An AI agent skill that helps you scrape LinkedIn pages reliably.
- 🎯 Right company, first try — never grabs a feed post or school by mistake
- 🔁 Survives LinkedIn's reshuffles — works across every known page variant
- 💸 $0 added cost — runs inside the AI you already use
- 🧩 No setup — one Markdown file, no packages, no build
The skill starts from one fact most AIs miss: LinkedIn serves a different layout of the same page on each load, so it checks every known variant and picks by position, not by class name.
- 🏢 Current company from any profile — a ready-to-run scraper for all three top-card layouts
- 💼 Jobs pages mapped — stable anchors for job cards, details, filters and the scroll container
- 📡 Data the page never rendered — job descriptions and company websites via LinkedIn's own internal API
- 🚫 Dead ends skipped — no time wasted on meta tags, JSON-LD or old data blobs
- 🩺 A debugging routine that finds the real cause — probe the live tab, list candidates, test on fresh loads
Needs an AI assistant that supports Agent Skills and a Chromium browser you're logged in to LinkedIn with. For scripts that drive Chrome, turn on View ▸ Developer ▸ Allow JavaScript from Apple Events.
# Claude Code
git clone https://github.com/adriangrantdotorg/linkedin-ai-skill.git ~/.claude/skills/linkedin-ai-skill
# Cursor
git clone https://github.com/adriangrantdotorg/linkedin-ai-skill.git ~/.cursor/skills/linkedin-ai-skill
# ChatGPT & Codex
git clone https://github.com/adriangrantdotorg/linkedin-ai-skill.git ~/.agents/skills/linkedin-ai-skill| Platform | Skills folder |
|---|---|
| Claude Code | ~/.claude/skills/ |
| Cursor | ~/.cursor/skills/ |
| ChatGPT & Codex | ~/.agents/skills/ |
Ask for LinkedIn data as usual; the skill kicks in on its own.
| You say | The skill makes |
|---|---|
| "Get the current company from this LinkedIn profile." | A JavaScript scraper that checks every layout and returns the top-card company |
| "Grab the full description of this LinkedIn job, even in a background tab." | A same-origin API call with the right headers, cached and failing softly |
| "My LinkedIn scraper suddenly returns the wrong company." | A debugging run against the live tab that finds the new variant |
Built with ❤️ for the AI automation community & everyone tired of broken selectors ✌🏾
