The job of capture
Capture is the short-term buffer for your /Brain.
Its purpose is not elegance. Its purpose is to stop good thoughts, decisions, and observations from disappearing.
The highest-leverage capture system is usually the one with the least friction.
What to capture
Capture anything with a realistic chance of mattering later:
- decisions and why they were made
- new learnings
- hypotheses and ideas
- phrases that describe how you want to work
- recurring blockers or open loops
If you hesitate because you are unsure where something belongs, capture it anyway.
Good capture patterns
- one raw inbox for fast input
- a voice or mobile path for ideas away from the keyboard
- lightweight prefixes or tags only if they genuinely help
- a regular review habit so the inbox does not rot
Example capture file:
# chose PostgreSQL over SQLite
Needed reliable concurrent writes, better observability, and less migration risk later.
Revisit only if deployment overhead becomes the bottleneck.
What capture should not do
Capture is not the moment for deep organization. If you force every note into a perfect taxonomy up front, capture volume drops and the system gets worse.
Do just enough structure to find the item again. Save judgment for consolidation.
Common mistakes
- making capture depend on choosing the perfect folder
- capturing too little because the standard is too high
- leaving the inbox unreviewed for weeks
- storing raw capture in startup files
Next steps
- Read Consolidation to turn raw inputs into durable memory
- Read Maintenance for review cadence and cleanup rules
- Read Obsidian as the Foundation to see how agents read and write to your vault