get-me-a-job

get-me-a-job

An AI-powered job search pipeline built by a Berkeley Haas MBA student who used it to land a job. Now packaged for classmates.

GO BEARS.

Restricted to @berkeley.edu accounts. The Google OAuth app is configured as Internal to the Berkeley Workspace.

---

What it does

Six skills that work together to run your entire job search from Cowork:

• Resume Tailor — Paste a JD, get a tailored resume with rewritten bullets and a gap analysis
• Cover Letter — Writes cover letters that sound human, not like a robot summarizing your resume
• Interview Prep — Company research, 15-20 tailored questions, story mapping, practice mode
• Network Outreach — Finds contacts, drafts value-first outreach messages, sends via your Gmail after you approve
• Company Research — Deep research briefs with fit analysis against your profile
• Humanizer — Strips AI-sounding patterns from any text (mandatory pass on all outputs)

Three shortcut commands:

• /tailor — paste a JD, get a tailored resume
• /interview — name a company, get a full prep doc
• /network — name a company, get contacts and outreach drafts

And a one-time onboarding wizard:

• /setup — authenticates Google, then walks you through resume, stories, profile, outreach style, and network context

---

Getting started

1. Install the plugin

If you're using Cowork (the Claude desktop / web app — most classmates):

1. Go to the latest release and download get-me-a-job.plugin.
2. In Cowork, open the plugin browser (Settings → Plugins → "Add custom plugin" or "Upload .plugin file").
3. Select the downloaded file and confirm.

If you're using Claude Code (the terminal CLI):

/plugin marketplace add kope-kope/haas-job-plugins
/plugin install get-me-a-job@haas-job-plugins

2. Run /setup

That's it. /setup handles everything else:

1. Connect Google — opens your browser, you sign in with @berkeley.edu, click Allow once. Drive, Docs, and Gmail are all granted in a single consent screen.
2. Resume — upload your formatted resume (.docx or Google Doc link). /setup parses it for the skills to use, creates a Job Search folder in your Drive, and saves your formatted doc as the master template.
3. Behavioral stories — build 3–5 STAR stories conversationally, or upload existing ones.
4. Job search profile — target roles, industries, geographies, seniority, strengths.
5. Outreach style — how you want your messages to sound.
6. Network context — communities, alumni networks, former colleagues.
7. Connector check — verifies Drive/Docs/Gmail are working.
8. Smoke test — tailors a real JD end-to-end to confirm everything's wired up.

You can stop at any step and resume later — /setup is idempotent.

3. (Optional) Connect Hunter.io for email finding

Free tier: 25 searches/month. Add your Hunter API key as a connector in Cowork. Without it, outreach falls back to LinkedIn and email pattern guessing — still works fine.

---

How it works

Every skill reads from personal reference files that /setup creates under your home directory:

~/.claude/get-me-a-job/
├── credentials.json                ← OAuth tokens (from Google sign-in)
├── config.json                     ← Drive folder ID, master resume Doc ID
└── references/
    ├── resume.md                   ← your master resume
    ├── stories.md                  ← your STAR stories
    ├── profile.md                  ← your job search targeting
    ├── outreach-style-guide.md     ← your outreach preferences
    └── network-context.md          ← your network map

Your data lives in your home directory, not inside the plugin install. Updating or reinstalling the plugin doesn't touch any of it.

The Python helpers in lib/ (gdocs.py, gmail.py, google_auth.py) are invoked through lib/run.py, which picks the best Python install path automatically:

1. uv if available — uses PEP 723 inline deps, zero install
2. Existing pip install — just runs
3. Auto pip install — installs requirements.txt once on first use, then runs

Works the same on macOS, Linux, and Windows.

---

Safety rules

• No emails sent without your approval. Outreach is drafted first, you see it in chat and (after Gmail-draft) in your inbox, and only an explicit "send it" triggers delivery.
• No fabricated metrics. If a resume bullet needs a number, the plugin asks you for the real one instead of inventing one.
• Humanizer pass on everything. Cover letters, outreach messages, and interview prep docs all go through the humanizer before you see them. Recruiters spot AI writing instantly.

---

Examples

The examples/ folder shows what your reference files will look like after /setup: