Skip to content

USAGE.md

Latest commit

 

History

History
122 lines (99 loc) · 3.76 KB

USAGE.md

File metadata and controls

122 lines (99 loc) · 3.76 KB

OfferAgent · Usage Guide

0. Prerequisites

1. Setup

cd /Users/clarkfan/OfferAgent

# 1) deps already installed via `npm install`

# 2) set your API key
cp .env.example .env
# then edit .env and paste your sk-ant-... key

# 3) initialize SQLite DB
npm run init-db

2. Daily Workflow

A) Score a JD against your CV

node bin/match.js --jd ./examples/jd-sample-vp-growth.txt
# → outputs ./out/match-report.md

B) Generate a tailored cover letter

node bin/cover.js \
  --jd ./examples/jd-sample-vp-growth.txt \
  --company "群核科技" \
  --role "VP Growth" \
  --lang zh
# → outputs ./out/cover-群核科技-VP-Growth-zh.md

For an English version:

node bin/cover.js \
  --jd ./out/jd-vercel-headofapac.txt \
  --company "Vercel" \
  --role "Head of APAC GTM" \
  --lang en

C) Crawl a public JD URL

node bin/crawl.js --url "https://example.com/jobs/12345"
# → outputs ./out/jd-<company>-<title>.json + .txt

⚠️ For platforms requiring login (BOSS, LinkedIn jobs, Liepin executive search), copy-paste the JD into a .txt file instead — we don't bypass auth walls.

D) Generate interview prep brief

node bin/prep.js --company "群核科技" --role "VP Growth"
# → outputs ./out/prep-群核科技.md
#   (includes: company brief, 30 likely Qs+A frameworks, 5 reverse Qs,
#    100-day plan, salary negotiation strategy)

3. Recommended Daily Loop

Morning (45 min)

  1. Open target-companies/list.md — pick 3 S/A-tier companies
  2. For each: find a real JD link or paste JD text
  3. Run match.js → reject anything < 70 score
  4. For passing JDs: run cover.js

Midday (30 min)

  1. Polish the generated cover letters — make them feel human
  2. Send via the appropriate channel (LinkedIn / email / WeChat)
  3. Log in outreach-tracker/tracker.md (or via init-db + a query)

Evening (15 min)

  1. Check responses, schedule follow-ups
  2. For interview invites: run prep.js 24h before

Weekly (1 hr Sunday)

  1. Review outreach-tracker stats — what's working?
  2. Update target-companies/list.md with new finds
  3. Reach out to 5 new headhunters from headhunter-net/list.md

4. Project Structure

OfferAgent/
├── bin/                    # CLI scripts
│   ├── match.js           # JD ↔ CV scoring
│   ├── cover.js           # cover letter generator
│   ├── crawl.js           # JD URL → structured data
│   ├── prep.js            # interview prep
│   ├── init-db.js         # SQLite schema setup
│   └── cli.js             # unified entry (placeholder)
├── cv/                     # master CVs (CN + EN)
├── target-companies/       # 100+ scored target list
├── cover-letter-gen/       # templates + generated outputs
├── outreach-tracker/       # CRM (Markdown + SQLite)
├── interview-prep/         # playbook + generated briefs
├── headhunter-net/         # executive recruiter contacts
├── examples/               # sample JDs for testing
├── data/                   # SQLite database (gitignored)
├── out/                    # generated artifacts (gitignored)
├── README.md               # project overview
├── USAGE.md                # this file
└── package.json

5. Tips

  • CV polish iteratively: After 5 applications, see which CV bullets get reactions. Update cv/范大勇-CV-V2-CN.md and re-run match scoring.
  • Cover letters are 70 % of the value: A great cover letter on a decent CV beats a great CV with a generic cover letter.
  • Track everything: If you don't log it, you can't improve it.
  • Don't spam: 5 great applications/day > 50 spam applications/day.