This receipt captures stabilization slices landed on main after the prior partial-progress reference in PLANNING_REGISTER.md and records issue-facing evidence for camera/input/focus/lasso regressions.

• 001a121 — keyboard zoom commands now requeue when graph metadata is temporarily unavailable, preventing dropped zoom intent behavior.
• 004fd13 — lasso boundary hit-testing updated to center-inclusive semantics with regression coverage.
• 18c6ae9 — node-open focus handoff hardened with timeline tracing to reduce blank-first-frame activation races.
• 874e2a6 — stale focus hints cleared during pane-close handoff.
• 441aded — active graph pane recovery after node-pane close handoff, plus deterministic scenario/unit coverage.
• d8983c9 — focus activation defers unmapped primary target and applies mapped fallback while preserving retry hint.
• 4708e55 — deferred focus activation diagnostics receipts added.
• f12e0cc — regression coverage added for deferred focus activation diagnostics emission.
• 37350e7 — background click deselect now ignores frames where graph primary-click actions were already handled.
• 4f1011f — unmapped semantic child-webview opens are surfaced via warn-level diagnostics instead of silent drop.
• e755f48 — unmapped semantic child-webview opens are deferred/retried across frames via runtime queueing.
• 57a55e6 — semantic child-webview pane opens now route through OpenNodeFrameRouted intent flow (same frame-routing semantics as other node opens).
• 7782fd9 — orchestration scenario coverage added for deferred child-webview retry and routed pane-open on first mapped frame.
• 1a6e8a8 — compositor pass-contract diagnostics coverage added for overlay-without-content pass-order violation and overlay style/mode emission channels.
• 67a4ad9 — scenario-level healthy compositor overlay diagnostics assertion added (style/mode channels present, pass-order violation absent).
• 936073e — background click deselect is now deterministic (plain click only), with modifier/radial-open regression coverage.
• b6b931b — compositor pass-contract invariants hardened: pass-order violations are now render-mode-aware (composited-only), native-overlay chrome diagnostics are regression-tested, and GL framebuffer binding restoration now honors captured state.
• 0b837b4 — compositor render-mode policy regressions expanded to cover embedded/placeholder pass-order non-violation paths and hover-overlay style mapping for composited/native-overlay modes.
• b77fd1e — background pan no longer blocks on radial_open, preventing sticky radial-menu state from disabling graph pan controls; policy regression added.
• 2c1f3e1 — lasso metadata state keying is now explicitly regression-covered for per-view scoping (metadata_id-derived lasso state IDs differ across views).
• 64cd66d — tile rearrange overlay scheduling regressions added: focused composited tiles now have explicit Focus overlay-pass scheduling evidence, and hovered native-overlay tiles have explicit Hover overlay scheduling evidence (fallback policy path).
• 5c3a175 — compositor replay-forensics substrate landed for #166: bounded replay sample ring capture in the compositor adapter, replay sample channels registered in phase diagnostics, and snapshot export now includes compositor_replay_samples evidence.
• 82f3712 — differential composition contract landed for #167: composited content pass now skips unchanged tile signatures (webview + pixel rect), preserves overlay affordance scheduling, emits skip/composed/fallback-reason diagnostics channels, and records per-frame skip-rate basis points.
• 37f2ba8 — diagnostics/profiling hook slice for #184: diagnostics snapshot JSON now exports compositor_differential summary metrics (composed/skipped/fallback counts and skip-rate basis points), and Diagnostics Inspector Compositor tab surfaces the same differential metrics in a dedicated summary grid.
• 187cb2e — optimization/degradation policy slice for #184: compositor now culls off-viewport tile content callbacks, enforces an explicit per-frame composited content budget with placeholder-mode degradation under GPU-pressure conditions, and emits culling/degradation diagnostics channels surfaced via diagnostics summary/export.
• 0a25448 — batching/resource-reuse observability slice for #184: compositor now emits context reuse hit/miss channels and per-frame overlay batch-size samples, with diagnostics summary/export surfacing reuse and batching metrics alongside degradation and skip-rate signals.
• 045c8ca — replay forensics inspector completion slice for #166: Diagnostics Inspector Compositor tab now surfaces replay availability summary (sample count, violation count, latest sequence/node/duration), snapshot export feedback explicitly reports replay sample/violation totals with export path, and snapshot JSON shape includes compositor_replay summary metadata alongside replay artifacts.
• 4b5e324 — selection-contract hardening slice for #185: node select/deselect hit resolution now emits explicit ambiguous-hit diagnostics when event node index cannot be resolved to a Graphshell node key, preserving prior selection truth instead of drifting state; regression tests cover both ambiguous and valid-hit paths.
• 5f61430 — open-routing contract slice for #186: selected-node open execution now consumes queued open mode as a one-shot signal per action, pending open requests are routed through one deterministic node+mode path, split-mode node opens normalize leaf-root wrapping before split construction, and targeted regressions cover one-shot mode consumption, split-route focus handoff, and split root wrapping.
• d9422f0 — close-handoff contract slice for #187: tab-close now deterministically preselects successor focus in the same tab group (right sibling, then left fallback) before tile removal, covering node-pane and tool-pane close flows so the next context remains immediately visible/usable without ambiguous active-target repair.
• d67ffa9 — GUI boundary-contract test slice for #244: added regression guards proving Servo callback semantic events remain queue-only until reducer intent application and proving GUI orchestration/presenter authority preserves semantic intents for reducer-owned mutation, enforcing decomposition invariants from the embedder-debt split plan.

