Key Finder analyzes LinkedIn job postings to surface the most in-demand skills and qualifications for any job title. **How it works:** 1. Constructs a LinkedIn search URL from the input 2. Scrapes up to 10 pages (~25 jobs each) 3. Fetches the full description for each posting 4. Extracts single words, two-word phrases, and three-word phrases 5. Filters noise via a customizable `noise.txt` file (common English words, HR boilerplate, legal language) 6. Counts how many postings contain each keyword and calculates the percentage 7. Returns results sorted by frequency Analysis runs in a background thread. The frontend polls `/progress` every few seconds with AJAX to show live updates — current page, jobs found, most recent title. Results export to CSV (keywords with frequency, or full job listings). **Interface:** Pure HTML/CSS/JS, black monospace theme, no framework or build step. Rate-limited with intentional delays (1s between job fetches, 2s between pages) to be polite to LinkedIn's servers.