Quickstart¶
Get from zero to your first automated LinkedIn post in under 5 minutes.
1. Install (30 seconds)¶
Verify:
Expected output:
linkedin-mcp-pro v2.3.1 — ready
Python 3.11+
DB: ~/.linkedin-mcp/profile/state.db
Browser: available
2. Start the dashboard (5 seconds)¶
Open http://localhost:8080.
3. Connect LinkedIn (60 seconds)¶
Open http://localhost:8080/connect and pick a method:
- Click Launch browser
- A Playwright window opens — log into LinkedIn normally
- Once the dashboard shows your profile, you're connected
- Cookies are saved to
~/.linkedin-mcp/profile/cookies.json(encrypted at rest with your passphrase)
- Open LinkedIn in your own browser
- DevTools → Application → Cookies →
li_at→ copy value - Paste it in the li_at field on the connect page
- Click Save
4. Compose your first draft (45 seconds)¶
Open http://localhost:8080/drafts.
The composer has three modes:
| Mode | What it does | When to use |
|---|---|---|
| Manual | Rich-text editor with live preview | When you want full control |
| AI assist | LLM drafts from a 1-line prompt | When you're out of ideas |
| Template | Pick from saved post templates | When you have a repeatable pattern |
Write your post, hit Save draft, then Post now (subject to safety gates).
5. (Optional) Wire an AI agent¶
Restart your agent. It will discover 54 new MCP tools automatically.
6. (Optional) Try the jobs module¶
Open http://localhost:8080/jobs.
- Setup tab → upload your CV (PDF or DOCX)
- Profile tab → fill out skills, target roles, salary range
- Search tab → click Find jobs (uses LinkedIn search with your filters)
- Review match scores → click Apply on ones you like → cover letter auto-generated
- Track status in Tracker tab
What now?¶
- Architecture overview — understand how the pieces fit
- Safety system — read this before going autonomous
- Dashboard tour — every page explained
- MCP tools reference — full tool inventory