Skip to content

aintoniodev/linkedin-profile-archive

v1.0.0MIT

Archive a LinkedIn profile's full posting history (text + images + engagement metrics) into a local folder for offline analysis. Drives a logged-in Chromium session over CDP to capture LinkedIn's internal voyager API, paginates until the last post, and downloads images through the browser. Ships capture_feed.py, fetch_all_pages.py, linkedin_archive.py, download_images.py, linkedin_report.py.

GitHubOpen Website

linkedin-profile-archive

Archive a LinkedIn profile's entire posting history — text, images, and engagement metrics — into a structured local folder for offline analysis. No premium API.

This is an Agent Plugin (Agent Plugins 1.0.0) that ships one skill (skills/linkedin-profile-archive/SKILL.md) plus executable scripts.

What it does

  • Captures LinkedIn's internal voyager/api activity-feed request from a live, logged-in browser (Chrome/Brave) via CDP — no manual curl copying, no cookies stripped from the profile.
  • Paginates through every post using the feed's paginationToken (the start offset alone returns the same first page every time).
  • Extracts each post's text, URL, date, and like/comment/repost counts into index.json + one posts/*.md per post.
  • Downloads each post's image through the browser (media.licdn.com 403s plain HTTP clients via Cloudflare TLS fingerprinting) with resume + reconnect.
  • Summarizes cadence, top posts, engagement, and topics into a REPORT.md.

Why the browser path

LinkedIn blocks plain HTTP scraping and encrypts session cookies with Chromium v20 "app-bound" protection, so they don't transfer to a copied profile and can't be replayed by curl. The robust route is to drive the user's real, currently logged-in browser.

Install

Copy this directory (plugin) anywhere, or copy just the skill:

cp -r skills/linkedin-profile-archive ~/.pi/agent/skills/linkedin-profile-archive

Prereqs:

pip install websocket-client requests

Quick pipeline

# 1. start the logged-in browser with CDP on :9222 (your real profile dir)
brave --remote-debugging-port=9222 --user-data-dir="$HOME/AppData/Local/.../User Data"

# 2. capture the feed request
python skills/linkedin-profile-archive/scripts/capture_feed.py \
  --url "https://www.linkedin.com/in/<PUBID>/recent-activity/all/" --out session/capture

# 3. (edit session/capture/curl.txt to the count:20,start:0,profileUrn feed) then paginate
python skills/linkedin-profile-archive/scripts/fetch_all_pages.py \
  --endpoint session/capture/curl.txt --out session/pages

# 4. extract posts
python skills/linkedin-profile-archive/scripts/linkedin_archive.py \
  --raw session/pages --out output/<PUBID>

# 5. download images
python skills/linkedin-profile-archive/scripts/download_images.py \
  --images output/<PUBID>/images.json --dir output/<PUBID>/images

# 6. analyze
python skills/linkedin-profile-archive/scripts/linkedin_report.py \
  --index output/<PUBID>/index.json --md output/<PUBID>/REPORT.md

See SKILL.md for full detail and pitfalls.

Components

ScriptRole
capture_feed.pyNavigate the activity page + capture voyager requests via CDP
fetch_all_pages.pyPaginate the feed with paginationToken until the last post
linkedin_archive.pyExtract posts (text/metrics/images manifest) from raw pages; --curl replay for replayable sessions
download_images.pyBinary-download post images through the browser (CDP)
linkedin_report.pyCadence / top-posts / topics / engagement summary

Output

output/<PUBID>/
├── index.json      # every post (id, date, url, text, metrics, images)
├── images.json     # {url, file} manifest for download_images.py
├── posts/          # NNNN-YYYY-MM-DD-slug.md
├── images/         # downloaded JPEGs
└── REPORT.md       # analysis summary (optional)

License

MIT