job-apply

name	job:apply
description	Apply to the next suitable job from your queue. Picks the next pending role from job_search.md, verifies the posting is live, fetches the JD, presents the link and how to apply — then (after user confirms) adapts the CV and generates a PDF. Updates status and source stats after user confirms they applied.
argument-hint	[optional: company name or # to apply to a specific role]
user-invocable	true
disable-model-invocation	false

Apply Next

Job tracking file

~/.claude/job_search.md

CV skill

job:cv — call after user confirms they want to apply.

Workflow — follow this order exactly

Step 1 — Pick the next role

Read job_search.md:

• Find the first role with status = pending in the ## Status table, following the queue order in ## Queue.
• If an argument was given (company name or #), use that role instead.

Step 2 — Verify the posting is live

Before doing anything else with the CV:

1. Find the apply URL for the role (from the Notes column, or search if not listed: "<Company>" "<Role>" job site:jobs.ashbyhq.com OR site:lever.co OR site:greenhouse.io OR site:wellfound.com).
2. Fetch or search the URL to confirm the job page shows a real posting — NOT "Job not found", "No jobs found", or a 404.
3. If closed/removed:
   • Update job_search.md: set status to closed, add reason in Notes, mark [x] in Queue.
   • Do not adapt the CV. Announce "Closed — moving to next" and repeat from Step 1 with the next pending role.
4. If open: continue to Step 3.

Step 3 — Extract the JD

Fetch the job description:

• Requirements, responsibilities, tech stack, salary range, timezone/location policy.
• Note any "not a fit" criteria that might be a blocker (US-only visa, specific language required, etc.).
• If there's a blocker → close the role and move on (same as Step 2 closed case).

Step 4 — Present to user (STOP here and wait)

Show:

**#N — Company — Role**
💰 Salary: ...
🌍 Location: ...
🛠 Stack: ...
✅ Why it fits: ...
🔗 Apply: <URL>
📋 How to apply: [describe: form / email / LinkedIn / etc.]

Then ask: "Shall we apply?"

Do NOT start adapting the CV until the user confirms. Wait for confirmation ("yes", "go", "let's do it", or similar).

Step 5 — Adapt the CV

After user confirms:

1. Invoke job:cv with the company name and key JD signals as argument.
2. job:cv handles: archive → edit → generate PDF → open in Preview.
3. Show the user the diff summary (what changed in the CV).

Step 6 — Wait for apply confirmation

After CV is ready, ask: "Did you apply?"

When user confirms (says "yes", "sent", "applied", etc.):

1. Update job_search.md Status table:
   • Set status to applied
   • Add date to Notes: YYYY-MM-DD
   • Mark [x] in Queue
2. Update ## Sources — Effectiveness Tracker:
   • Increment Applied for the role's source
3. Confirm: "Done, marked as applied. Next?"

Step 7 — Offer to continue

Ask if they want to move to the next role or stop.

---

Status update rules

Event	What to update
Role closed/removed	status → closed, mark [x] in Queue, add reason to Notes
Role skipped (stack mismatch, geo block, etc.)	status → skipped, mark [x] in Queue, add reason to Notes
CV adapted, waiting to apply	status stays pending (do NOT mark applied yet)
User confirms they applied	status → applied, add date, increment source Applied count
Got response (recruiter reached out)	status → response, update Notes, increment source Response count
Got interview invite	status → interview, update Notes, increment source Interview count

CRITICAL: Never mark applied until user explicitly confirms they submitted the application.

---

Queue order

Follow the order in ## Queue. Top = highest priority. Unchecked [ ] = pending.

---

Notes on common apply flows

• Ashby form (jobs.ashbyhq.com): fill form on page, attach PDF CV
• Email (e.g. [e-mail]): send email with CV attached, subject "Senior [Role] — [Your Name]"
• LinkedIn Easy Apply: use LinkedIn, attach PDF
• Greenhouse / Lever: fill form, attach PDF