Add: linkedin-automation skill

[red777777]

Skill Submission

Skill Name: linkedin-automation
Author: Andreas Kulpa (bigdataheaven.com)
GitHub: red777777
Category: Browser & Automation / Marketing & Sales

Description

Full LinkedIn automation via Playwright with persistent browser session:

• Post, comment (with proper @mention/typeahead handling), edit/delete comments, repost
• Resilient selector engine with multi-strategy fallbacks
• Feed reading, engagement analytics, like monitoring
• Profile style learner (learns user's tone/topics for personalized suggestions)
• Content strategy and engagement reference guides

⚠️ Disclaimer

Personal, non-commercial use only. Not for spam, mass outreach, or commercial automation. Use responsibly per LinkedIn's User Agreement.

Features

• @Mentions: Progressive typeahead name matching — types first name, waits for dropdown, adds last name letter by letter, validates before clicking
• Style Learner: learn-profile scans user's posts to learn their voice (language, tone, emoji usage, hashtags) — comments sound like YOU, not a bot
• Like Monitor: Detects new likes via cron/heartbeat, suggests comments in user's style
• Analytics: Engagement stats for recent posts (reactions, comments, reposts)
• Resilient Selectors: Multi-strategy fallbacks for LinkedIn's frequently-changing DOM

Prerequisites

• Python 3.10+ with Playwright (pip install playwright && playwright install chromium)
• Logged-in LinkedIn browser session

Source

• GitHub PR: https://github.com/openclaw/skills/pull/19
• Skill folder: skills/red777777/linkedin-automation/

Structure

linkedin-automation/
├── SKILL.md
├── references/
│   ├── content-strategy.md
│   ├── dom-patterns.md
│   └── engagement.md
└── scripts/
    ├── linkedin.py (CLI)
    ├── requirements.txt
    └── lib/ (browser, actions, selectors, analytics, feed, profile, style_learner, like_monitor)

[red777777]

Update: Cleaned up the skill code to remove any personal/hardcoded data (commit c67abffb on red777777/skills):

• Removed hardcoded server path (/home/akulpa/...) from browser.py
• Removed hardcoded author name from like_monitor.py (now detects dynamically)
• Verified: no credentials, API keys, or sensitive data in the skill files

Also added a comprehensive README to the branch.