nightshift-hook

Research Hook: Nightshift Integration

Command Context

When to Reference Research Module

Always reference when:

• Nightshift overnight pipeline is executing
• Research queue has pending TODO items in research_learning.org
• Need to process links and generate podcasts before morning briefing

Key decisions this hook informs:

• How many items to process this cycle (respects max_sources_per_night)
• Whether to generate podcasts (checks nlm availability)
• Which focus areas have the most pending items

Quick Reference

Question	Answer
When does this run?	Nightshift research timer (02:00 UTC)
What does it process?	TODO items from research_learning.org
Does it generate podcasts?	Yes, if nlm CLI is available and >= 3 sources processed
What's the processing limit?	max_sources_per_night setting (default 20)
What outputs are created?	Literature notes, zettels, podcasts, journal entry, action items

Agents This Hook Invokes

Agent	Purpose
research-orchestrator	Full pipeline: discovery, extraction, synthesis, podcasts, post-processing

Integration Points

• research_learning.org - Source of TODO items to process
• research-orchestrator agent - Invoked in nightshift mode (non-interactive)
• podcast-creator agent - Spawned by orchestrator for podcast generation
• 0-personal/content/podcasts/ - Podcast output directory
• Journal - Session summary written to daily journal

---

This hook is invoked by nightshift for overnight research processing.

Trigger

Called by nightshift scheduler via nightshift run --command=/research-daily.

Mode

Non-interactive. No user confirmation needed — process all eligible TODO items.

Workflow

Step 1: Count and Prioritize Research Queue

Read research_learning.org and count pending items:

import sys
from pathlib import Path

sys.path.insert(0, '.datacore/modules/research/lib')

data_root = Path.home() / "Data"
research_org = data_root / "0-personal" / "org" / "research_learning.org"

# Count TODO items
todo_count = 0
if research_org.exists():
    content = research_org.read_text()
    todo_count = content.count("** TODO") + content.count("*** TODO")