DIS-069: wtdir/resolve lack nearest-hive awareness — `//CELL` doesn't round-trip from a nested hive; `//JS-112` ALIASES flat `~/src/JS-112` vs the journal cell; `///` falls back to the launch tree

Now: OPEN
Sev: MED

The naming side got the hive fix; the resolving side didn't. BRO-024 iter 5 taught navCwd to name a hive cell by its CELL name (work/BRO-024's core/discover.js:361-368, unmerged); trunk navCwd still slices under srcRoot() (discover.js:355-373 — the live ///work/JS-112 invite bug). But wtdir/resolve in BOTH trees still join srcRoot()+host flat (discover.js:325-349, 313-315): //CELL typed from a nested hive doesn't round-trip (falls back to cwd scope, lossy after cross-tree nav), and with SRC_ROOT=~~/src //JS-112 is AMBIGUOUS between flat `~~/src/JS-112 and the journal cell ~~/src/journal/work/JS-112. Adjacent: rootTop (discover.js:279-287, BE-045) names the hive meta only when a wt anchors exactly at dirname(srcRoot()); with no wt at~~ /src it falls back to launchTop() — so /// from a ticket worktree targets the LAUNCH tree, not the main tree ([trees] example 5's ///jab/dog breaks). This is [BRO-024]:54's "own ticket due" — filed now; the ///`→main-tree mapping + pwd-outside-SRC_ROOT rejection stay BRO-024:52.

Input

Context

Trunk is pre-ONE-VAR: the whole BRO-024 context machinery ships only in work/BRO-024 (5ebf0c8d+) — merging is the user's call and dissolves the navCwd half; the wtdir/resolve half survives the merge.
Discovery boundaries disagree by design today: naming is SRC_ROOT-bounded (wtdir/navCwd), the find-climb is $HOME-bounded (discover.js:121-143, $HOME/.be probed inclusively at :136-137), topWt stops at the */work/ hive boundary (discover.js:173-174, BE-031). One nearest-hive rule should serve name AND resolve, or aliasing returns.
CLI side effect of the same gap: a pwd outside SRC_ROOT but under any .be under $HOME silently adopts that repo (loop.js:398-406) — the defined no-context error is BRO-024:52's TODO, coordinate.

Goals

wtdir/resolve resolve //NAME via the nearest hive first (cell <tree>/work/NAME), then the flat <srcRoot>/NAME; a name matching BOTH is a defined ambiguity error, not a silent pick.
//CELL round-trips: navCwd(wtdir(//CELL)) == //CELL from any cwd, nested hives included.
One shared nearest-hive helper for navCwd + wtdir (no logic fork).
Tests: round-trip legs (nested fixture kin of work/BRO-024's leg k), the alias-ambiguity error, ///-from-a-cell targeting the main tree (with BRO-024:52's mapping).

Constraints

Coordinate with the work/BRO-024 merge; do not duplicate iter-5's navCwd fix on trunk by hand.
Resolution stays verb-side per the law (verbs resolve args, pager arg-blind); no pager special-casing.

WIP

Design decisions

(none yet)

TODOs

shared nearest-hive helper; wtdir/resolve adopt it
ambiguity = defined error (name in both flat and cell positions)
round-trip test legs; nested hive fixture
/// mapping test once BRO-024:52 lands

Blockers and bummers

Soft-blocked by the work/BRO-024 merge decision (user's call).

Outcome

FILED 2026-07-12 (trees.mkd audit; discharges BRO-024:54 "own ticket due"); nothing landed yet.

DIS-069: wtdir/resolve lack nearest-hive awareness — //CELL doesn't round-trip from a nested hive; //JS-112 ALIASES flat ~/src/JS-112 vs the journal cell; /// falls back to the launch tree