pooriaarab Inbox Triage
LLM-assisted processing of brain/inbox/ using Claude Haiku. Suggests category, filename, and [[wiki-links]] for each note. User approves, edits, or skips before anything moves.
Run
cd ~/Documents/pooriaarab/code/personal-os
# Dry run — see suggestions without moving anything
.venv/bin/python scripts/inbox-triage.py
# Execute — move approved notes
.venv/bin/python scripts/inbox-triage.py --execute
Categories
| Category | Use for |
|---|---|
ideas/ |
Atomic idea notes |
learnings/ |
Things learned from content or experience |
meetings/ |
Meeting or conversation notes |
journal/ |
Personal reflections |
content/ |
Saved articles, videos, tweets |
people/ |
Notes about a specific person |
projects/ |
Project context notes |
areas/ |
Ongoing area notes (health, finance, career...) |
goals/ |
Goals and aspirations |
archive/ |
Low-value, archive |
Review Flow
For each note the script:
- Calls Claude Haiku → gets category, filename, links, confidence, reason
- Shows a rich table of all suggestions (dry run)
- On
--execute: prompts per note —y(accept),n(edit),skip - On accept: injects
## Connectionssection with [[wiki-links]], moves file
Non-interactive (cron): auto-accepts all suggestions.
After Triage
- Update relevant MOC files (
MOCs/MOC-Ideas.md, etc.) to link to new notes - Run
brain-backup.shto push changes to GitHub
Dependencies
ANTHROPIC_API_KEYin~/Documents/pooriaarab/code/personal-os/.env- Python venv:
cd personal-os && python3 -m venv .venv && .venv/bin/pip install anthropic rich click python-dotenv