job-batch

name: job-batch description: Process a batch of job applications - discover roles, prepare materials, pre-fill forms. Use when ready to apply to multiple positions.

Job Application Batch Processor

Slash command: /job-batch

Purpose

Process a batch of job applications using the relationship-focused methodology. This is NOT mass-applying - each role gets proper research.

CRITICAL: This is Phase 6

Per the /apply skill methodology, formal applications are the LAST step, not the first.

For each role in the batch, we do:

1. Company intel (challenges, pain points, recent news)
2. People mapping (hiring manager + champions)
3. Email pattern detection
4. Outreach draft (user sends)
5. THEN formal application (user submits)

Batch processing doesn't skip steps - it parallelizes them.

Workflow

Phase 1: Discovery

1. Search for roles matching CLAUDE.md criteria
2. Verify each role is still open via Playwright
3. Check applications/history.json for duplicates
4. Create batch of 10 candidates

Phase 2: User Selection

1. Present batch with compensation, location, fit assessment
2. User selects which to proceed with (typically 3-5, not all 10)
3. Create folders in applications/pending/{company}_{role}_{date}/

Phase 3: Company Intel (NEW - REQUIRED)

For each selected role:

1. Research recent news, funding, product launches
2. Identify challenges and pain points
3. Find engineering blog, tech stack info
4. Document in notes.md:
   • 3-5 bullets on what they care about RIGHT NOW
   • Why this role exists (growth? backfill? new initiative?)

Phase 4: People Mapping (NEW - REQUIRED)

For each selected role:

1. Hiring Manager: Who owns this headcount? LinkedIn search
2. Champions: 2-3 employees with aligned interests (publications, talks, GitHub)
3. Mutual Connections: Does user know anyone who knows them?
4. Document in notes.md:
   • Names, titles, backgrounds, LinkedIn URLs
   • Email patterns (use email_finder.py)

Phase 5: Outreach Drafts (NEW - REQUIRED)

For each selected role, draft in outreach_draft.md:

1. Informational email to a champion (if genuine interest exists)
2. Or hiring manager email (after application)
3. Follow /apply skill templates - specific, about THEM, easy yes/no

User sends these - Claude NEVER sends.

Phase 6: Materials Preparation

1. Navigate to each job posting via Playwright
2. Extract full job description to notes.md
3. Generate cover letter following /anti-ai-writing guidelines
4. Fill any application-specific questions in responses.md

Phase 7: Form Pre-fill

1. Open each application in separate Playwright tab
2. Fill standard fields:
   • Name, email, phone, location
   • LinkedIn, GitHub URLs
   • Resume upload (CV PDF)
   • Cover letter (paste or upload)
3. Leave tabs open for user review
4. DO NOT submit - wait for user

Phase 8: Report

Generate summary with:

• Tab index for each application
• Company, role, compensation
• Hiring manager identified: Yes/No
• Outreach draft ready: Yes/No
• Cover letter preview
• Any fields that couldn't be auto-filled

Output Per Role

Each applications/pending/{company}_{role}_{date}/ folder contains:

├── notes.md           # JD + company intel + people mapping
├── cover_letter.md    # Generated cover letter
├── outreach_draft.md  # Email to champion or hiring manager
└── responses.md       # Application-specific questions (if any)

The Test

Before proceeding with any role, ask:

"Could I email someone at this company even if they had no job openings?"

If the answer is "no" - the research isn't done yet.

Job Board Handling

Greenhouse (AI2, many others)

• Standard form structure
• Resume upload, cover letter text area
• May have custom questions

Ashby (Genesis, Insitro)

• Clean modern forms
• Often has "How did you hear about us?"
• Resume upload supported

Workday (Flatiron/Simons, PacBio)

• Complex multi-step forms
• May require account creation
• Often has duplicate entry issues

Lever

• Simple single-page forms
• Resume + cover letter + optional fields

Interfolio (Academic)

• Academic-specific
• May require separate documents (CV, research statement)

Standard Field Values

Read from user's .claude/CLAUDE.md and profile.json:

• Name, email, phone, location
• LinkedIn, GitHub URLs
• Resume path (typically documents/*.pdf)

Users must configure these files before using batch processing. See GETTING_STARTED.md.

Anti-AI Writing Checklist

Before finalizing any cover letter, verify:

• No "delve," "tapestry," "multifaceted," "pivotal"
• Em dashes < 2
• Contains at least one contraction
• Varied paragraph lengths
• Specific technical details
• No generic company praise

See: /anti-ai-writing skill

Final Report Format

## Applications Ready for Review

| Tab | Company | Role | Comp | HM Found | Outreach Ready |
|-----|---------|------|------|----------|----------------|
| 1 | Flatiron | Sr SWE Lead | $190-235K | ✓ | ✓ |
| 2 | Genesis | ML Scientist | Competitive | ✓ | ✓ |
...

### Tab 1: Flatiron Institute
- URL: [link]
- Hiring Manager: [Name, Title]
- Champion for outreach: [Name] - [why interesting]
- Cover letter preview: [first 200 chars]
- Outreach email preview: [first 100 chars]
- Unfilled fields: None

[etc for each tab]

### Next Steps
1. Review outreach drafts - send if genuine interest
2. Review each application tab
3. Edit any fields as needed
4. Submit when ready
5. Follow up per `/apply` timing guidelines