Patterns
Not new commands – ways to put the ones you already have together. What to reach for, and when.
Which shape
Section titled “Which shape”phosphor init asks once; here’s the reasoning behind each answer.
- homelab (the default): you have, or expect to have, more than one machine. You want the fleet panel, the file browser and the monitoring from day one. Most people start here.
- revived: one machine, brain and only screen at once – often older or
weaker hardware you’re giving a second life as the brain (see concepts:
that’s the whole idea of reusing what you have). Skips CLOUD, the
heaviest tab (three TUIs at once), because there’s no fleet to browse yet
and the machine may not have the headroom to spare. Add CLOUD back later
with
phosphor keeponce you do have somewhere to point yazi. - dev: you already know you’ll spend most of your time with an AI assistant. Puts a two-assistant tab first (Alt-1) instead of last.
None of this is permanent. phosphor recipe adds what a shape left out;
phosphor tabs reorders; Alt-r changes any tab by hand.
Growing it
Section titled “Growing it”Three ways to add a tab, from least to most committed:
phosphor recipe NAME– a bundle Phosphor ships (homelab,dev,bubble,workbench). Fastest, and undoing it isphosphor tabs,f.- A
tabs.d/*.tomlfile of your own – same mechanism recipes use, but the content is yours: something you built once and want on every deck you run, or a tab a friend sent you as a file. Drop it in,phosphor gen. See profile. phosphor keep– you arranged a tab by hand (split it, swapped a program in) and want to keep it. This writes straight into your profile: the most permanent of the three, and the only one Alt-r’s s andphosphor keepwill ever touch again.
Recipes and tabs.d are how you move a tab to a deck. phosphor keep is
how you turn something you built live into one.
Moving to a new brain
Section titled “Moving to a new brain”Nothing about a deck is tied to the machine it runs on except the machine itself. To move the whole thing:
- On the new machine:
sh install.sh(see install), stopping beforephosphor init– you’re not building a new profile, you’re bringing yours. - Copy over, from the old brain:
~/.config/phosphor/(your profile andapps.toml,tabs.d/), and your notebook –~/.local/share/phosphor/notes.mdand-archive.md, unless[notes] folderalready points it at a synced vault, in which case it’s already there. phosphor gen && phosphor upon the new machine. Hosts withlocal = truein your profile describe the old brain: edit that block (or runphosphor setupand let it walk you through the new machine’s own disks) before your firstgenthere.- Point every screen’s
deckcommand at the new brain: rerunphosphor phone/phosphor screenfrom it for each one (see phones).
Nothing is deleted from the old machine by any of this – phosphor down
there when you’re done, by hand, once you’ve confirmed the new one works.
Sharing one tab
Section titled “Sharing one tab”The same idea as moving a whole deck, scaled down: put the [[tabs]] block
in a file, hand the file over. phosphor keep --dry-run prints a tab’s
block without writing it, ready to paste into a tabs.d/*.toml – yours to
keep, or someone else’s to drop in.
Backing out
Section titled “Backing out”Every write phosphor’s tools make to your profile keeps a deck.toml.bak
first (the previous version – only one deep, not a history). phosphor gen --dry-run shows what a change would write without writing it. Nothing
here replaces real version control: a profile is one text file, so git init ~/.config/phosphor (kept out of any repo you publish, see privacy) is
a real history if you want one.