing slips through

Install

uv tool install job-hunter-kit

Standard install supports both agent and llm-api modes.

Quick Start

job-hunter init my-workspace
cd my-workspace
job-hunter doctor

Open workspace in VS Code, then run /setup onboard, /setup context, /setup stories, and /setup resume. job-hunter doctor validates config and reports exact fixes.

Modes

Mode	What runs	When to use
agent	Python prepares context; Claude Code or Codex (VS Code) skills handle scoring, tailoring, and writing	Interactive daily review
llm-api	Full autonomous pipeline; LLM APIs called inside Python	Unattended runs and GitHub Actions

Set mode: in config/job_hunter.yml. Default is agent.

Daily Workflow

job-hunter hunt --region primary
job-hunter brief
job-hunter dashboard --no-interactive

In agent mode, open the workspace in VS Code with Claude Code or Codex and use:

/job-hunter brief
/job-hunter batch
/job-hunter one <url>
/job-hunter finalize

In llm-api mode, job-hunter hunt runs scrape → score → tailor → cover letter → PDF → tracker in one pipeline.

Optional Company Hunt

For company career pages that need a real browser, add targets to config/career_pages.yml, then run the Company Career Hunt workflow from GitHub Actions. Results are committed to outputs/browser_hunt/jobs.json.

CLI Reference

• job-hunter init <workspace> — create a workspace
• job-hunter doctor — check setup health
• job-hunter hunt — discover and enrich jobs
• job-hunter brief — write the daily briefing
• job-hunter tailor — tailor resume for one or more job postings
• job-hunter dashboard, applications — inspect application state
• job-hunter update — update workspace assets, skills, and workflows after a package upgrade
• job-hunter update --skills-only or --workflows-only — targeted refresh
• job-hunter version — version and upgrade guidance

Bundled skills use hidden job-hunter internal ... commands. They are not part of normal user workflow.

Data Contract

Your data stays yours. Product updates must not overwrite config/, profile/, outputs/, or .env. Deterministic choices live in config/job_hunter.yml; career and writing guidance lives in profile/career_context.md; URL dedup state lives in outputs/state/discovered_urls.yml.

See DATA_CONTRACT.md for the full contract.

Development

uv sync --extra dev
uv run pytest tests/ -q --tb=short
uv run ruff format --check job_hunter tests .github/scripts
uv run ruff check job_hunter tests .github/scripts
uv run ty check job_hunter tests
uv build

MIT licensed. See CONTRIBUTING.md.

Lineage

Job Hunter is the evolution of a single experiment: job-hunter-core and job-hunter-template worked as one system to prove the pipeline, the workspace model, and the agent skill layer — everything this package ships today.