foundation
The base of your own system. Remove it and the rest stops working.
the marketplace
start setup to open modules.
upkeep
When something changes, keeps every dependent part current.
axioms
Why this system exists and what it protects.
editor
Draws out and sharpens your own thinking.
mercury
Brings useful ideas in and keeps important ones alive.
methodology
How your model captures, develops, and preserves your thinking.
publisher
Turns developed thinking into finished work.
capture
Turns what you save, and the sources you point at, into your record, one capture at a time.
audit
Checks the whole system against what should be true now.
airlock
Lets you try a new model on files you choose, behind a separate GitHub account, and treats what it sends back as untrusted.
bookshelf
Benjamin's shelf of books and ideas worth returning to, for anyone who wants to borrow it.
close
Ends the task you're in without losing anything. Saves what the session taught about you into your own files, finishes or parks every open thread, and asks only what genuinely needs you.
devils-advocate
Every idea, interpretation or recommendation the model gives you comes with the strongest case against it, so you pick instead of riding along. Invoke it directly for a full red-team.
optimise
Re-examines what was just made against your own principles and keeps fixing until the honest answer to "would you change anything?" is no.
model-resync
When a clearly better model arrives, runs it back over your own files once to pull out what the old one missed and drop instructions it no longer needs.
host-memory
Keeps every ai tool's own memory on and, for each tool you say yes to, draws what it learned about you into your own files at each Alexandria session, so what one tool remembers every model knows.