LinkedInalyzer - AI-Powered LinkedIn Feed Cleanup Tool
3 min read
LinkedInalyzer analyzes your LinkedIn feed for political content and AI-generated slop, then gives you a dashboard to decide who to unfollow or disconnect.
Why:
After 20 years on LinkedIn, the network accumulates people who've drifted from professional posting to political commentary and AI engagement bait. LinkedInalyzer identifies those connections so you can clean up your feed.
How it works:
• Feed scraper — Playwright scrolls through your feed, extracting posts, authors, and how each one reached you (suggested, liked by a connection, reposted, promoted, etc.)
• Gemini analysis — each post goes through Google Gemini 2.5 Flash with structured scoring for political content (0.0–1.0) and AI slop probability (0.0–1.0)
• SQLite storage — everything persisted locally for historical tracking
• Dashboard — FastAPI + React app with a filterable author list, direct profile links for quick unfollowing, and feed-context tracking
• Analytics — score histograms, feed-context donut, scrape timeline, top offenders, and an amplifier graph showing which connections spread flagged content
Scoring:
Political score goes from 0.0 (completely non-political) to 1.0 (overtly partisan). AI slop score flags generic engagement bait vs. original human writing. Posts scoring ≥ 0.4 political are flagged for review.
Stack: Python 3.11, Playwright, SQLAlchemy, FastAPI, Click on the backend. React 18 + TypeScript + Vite + Tailwind + TanStack Query + Recharts on the frontend. AI via
Workflow:
Note: LinkedIn's ToS prohibits scraping. This is built for personal feed management, not bulk data collection — use responsibly.
Repo: github.com/nerdymark/LinkedInalyzer Read more...
Why:
After 20 years on LinkedIn, the network accumulates people who've drifted from professional posting to political commentary and AI engagement bait. LinkedInalyzer identifies those connections so you can clean up your feed.
How it works:
• Feed scraper — Playwright scrolls through your feed, extracting posts, authors, and how each one reached you (suggested, liked by a connection, reposted, promoted, etc.)
• Gemini analysis — each post goes through Google Gemini 2.5 Flash with structured scoring for political content (0.0–1.0) and AI slop probability (0.0–1.0)
• SQLite storage — everything persisted locally for historical tracking
• Dashboard — FastAPI + React app with a filterable author list, direct profile links for quick unfollowing, and feed-context tracking
• Analytics — score histograms, feed-context donut, scrape timeline, top offenders, and an amplifier graph showing which connections spread flagged content
Scoring:
Political score goes from 0.0 (completely non-political) to 1.0 (overtly partisan). AI slop score flags generic engagement bait vs. original human writing. Posts scoring ≥ 0.4 political are flagged for review.
Stack: Python 3.11, Playwright, SQLAlchemy, FastAPI, Click on the backend. React 18 + TypeScript + Vite + Tailwind + TanStack Query + Recharts on the frontend. AI via
google-genai SDK.Workflow:
linkedinalyzer login once to save a persistent Chromium profile, then linkedinalyzer scrape whenever you want to pull new posts. The dashboard lets you mark each author as Pending → Reviewed → Keep / Unfollowed / Disconnected.Note: LinkedIn's ToS prohibits scraping. This is built for personal feed management, not bulk data collection — use responsibly.
Repo: github.com/nerdymark/LinkedInalyzer Read more...