/apply

name: apply
description: Fill out a job application on Greenhouse, Lever, or Workday
argument-hint: "job URL, 'last' to use most recent job, or 'current' to fill the active browser tab"

Apply Skill

> **Priority hierarchy**: See `shared/references/priority-hierarchy.md` for conflict resolution.

Fill out job application forms on Greenhouse, Lever, and Workday using browser automation.

Quick Start

• `/proficiently:apply` - Start the flow (will ask for a job URL or use the most recent job)
• `/proficiently:apply https://...` - Apply to a specific job posting
• `/proficiently:apply last` - Apply using the most recent job folder
• `/proficiently:apply current` - Fill the application form already open in the active browser tab

File Structure

scripts/
  fill-page.md              # Form-filling subagent prompt

Data Directory

Resolve the data directory using `shared/references/data-directory.md`.

---

Workflow

Step 0: Check Prerequisites

Resolve the data directory, then check prerequisites per `shared/references/prerequisites.md`. Resume file is required. Load `DATA_DIR/application-data.md` if it exists (created in Step 2 if not).

Step 1: Determine Target Job

Parse `$ARGUMENTS`:

**If a URL:**

• Check if a matching job folder exists in `DATA_DIR/jobs/` (match by company slug in folder name or by URL). If found, load `posting.md`, `resume.md`, `cover-letter.md` from that folder.
• If no match, set up browser per `shared/references/browser-setup.md`, fetch the posting, save it to a new folder at `DATA_DIR/jobs/[company-slug]-[date]/posting.md`.

**If "last" or empty:**

• Find the most recently modified job folder in `DATA_DIR/jobs/`
• Load its `posting.md`, `resume.md`, `cover-letter.md`
• Confirm with the user which job this is for

**If "current":**

• Skip navigation — will use the current browser tab as-is
• Match the tab's URL against saved job folders to load context if possible

Report what's loaded:

Applying to [Role] at [Company].

Step 2: Build/Load Application Data

If `DATA_DIR/application-data.md` exists, read it and load the values.

If it does NOT exist: 1. Extract what you can from the resume: name, email, phone, LinkedIn, location 2. Present extracted data to the user. Ask them to confirm and fill in gaps: work authorization, visa sponsorship, EEO preferences (default all EEO to "Decline to self-identify") 3. Save to `DATA_DIR/application-data.md` using this format:

# Application Data

## Personal Information
- First Name: ...
- Last Name: ...
- Email: ...
- Phone: ...
- City: ...
- Country: United States

## Online Profiles
- LinkedIn: ...
- GitHub: ...
- Portfolio: ...

## Standard Answers
- How did you hear about us: Job Board
- Previously worked at this company: No
- Authorized to work in the US: Yes
- Requires visa sponsorship: No

## EEO / Voluntary Disclosures
- Gender: Decline to self-identify
- Race/Ethnicity: Decline to self-identify
- Veteran status: I am not a veteran
- Disability: I don't wish to answer

Step 3: Navigate to Application Form and Scout Requirements

Set up browser per `shared/references/browser-setup.md` (`tabs_context` → `tabs_create` → `navigate`).

**If `$ARGUMENTS` is "current"**: Skip navigation. Call `tabs_context_mcp` to get the active tab.

**Otherwise**, detect ATS type from URL patterns (see `shared/references/ats-patterns.md`) and navigate accordingly:

**Lever** (`jobs.lever.co/...`):

• Navigate to the posting URL with `/apply` appended, or navigate to the posting and click "APPLY FOR THIS JOB"

**Greenhouse** (`boards.greenhouse.io/...` or page with `grnhse_iframe`):

• Navigate to the posting URL
• Extract iframe tokens via `javascript_tool`:

  const iframe = document.getElementById('grnhse_iframe');
  const url = new URL(iframe.src);
  JSON.stringify({
    boardToken: url.searchParams.get('for'),
    jobToken: url.searchParams.get('token')
  });

• Navigate to direct form URL: `https://job-boards.greenhouse.io/embed/job_app?for={boardToken}&token={jobToken}`

**Workday** (`*.myworkdayjobs.com/...`):

• Navigate to the posting. Click "Apply Now".
• If a landing page appears with Autofill/Manual options, click "Apply Manually".
• If an auth gate appears, **tell the user to sign in, then say "continue" when ready**. Account creation is a prohibited action — the user must handle authentication themselves.

**Unknown ATS**:

• Navigate to the URL, take a screenshot
• Attempt to identify the form. If unrecognizable, tell the user and ask for guidance.

**Scout the form.** Once on the application form, do a quick scan (`read_page(filter="interactive")` or scroll through for Workday) to determine:

• Does the form have a **resume/CV upload** field?
• Does the form have a **cover letter** upload or text field?
• Are there any **unusual required fields** that need special attention?

Record these requirements — they determine what materials to generate in Step 4.

Step 4: Generate Missing Materials

The goal is to have everything ready before filling, so the user does minimal work.

**Always tailor the resume.** Check if `DATA_DIR/jobs/[job-folder]/resume.md` exists for this job:

• If YES: the resume is already tailored for this role. Skip.
• If NO: Run the tailor-resume skill inline. Follow the workflow in `skills/tailor-resume/SKILL.md` — use the job posting (already loaded), the original resume, and the work history profile to generate a tailored resume. Save to the job folder. Present it to the user for quick review before continuing.

**Generate a cover letter only if the form requires one.** If the scout in Step 3 found a cover letter field:

• Check if `DATA_DIR/jobs/[job-folder]/cover-letter.md` exists
• If YES: already done. Skip.
• If NO: Run the cover-letter skill inline. Follow the workflow in `skills/cover-letter/SKILL.md` — use the posting, tailored resume, and profile. Save to the job folder. Present it for quick review.

**If the f