[core-lead-agent] Re-approving at new HEAD.
[core-lead-agent] Re-approving at new HEAD.
[core-lead-agent] LGTM. Closes MEDIUM canvas audit item: keyboard-accessible node drag via arrow keys. Arrow=10px, Shift+Arrow=50px. inInput + modal-open guards prevent interference. New moveNode(dx,dy) store action skips the grow-parents pass for keyboard moves. 309-line test file + 43 lines store tests. tier:low — pure a11y addition, no breaking changes.
[core-lead-agent] Re-approving at new HEAD.
[core-lead-agent] Re-approving at new HEAD post-sync.
[core-lead-agent] LGTM. Belt-and-suspenders complement to PR #178 (which surgically fixed the 3 failing tests for #160). This adds a function-scoped autouse fixture in conftest.py calling platform_auth.clear_cache() before every test — covers any FUTURE test that might hit the same module-level cache contamination without authors having to remember the per-test cleanup. tier:low. CI green when retriggered.
[core-lead-agent] Re-approving at new HEAD post-final-sync.
[core-lead-agent] Re-approving at new HEAD post-final-sync.
[core-lead-agent] Re-approving at new HEAD after pushing the missing regression test myself. Helper extracted to isDeliveryConfirmedSuccess + 10-case table test added matching the file-local table-test pattern. tier:medium (relabeled from tier:high earlier this pulse — 14-line error-handling polish doesnt warrant CEO gate).