linkedin-connection-agent

LinkedIn Connection Agent

Fetches LinkedIn connection request results from a PhantomBuster agent, calculates acceptance metrics, and stores analytics in the database. Runs as a daily cron job via Modal.

What It Does

Calls PhantomBuster API to get latest LinkedIn connection results
Downloads structured JSON profile data from S3
Calculates metrics: total sent, accepted, pending, acceptance rate
Stores aggregate stats and individual profiles in linkedin_agent_data table
Tracks status changes (e.g., "Invitation sent" → "Request accepted")

How to Run

Local

# View analytics only
python3 skills/linkedin-connection-agent/scripts/linkedin_connection_agent.py --analytics

Store results in database

python3 skills/linkedin-connection-agent/scripts/linkedin_connection_agent.py --analytics --store


### Modal Cron (Daily at 6 PM EST / 11 PM UTC)

```bash
modal deploy skills/linkedin-connection-agent/scripts/modal_linkedin_connections.py
modal run skills/linkedin-connection-agent/scripts/modal_linkedin_connections.py

Configuration

Environment Variables

PHANTOMBUSTER_API_KEY=xxx
PHANTOMBUSTER_AGENT_ID=685897070795556  # LinkedIn connections agent
DATABASE_URL=postgresql://...

Dependencies

pip install requests psycopg2-binary python-dotenv

Metrics Parsing (Important)

total_sent = profiles with invitationDate (we sent them an invite)
total_accepted = profiles with status = "Request accepted"
pending = profiles with status = "Invitation sent"
acceptance_rate = total_accepted / total_sent × 100

Do NOT count "Already connected" as accepted — those are pre-existing connections.

Important Notes

PhantomBuster daily limit: 20 connection requests
Agent ID 685897070795556 is required — it's the specific LinkedIn connections agent
Individual profiles tracked by unique linkedin_url for status updates
Aggregate stats stored daily for trend tracking

Scripts

File	Purpose
`linkedin_connection_agent.py`	Local analytics script
`modal_linkedin_connections.py`	Modal cron deployment