FAQ
What yori is, what it deliberately is not, and why.
Does yori call an LLM?
No. yori is a pure renderer — text in, text out. It produces the prompt and
gets out of the way. You own the model invocation: pipe the output into
claude, llm, or anything else.
cat bug.log | yori run triage | claudeIs there a server or database?
No. The store is just a directory of markdown files. There's no daemon, no
service, no account. Versioning is your own git, and sharing is git push /
git clone.
How is this different from a snippets folder?
A folder gives you files; yori gives you composition and resolution on top of
them: Liquid templating with variables, partials, and inheritance; a layered
store (project → global → package) that resolves names with override semantics;
and yori sync, which renders artifacts into the dirs your coding agent reads
from.
What's the difference between a package install and an item install?
A package install (yori install <git-url>) adds the whole repo as a
read-only layer you resolve through. An item install
(yori install <reg> <item>) copies a single artifact plus its dependency
closure into your store as editable source you own. See
Registry.
Why "yori"?
Named after Tron's Yori, who ran the I/O Tower — the gateway where a User's words reached their program. yori is where the words you tell your AI live.
Does it phone home?
No telemetry, ever. The only network use is git, and only when you run a
registry command.