Setup Wizard

Setup Wizard

Guide users through first-time repository configuration.

More Input = Better Output

The system's quality is directly proportional to the reference materials provided. Your job is to extract EVERYTHING the user has. Old CVs, cover letters, LinkedIn, project docs, performance reviews. All of it.

The user does NOT fill in CANDIDATE-PROFILE.md manually. You build it from their materials.

Phase 0: Install Playwriter Extension

This enables browser automation for filling lengthy, multi-page application forms.

Many companies use applicant tracking systems (Workday, Greenhouse, Lever) that require filling the same information across 5-10 pages. Playwriter lets the agent do this for you.

1. Direct user to install:

   • URL: https://chromewebstore.google.com/detail/playwriter-mcp/jfeammnjpkecdekppnclgkkffahnhfhe

2. Instruct them to:

   • Pin the extension to Chrome toolbar (click puzzle icon, then pin)
   • Gray icon = not connected
   • Green icon = connected and ready

3. Explain the use case:

   • "When you find a job with a long application form, click the Playwriter icon on that tab."
   • "I can then fill in the form using your profile data."
   • "You review before submitting. I never submit without your approval."

4. Confirm: "Is Playwriter installed and pinned?"

Phase 1: Collect Reference Materials

Be greedy. Ask for everything.

Required

1. "Paste your current CV here, or give me a file path."

   • Save to 01-Core-Materials/CVs/[Name]-CV-Current.md

2. "Do you have older CV versions? Different formats for different roles?"

   • Save each to 01-Core-Materials/CVs/[Name]-CV-[Version].md

3. "Copy and paste your LinkedIn profile. The whole thing: headline, about, experience, skills, recommendations."

   • Save to 01-Core-Materials/Portfolio/[Name]-LinkedIn-Profile.md

Highly Valuable

4. "Any cover letters you're proud of?"

   • Save to 01-Core-Materials/Cover-Letters/[Name]-CL-[Company].md

5. "Portfolio pieces, case studies, or project write-ups?"

   • Save to 01-Core-Materials/Portfolio/[Name]-[Project].md

Nice to Have

6. "Performance reviews, recommendation letters, award citations?"

   • Save to 01-Core-Materials/Portfolio/[Name]-[Type].md

7. "Old job descriptions from roles you've held?"

   • Save to 01-Core-Materials/Portfolio/[Name]-JD-[Role].md

Phase 2: Build CANDIDATE-PROFILE.md

You build this. The user verifies.

Read ALL collected materials in 01-Core-Materials/. Then populate CANDIDATE-PROFILE.md:

For each section:

1. Show what you synthesized: "From your materials, I built this: [section]"
2. Ask: "Anything to change or add?"
3. Save after confirmation

For gaps:

1. Point out what's missing: "I don't have [X]. Can you tell me?"
2. Ask directly, offer to draft from bullet points

Phase 3: Capture Voice and Preferences

1. "Looking at your cover letters, I notice you use phrases like [examples]. Is that your natural voice?"
2. "Any words or phrases I should NEVER use?"
3. Save to Writing Style section

Phase 4: Configure Agents

job-application-automator.md

• "Any style rules or pet peeves for your documents?"
• Update <user_preferences> block

researcher.md

• "What are your job search parameters? (remote/hybrid, industries, salary range, location)"
• Update <user_context> block

Phase 5: Verify and Handoff

1. Summarize:

   • Files collected in 01-Core-Materials/
   • CANDIDATE-PROFILE.md status
   • Agent configurations

2. Offer: "Want me to scan for any gaps in your profile?"

3. Explain next steps:

   • job-application-automator for CVs and cover letters
   • researcher for company research
   • /setup to return here anytime

Before completing setup, verify:

• At least one CV saved
• LinkedIn profile saved (strongly encourage)
• CANDIDATE-PROFILE.md fully populated (no [brackets] in key sections)
• Contact info complete
• At least 2 work experiences documented
• Skills populated with real skills from materials
• Writing Style section captures user's actual voice
• Agent preferences configured