Elite "career coaches" utilize well-known, long-standing research - package it as their own, and sell you "secrets" for $3,000. Instead, this repo provides Claude Code skills for relationship-focused job searching, implementing methodology grounded in career development research.
Online job applications have notoriously low success rates. Research suggests 50-80% of positions are filled through networking and referrals before being publicly posted, with employee referrals accounting for 30-50% of hires despite comprising only 7% of applicants (Wanberg et al., 2020).
Most AI job tools automate the traditional approach: generate more cover letters, apply to more positions, faster. This toolkit does the opposite—fewer applications, more relationships. The AI handles research and drafting; you decide whether you could genuinely talk to someone about their work for 20 minutes before reaching out.
- Research people whose work genuinely interests you
- Build relationships through informational conversations
- Demonstrate value through signals (portfolio, proposals, demonstrated understanding)
- Apply formally as backup/formality
git clone https://github.com/rhowardstone/AI-Job-Coach.git
cd AI-Job-Coach
pip install python-jobspy pandas requests
./install.sh
# Edit .claude/CLAUDE.md with your profile
claudeThe install script creates the directory structure and copies templates. Edit .claude/CLAUDE.md with your name, email, target companies, and preferences before running claude.
Browser automation (optional): For form-filling, run /plugin inside Claude Code and install Playwright.
See GETTING_STARTED.md for complete setup guide.
python toolkit/scripts/job_search.py "software engineer" --location "Seattle" --min-salary 150000
python toolkit/scripts/greenhouse_search.py stripe anthropic --keyword "engineer"
python toolkit/scripts/email_finder.py "Jane Smith" company.com --verify| Command | Description |
|---|---|
/apply |
Relationship-focused application workflow |
/job-search |
Multi-board job search |
/job-batch |
Batch processing |
/ats-review |
Resume vs job description analysis |
/interview-prep |
Question generation + STAR feedback |
/anti-ai-writing |
Human-sounding text guidelines |
Claude fills, you submit.
- Claude drafts → You send
- Claude researches → You validate interest
- Claude fills forms → You click submit
Granovetter's foundational research found that 55.6% of job-finders learned about their position through personal contacts—predominantly "weak ties" (acquaintances) rather than close friends.
Granovetter, M. S. (1973). The Strength of Weak Ties. American Journal of Sociology, 78(6), 1360-1380. DOI: 10.1086/225469
Spence's Nobel Prize-winning work established that candidates can demonstrate quality through costly signals—credentials, portfolios, and demonstrated work that correlate with actual ability.
Spence, M. (1973). Job Market Signaling. The Quarterly Journal of Economics, 87(3), 355-374. DOI: 10.2307/1882010
Recent research validates that structured networking interventions improve both networking self-efficacy and reemployment quality.
Wanberg, C. R., van Hooft, E. A. J., Liu, S., & Csillag, B. (2020). Can job seekers achieve more through networking? Personnel Psychology, 73(4), 559-585. DOI: 10.1111/peps.12380
The informational interview—conversations to learn about someone's work rather than to pitch yourself—has empirical support for improving networking outcomes.
Kanar, A. (2023). Effectiveness of informational interviewing for facilitating networking self-efficacy. The Career Development Quarterly. DOI: 10.1002/cdq.12318
Effective outreach follows established principles: reciprocity (offer value first), specificity, and making requests easy to accept or decline.
Cialdini, R. B. (1984). Influence: The Psychology of Persuasion. William Morrow. Internet-Archive
MIT