job-hunter - SKILL.md Agent Skill

name: job-hunter description: > Automated job application pipeline. Finds jobs across LinkedIn, Indeed, Greenhouse, and Lever, then guides you through company research, resume tailoring, and cover letter drafting. Use when: (1) user says "find me jobs", "job search", "job hunt", "look for roles", "/job-hunter", "help me apply", (2) user wants to research a company before applying, (3) user needs a tailored resume or cover letter for a specific role, (4) user wants to track their application pipeline. NOT for: actually submitting applications, salary negotiation, or interview prep questions.

Job Hunter

Agentic job application pipeline: search → research → tailor resume → draft cover letter → package materials.

CLI Tool

All commands run from this skill directory:

cd ~/.claude/skills/job-hunter

Search

bun run job-hunt.ts search "<query>" [options]

Options:

--site linkedin|indeed — restrict to one job board
--location "City, ST" — location filter
--remote — remote jobs only
--results N — number of results (default: 15)
--hours-old N — max posting age in hours
--job-type fulltime|parttime|contract|internship — filter by type
--save — auto-save all results to job tracker
--json — raw JSON output

Also searches Greenhouse boards and Lever sites configured in data/profile.json under preferences.greenhouseBoards and preferences.leverSites.

Examples:

bun run job-hunt.ts search "software engineer" --location "San Francisco" --results 10
bun run job-hunt.ts search "senior backend engineer" --remote --save
bun run job-hunt.ts search "ML engineer" --site linkedin --results 20 --save

Job Tracker

bun run job-hunt.ts jobs                              # List all tracked jobs
bun run job-hunt.ts jobs show <id>                    # Show job + materials
bun run job-hunt.ts jobs status <id> <status>         # Update status
bun run job-hunt.ts jobs remove <id>                  # Remove from tracker

Statuses: new → researched → tailored → applied → interviewing → offer / rejected

Pipeline Steps

Each step prints a structured prompt. Execute the instructions Claude gives you.

bun run job-hunt.ts research <id>       # Company research prompt
bun run job-hunt.ts tailor <id>         # Resume tailoring prompt
bun run job-hunt.ts letter <id>         # Cover letter prompt
bun run job-hunt.ts package <id>        # All three steps for one job

Full Pipeline

bun run job-hunt.ts pipeline "<query>" --top 3 [search options]

Searches, saves top N jobs, and prints all prompts for each.

Profile

bun run job-hunt.ts profile             # Show current profile
bun run job-hunt.ts profile init        # Create profile.json from example

Cache

bun run job-hunt.ts cache clear         # Clear search cache

Agentic Pipeline Loop

When the user asks to find and apply to jobs, follow this loop:

1. Search & Save

Run a search with --save to find and track relevant jobs:

cd ~/.claude/skills/job-hunter
bun run job-hunt.ts search "<query>" --location "<loc>" --save --results 10

Present the results to the user. Let them pick which jobs to pursue, or use pipeline for the top N.

2. Research Each Company

For each selected job, run:

bun run job-hunt.ts research <id>

This prints a research prompt. Execute it: use web_search and web_fetch to research the company, then write findings to the output directory specified in the prompt.

3. Tailor Resume

After research is complete:

bun run job-hunt.ts tailor <id>

This prints a tailoring prompt with the user's full profile, the job description, and the research brief. Follow the instructions to write a tailored resume.

4. Draft Cover Letter

After the resume is tailored:

bun run job-hunt.ts letter <id>

Uses all prior context (research + tailored resume) to generate a personalized cover letter.

5. Review & Package

Show the user:

Research brief (data/output/<id>/research.md)
Tailored resume (data/output/<id>/resume.md)
Cover letter (data/output/<id>/cover-letter.md)

Ask for feedback before proceeding to the next job. The user may want to revise.

6. Track Progress

Update job status as the user progresses:

bun run job-hunt.ts jobs status <id> applied
bun run job-hunt.ts jobs status <id> interviewing

Profile Setup

Before tailoring resumes, the user needs a profile:

Run bun run job-hunt.ts profile init
Edit data/profile.json with their details
Include: experience, education, skills, target roles, preferences

The profile feeds into every tailoring and cover letter prompt.

File Structure

skills/job-hunter/
├── SKILL.md              (this file)
├── job-hunt.ts           (CLI entry point)
├── lib/
│   ├── search.ts         (ts-jobspy + Greenhouse + Lever)
│   ├── research.ts       (company research prompt builder)
│   ├── tailor.ts         (resume tailoring prompt builder)
│   ├── cover-letter.ts   (cover letter prompt builder)
│   ├── tracker.ts        (job CRUD — data/jobs.json)
│   ├── cache.ts          (file-based cache, 30min TTL)
│   └── format.ts         (console + markdown formatters)
├── data/
│   ├── profile.json      (user's resume/profile — fill this in)
│   ├── jobs.json         (tracked jobs)
│   ├── cache/            (auto-managed search cache)
│   └── output/<id>/      (per-job materials)
└── references/
    ├── ts-jobspy-api.md
    ├── greenhouse-api.md
    └── lever-api.md