Antigravity LinkedIn Automation

Appearance

Standard Operating Procedure (SOP) User Guide ​

Quick Summary

This SOP provides clear, actionable instructions for operating the LinkedIn Automation System in preview, live, and fully automated modes.

1. Operating Instructions ​

Procedure 1: Generating a Single Post Preview ​

  1. Open terminal and navigate to project root.
  2. Run preview command with your target topic:
    bash
    python main.py --action post --topic "Future of Agentic AI" --preview
  3. Inspect generated text and character counts output in terminal.

Procedure 2: Publishing a Live Post to LinkedIn ​

  1. Verify LINKEDIN_ACCESS_TOKEN is active in .env.
  2. Execute live posting action:
    bash
    python main.py --action post --topic "Future of Agentic AI" --live
  3. Check returned URL in stdout and verify post appearance on LinkedIn.

Procedure 3: Running End-to-End Orchestration Workflow ​

  1. Execute full orchestration with deep research mode:
    bash
    python main.py --action orchestrate --topic "AI 2026" --research-mode deep --live
  2. The orchestrator will automatically:
    • Query NotebookLM for insights
    • Run deep web research
    • Draft and validate post against soul.py
    • Publish to LinkedIn
    • Stream telemetry to OBS HUD

2. Troubleshooting & FAQ ​

Error: NLM Auth Invalid

Cause: Stale browser cookies or expired session tokens in NotebookLM MCP.
Resolution: Run nlm login in terminal to refresh cookies, then execute python main.py --action health.

Error: JudgeGuard WORK_LOG.md is stale

Cause: No active 🟡 Starting entry logged in WORK_LOG.md within the past 5 minutes.
Resolution: Append echo "🟡 Starting [YOUR_ACTION]" >> WORK_LOG.md before running your command.

Error: LinkedIn Access Token Invalid

Cause: Expired OAuth2 bearer token.
Resolution: Run python setup_linkedin_auth.py to refresh OAuth tokens.

3. Related Documentation ​