Beagle worktree lifecycle (be repos)

The ticket work loop over beagle-native repos, in a VM with WORK_ROOT in the work/ dir of the project. be=jab is the JavaScript impl (reads js files from be, not node_modules). Agents commit or push only on the user's explicit request (work).

Fork

mkdir $WORK_ROOT/<TICKET>, then cd $WORK_ROOT/<TICKET> (e.g. ticket TKT-123, topic TKT).
Clone options, in the order of preference (asc), assuming we clone be:
1. a full clone: $be get be://localhost/project (Victorian stale)
2. a store-backed worktree: $be get file:/$SRC_ROOT/.be/project (Edwardian)
3. a main tree clone: be get ///be/ (preferred by default)
4. a topic tree clone: be get //<TOPIC>/ (preferred for interdependent ticket families, to minimize main tree disturbance)
For tickets touching several repos, use //<TICKET>-be, //<TICKET>-jab, etc worktrees for different repos.

Work and review

$be or $be status see the state of the tree.
$be diff or $be diff path see the changes, including branch ahead/behind.
$be get update from the tracked branch, $be get \? from the trunk.
Update and rerun the tests before reporting the work done.

Land

$be put <file> staging (when wrapping up the work).
$be post "TKT-123: comment" committing (explicit request only).
$be post push to the upstream (explicit request only).