name: review-saved-jobs
description: Review saved LinkedIn jobs systematically. Use when user says "review saved jobs", "check saved jobs", "go through my saved jobs", or wants to batch process their LinkedIn saved jobs list.
requires: browser-automation
platform_note: Requires browser automation with LinkedIn login. Works on Claude Code (Chrome MCP) and OpenClaw (browser tool).

Review Saved Jobs

Systematically review LinkedIn saved jobs, quick-filter obvious passes, and run /evaluate-job for promising roles.

Prerequisites

• Chrome browser automation (Claude in Chrome MCP)
• User logged into LinkedIn in Chrome

Workflow

1. Open LinkedIn Saved Jobs

tabs_create_mcp → create new tab
navigate to https://www.linkedin.com/my-items/saved-jobs/
wait 2 seconds for page load
screenshot to see current list

2. Get Job List

Take screenshot and identify visible jobs. Note:
- Job title
- Company name
- Posted date (freshness)
- "Easy Apply" badge if present
- "Actively reviewing applicants" if shown

3. Load Context (parallel)

• Read [[Job Hunting]] for:
• Applied Jobs list (avoid duplicates)
• Passed On list (already evaluated)
• Anti-Signals patterns
• Pipeline health (active count)
• Read user's CLAUDE.md for background/criteria

4. Quick-Filter by Title

Before clicking into each role, quick-filter visible titles:

5. Process Each Promising Role

For roles that pass quick-filter:

1. Click on job title to open details panel
2. Wait for load (2 seconds)
3. Get the job URL from the browser tab
4. Chain to /evaluate-job:
5. Invoke the skill with the LinkedIn job URL
6. /evaluate-job handles: full JD extraction, fit analysis, vault note creation, Job Hunting updates
7. Wait for evaluation to complete before moving to next role
8. Record outcome in session tracking (APPLY/CONSIDER/PASS)

6. Track Progress

Use todo list to track progress through saved jobs:

- [ ] Role A - Company X
- [x] Role B - Company Y → PASS (too junior)
- [x] Role C - Company Z → APPLY (drafted email)
- [ ] Role D - Company W

7. Summary Output

After processing batch, output summary:

Stats:
- Reviewed: X roles
- APPLY: Y
- CONSIDER: Z
- PASS: W

8. Handle APPLY Decisions

For APPLY recommendations:
- Easy Apply: Ask "Proceed with application now?"
- External ATS: Add to "To Apply" list with note

Scrolling Through Long Lists

If saved jobs list is long:
1. Process visible jobs first
2. Scroll down: scroll action with scroll_direction: down
3. Take new screenshot
4. Continue processing new visible jobs
5. Repeat until reaching previously evaluated roles or end of list

Quick Commands

Chaining with /evaluate-job

This skill invokes /evaluate-job for each promising role after quick-filtering:

1. Quick-filter identifies roles worth evaluating (by title, company, freshness)
2. For each promising role, extract the LinkedIn job URL
3. Call /evaluate-job <url> — this handles:
4. Full JD extraction
5. Fit analysis across all dimensions
6. Anti-signal checking
7. Vault note creation (mandatory)
8. Job Hunting tracking updates
9. Record the outcome and move to next role

Division of labor:
- /review-saved-jobs — Batch processing, quick-filtering, session tracking
- /evaluate-job — Deep evaluation of individual roles

Related Skills

• /evaluate-job — Deep evaluation of single job posting
• chrome-automation — Reference for Chrome browser best practices