Detect LinkedIn spam and prospection messages via Beeper MCP. Auto-generates suggested responses (FR/EN) with human-in-the-loop confirmation.

git clone https://github.com/manthis/openclaw-skill-linkedin-spam-filter.git
cd openclaw-skill-linkedin-spam-filter

# Test detection locally
python3 scripts/linkedin-spam-filter.py --test-text "Hi, I have an exciting opportunity for you"

# Full check (requires Beeper MCP)
export BEEPER_SERVER="beeper"
python3 scripts/linkedin-spam-filter.py --dry-run --json

⚠️ Security: Never store MCP tokens or credentials in config files.

1. Detect spam via linkedin-spam-filter.py --json
2. Review the suggested response
3. Send the response via scripts/send-response.py --chat-id <ID> --message "<text>"
4. Archive automatically (done by send-response.py)

⚠️ Important: Always archive the chat after sending a response to keep LinkedIn inbox clean.

• 🔍 Pattern-based prospection detection (FR + EN)
• 💬 Auto-generated response suggestions
• 🌍 Language auto-detection (French/English)
• 🛡️ Human-in-the-loop — never auto-sends
• 📊 JSON output for automation
• 🧪 Standalone test mode (--test-text)
• 📥 Auto-archive after response

• Python 3.8+
• mcporter CLI with Beeper MCP (for live checks)

MIT