Process Work History

• When scanning files: "Found [N] files — [list types]. I'll start with your resume since that's the highest-value artifact for building the timeline."
• When extracting: "Reading your resume now — I'm pulling out companies, dates, titles, and key bullet points. I'll preserve your own language from the resume bullets since that's how you've already positioned this work."
• When writing summaries: "Creating a structured summary for [Company]. This organizes everything from the resume + any other docs into a format that other skills can reference — the job search skill uses this for fit assessment, and the case study skill draws from it."
• When building the timeline: "Building your canonical work timeline — this is the master reference for your career arc."
• When identifying gaps: "Now checking for gaps — missing years, companies with no detail, projects mentioned but not captured."

Step 1: Where Do the Files Live?

Ask the user where their source documents are. Common patterns:

• Already in Work-History/[company]/ — they've organized per-company already
• In a single Work-History/inbox/ folder — dumped together, unsorted
• In an external folder they'll tell you about (e.g., ~/Downloads/career-docs/)
• Nowhere yet — they need to gather files

If they haven't gathered anything, give them clear instructions:

To get the most out of this skill, drop the following into Work-History/inbox/ (I'll create the folder if it doesn't exist):

• Your current resume (PDF, .docx, or .md)
• Your LinkedIn profile export (Settings → Data Privacy → Get a copy of your data → "Want something in particular?" → pick at least "Profile")
• Any performance reviews you have from past jobs
• Old project writeups, case studies, or presentation PDFs
• Portfolio docs or design files (I'll extract text where possible, otherwise note them for later)

Once they're in, come back and run /process-work-history again and I'll scan them.

If they already have files in per-company folders, skip this and move to Step 2.

Step 2: Inventory the Files

Once the user confirms files are in place:

1. List the source folder — use Glob or Bash ls
2. Categorize each file:
   • .pdf → process with pdftotext -layout (see Work-History/CLAUDE.md)
   • .md, .txt → read directly
   • .docx → try pandoc or textutil if available; otherwise note as "needs manual extraction"
   • Images (.png, .jpg) → optional, use vision to scan for text if the user wants
   • Design files (.fig, .sketch) → Skip, note as "design file, not processable"
   • Videos / audio → Skip
3. Create or update SOURCE_TRACKER.md in the target folder with status for each file

Step 3: Extract in Batches

Process 3-6 related files at a time. For each batch:

1. Read / extract the text
2. Identify:
   • Company name and dates
   • Role / title
   • Projects mentioned (with dates where possible)
   • Metrics (revenue, adoption, growth %, anything quantifiable)
   • Quotes / feedback worth saving for interview stories
   • Key decisions or turning points the user led
3. For each company found, ensure a folder exists: Work-History/[company]/summaries/
4. Create or update Work-History/[company]/summaries/[company]-overview.md following the template in Work-History/CLAUDE.md
5. If a project is substantial enough to deserve a deep-dive, create Work-History/[company]/summaries/[project].md and link it from the overview

Handling the resume specifically: The resume is the highest-value artifact. Process it first in its own batch. Extract:

• Full company list with dates (this becomes the canonical timeline)
• Role / title for each
• Top 2-3 bullet points per role (these are the user's own positioning language — preserve them verbatim)
• Any quantifiable metrics

Step 4: Build the Timeline

Once the first batch is processed, construct a canonical timeline from the resume + LinkedIn data. Save it to Work-History/timeline.md or display in conversation.

# Work Timeline

| Dates | Company | Role | Notes |
|-------|---------|------|-------|
| 2024-01 – present | [Current Company] | [Current Role] | |
| 2020-03 – 2023-12 | [Previous Company] | [Previous Role] | |
| ... | ... | ... | |

Reverse chronological. Flag any gaps (months or years with no entry) visibly in the Notes column.

Step 5: Identify Gaps

After the initial pass, do a gap analysis:

1. Timeline gaps — date ranges between companies with no coverage. Flag these and ask the user: sabbatical? Freelancing? Job search? Personal project?
2. Companies with no detail — a company appears in the resume but there's no Work-History/[company]/summaries/[company]-overview.md yet. Needs a summary.
3. Projects mentioned but not detailed — the resume says "Led redesign of onboarding, 40% activation lift" but there's no project deep-dive. Candidate for /project-interview.
4. Missing metrics — a project is described but no numbers. Ask if the user remembers any.
5. No recent entries — the most recent company has no current-state section or recent project details.

Present the gaps as a prioritized list:

Here's what I found and what's missing:

✅ Covered well:

• [Company A] — resume entry processed, 3 project deep-dives exist
• [Company B] — overview exists with metrics

⚠️ Needs a deep-dive:

1. [Company C] — [Project name] — mentioned in resume ("Led X"), no dossier yet
2. [Company D] — [Project name] — has short description, no outcomes captured

❌ Missing entirely:

1. [Company E] — [Dates] — appears in LinkedIn but no resume bullet or docs
2. Timeline gap: 2019-06 to 2020-02 — no entry

Want to fill any of these now? I can run /project-interview on each, or you can point me at more source docs.

Step 6: Offer to Fill Gaps

For each gap the user wants to fill:

1. Project gap → invoke /project-interview with the company + project name as context
2. Company gap → start a lightweight overview interview (just the basics: dates, role, what you did, key projects). Then offer /project-interview for the notable projects.
3. Timeline gap → ask what was happening during that period and capture a short note in timeline.md

Re-running the Skill

When the user re-runs /process-work-history:

1. Re-inventory the source folder — flag any new files since last run
2. Check SOURCE_TRACKER.md to see what's already been processed
3. Process new files only (skip Done entries)
4. Re-run gap analysis — gaps may have been filled by /project-interview runs since the last scan
5. Present updated gap report

Interview Style for Gap Filling

When interviewing the user to fill a gap, keep it light unless they want to go deep:

• Confirm the basics (company, dates, role, 1-line description of what the company does)
• Ask for 2-3 most notable projects or outcomes
• For each, ask: "Want me to do a full /project-interview on this, or is a short blurb enough for now?"

Let the user decide the depth. Not every past job needs a full dossier.

Output Location Summary

• Canonical timeline: Work-History/timeline.md
• Per-company overviews: Work-History/[company]/summaries/[company]-overview.md
• Source tracker per company: Work-History/[company]/summaries/SOURCE_TRACKER.md
• Project deep-dives: Work-History/[company]/summaries/[project].md
• Source files: Work-History/inbox/ or per-company folders (git-ignored)