LinkedIn CLI

Standalone CLI tool that extracts LinkedIn post analytics by clicking LinkedIn's own Export button and parsing the downloaded XLSX file. No DOM scraping, no API abuse. Everything runs locally.

What it does

Discovers your LinkedIn posts by scrolling through your activity feed (authenticated)
Captures analytics for each post via LinkedIn's Export to XLSX, then parses both sheets (Performance + Demographics)
Displays clean per-post analytics cards with impressions, reactions, comments, demographics, and audience breakdowns
Stores structured JSON snapshots locally in SQLite for history tracking and integration
Schedules follow up scans at 1h, 6h, 24h, 72h, and 7d intervals
All local. No backend, no cloud, no data leaves your machine

Requirements

Windows (10 or later)
Python 3.11+ (python.org/downloads)
A LinkedIn account

Step by step setup

1. Download and extract

Download the latest linkedin-cli-v*.zip from Releases and extract it to a folder on your PC.

2. Run setup

Double click setup.bat. This will:

Create a Python virtual environment (.venv)
Install all dependencies (playwright, openpyxl)
Install Chromium browser (needed by Playwright)
Initialize the local scanner database

Wait for "Setup complete!" before proceeding.

3. Log into LinkedIn

Double click login.bat. A Chrome browser window will open. Sign into your LinkedIn account, then close the browser. The scanner will auto detect your profile URL and save it.

Your login session (cookies) is saved locally in %LOCALAPPDATA%\LinkedInCLI\browser-profile\ and reused across runs. You only need to log in once.

4. Use the scanner

Double click scan.bat to open the interactive menu:

============================================
  LinkedIn CLI
============================================

  1. Discover + Scan (fresh import)
  2. Scan pending (process due queue)
  3. Status
  4. Pause / Resume

Choose (1-4):

First time: Choose 1 to discover all your posts and scan their analytics immediately.

Ongoing: Choose 2 to process any scans that are due (the scanner schedules follow ups at 1h, 6h, 24h, 72h, and 7d).

Check progress: Choose 3 to see how many scans are pending and when the next one is due.

CLI reference

If you prefer the command line over the interactive menu:

| Command | Description | |---------|------------| | linkedin-cli init | Create local SQLite database | | linkedin-cli login | Open browser to authenticate with LinkedIn | | linkedin-cli login --check | Verify existing LinkedIn session is valid | | linkedin-cli discover --limit 25 | Find posts on your profile, store in local DB | | linkedin-cli discover --limit 25 --exact-dates | Discover with precise date filtering | | linkedin-cli discover --headless | Discover with no visible browser window | | linkedin-cli import --scan-now | Discover + scan analytics immediately | | linkedin-cli import --limit 50 --scan-now | Full import up to 50 posts, scan right after | | linkedin-cli scan | Process due analytics captures | | linkedin-cli scan --force | Process ALL pending scans regardless of schedule | | linkedin-cli scan --headless | Run scans without visible browser window | | linkedin-cli status | Show queue state and last scan time | | linkedin-cli status --url-only | Print only the stored profile URL | | linkedin-cli pause / resume | Toggle scanning on/off | | linkedin-cli reset --force --reinit | Wipe everything and start fresh |

Output format (v2 capture payload)

{
  "schema_version": "linkedin_cli_snapshot_v2",
  "capture": {
    "capture_mode": "export",
    "capture_timestamp": "2026-05-14T22:00:00Z",
    "snapshot_window": "1h",
    "confidence": 0.98
  },
  "post": {
    "canonical_urn": "urn:li:activity:...",
    "top_job_title": "Frontend Developer",
    "top_location": "Karachi Division",
    "top_industry": "IT Services and IT Consulting"
  },
  "metrics": {
    "impressions": 105,
    "reactions": 3,
    "comments": 0,
    "reposts": 0,
    "members_reached": 29,
    "saves": 0,
    "sends": 0,
    "engagement_rate": 0.0286
  },
  "demographics": [
    {"category": "Company size", "value": "1-10 employees", "percentage": 41.4},
    {"category": "Job title", "value": "Software Engineer", "percentage": 28.6}
  ]
}

Where data is stored

Everything lives in %LOCALAPPDATA%\LinkedInCLI\:

| Path | Purpose | |------|---------| | scanner.sqlite | SQLite database (posts, scan queue, snapshots) | | config.json | Your LinkedIn profile URL and scanner settings | | browser-profile/ | Chromium profile with your LinkedIn session |

Run linkedin-cli reset --force --reinit to delete all of it.

Integration

The scanner stores all analytics as structured JSON in %LOCALAPPDATA%\LinkedInCLI\scanner.sqlite. Access it programmatically:

import sqlite3, json

db = sqlite3.connect(rf"{__import__('os').environ['LOCALAPPDATA']}\LinkedInCLI\scanner.sqlite")
db.row_factory = sqlite3.Row
snapshots = db.execute("SELECT * FROM snapshots ORDER BY captured_at DESC").fetchall()
for s in snapshots:
    payload = json.loads(s["payload"])
    print(payload["metrics"]["impressions"])

Or use the scanner's own Python modules:

from scanner.db import init_db, status_summary, due_scans
init_db()
summary = status_summary()
print(summary["pending_scans"], "scans pending")

Risk assessment

This scanner carries minimal risk for the following reasons:

No API abuse. It clicks LinkedIn's own "Export" button, the same one you would click manually. It does not hit any LinkedIn API endpoint, does not scrape the DOM with regex, and does not send unauthorized requests. Every action maps to a real user interaction.
No credentials stored in code. Your LinkedIn session lives in a local Chromium profile (%LOCALAPPDATA%\LinkedInCLI\browser-profile\), the same way Chrome remembers your logins. No passwords or tokens appear in the source code, config files, or database.
All data stays local. Analytics data is written to a SQLite database on your machine. Nothing is sent to any server, cloud, or third party. There is no telemetry, no analytics, no phoning home.
Rate limiting built in. The AdaptiveLimiter enforces random 3 to 5 second delays between actions, with automatic backoff if LinkedIn signals rate limiting. Scans of the same post are scheduled at least 1 hour apart.
Anti detection measures. The browser context uses randomized viewports, standard Chrome UA strings, and stealth scripts to avoid triggering automation detection. However, like any browser automation tool, LinkedIn could theoretically update their detection. This is an inherent risk of any Playwright/Puppeteer based tool, mitigated but not eliminated.
Not officially endorsed by LinkedIn. This is an independent tool. Use it responsibly and at your own discretion.

MCP Servers