# Lend models between your machines

Two Kaptains on your tailnet: one can run the other's models. Compute only; one code grants it, deleting the code ends it.

## What crosses (and what never does)
Compute only. The borrower runs inference on the owner's models — it cannot see chats, files, or settings. Transport is your tailnet (both machines on it): https://chynj.ca/kaptain/docs/how-to/tailscale/

## 1. Issue a connection code
Owner machine: Settings > Network Access > Share models between your computers > Issue a token (`kcat_...`). Copy it; note the owner's tailnet address.

## 2. Connect from the borrowing machine
Settings > Network Access > Share models between your computers > connect to another Kaptain: paste the owner's address + the kcat_ code.

## 3. Verify
The owner's models appear in the borrower's model picker GROUPED UNDER THAT COMPUTER'S NAME — usable in chat, kapsules, delegation. Run one small prompt.

## 4. Revoke
Delete the code on the owner machine; access ends immediately.

## kcat_ vs chynj_
- kcat_ (connection code): another KAPTAIN may run models here, send images for them to describe, and follows their settings — never chats or files.
- chynj_ (device access code): signs a PERSON into this Kaptain — full UI. Never interchangeable.