• Targeted workbench/unit tests for focus activation targeting and diagnostics emission are passing.
• Targeted replay-forensics tests are passing (replay_ring_is_bounded_to_capacity, guarded_callback_with_snapshots_returns_before_and_after_states, diagnostics_json_snapshot_shape_is_stable, snapshot_json_includes_compositor_replay_samples_section, diagnostics_registry_declares_phase3_identity_channels_with_versions).
• Targeted differential-composition tests are passing (differential_content_decision_skips_when_signature_is_unchanged, differential_content_decision_recomposes_when_signature_changes, focus_overlay_scheduling_is_preserved_when_content_signature_is_clean, hover_overlay_scheduling_is_preserved_when_content_signature_is_clean, diagnostics_registry_declares_phase3_identity_channels_with_versions).
• Targeted diagnostics snapshot tests for differential summary are passing (diagnostics_json_snapshot_shape_is_stable, snapshot_json_includes_compositor_differential_summary_section, snapshot_json_includes_compositor_replay_samples_section).
• Targeted optimization/degradation regressions are passing (should_cull_tile_content_when_disjoint_from_viewport, should_not_cull_tile_content_when_visible_in_viewport, gpu_pressure_degradation_triggers_at_budget_boundary, diagnostics_registry_declares_phase3_identity_channels_with_versions, snapshot_json_includes_compositor_differential_summary_section).
• Targeted batching/reuse diagnostics tests are passing (diagnostics_registry_declares_phase3_identity_channels_with_versions, snapshot_json_includes_compositor_differential_summary_section, should_cull_tile_content_when_disjoint_from_viewport, gpu_pressure_degradation_triggers_at_budget_boundary).
• Targeted replay inspector/export diagnostics tests are passing (diagnostics_json_snapshot_shape_is_stable, replay_export_feedback_includes_path_and_counts, snapshot_json_includes_compositor_replay_samples_section).
• Targeted selection-contract tests are passing (test_select_node_action_ctrl_click_adds_to_selection, test_select_node_single_click_does_not_affect_multi_selection, test_node_key_or_emit_ambiguous_hit_emits_diagnostic_on_none, test_node_key_or_emit_ambiguous_hit_does_not_emit_for_valid_node).
• Targeted open-routing contract tests are passing (pending_open_mode_is_one_shot_after_execution, pending_open_request_split_mode_uses_split_route_and_focuses_node, open_or_focus_node_pane_split_wraps_leaf_root_before_split).
• Targeted close-handoff contract tests are passing (close_handoff_from_active_node_tab_prefers_right_successor, close_handoff_from_active_tool_tab_prefers_left_when_no_right_successor, close_settings_tool_pane_restores_previous_graph_focus_via_orchestration, ensure_active_tile_recovers_after_active_node_tile_is_removed).
• Targeted GUI boundary-invariant tests are passing (servo_callback_events_are_enqueue_only_until_reducer_applies_intents, orchestration_preserves_semantic_intents_until_reducer_applies_them).
• cargo check passed for each stabilization slice.
• scripts/dev/smoke-matrix.ps1 quick (Windows quick profile) passed after the latest stabilization test coverage commit.
• scripts/dev/smoke-matrix.ps1 quick (Windows quick profile) passed after each newly landed stabilization slice above.

This lane remains partial in readiness terms because the stabilization bug register in PLANNING_REGISTER.md still includes unresolved items outside the landed focus/lasso/camera slices (notably compositor pass-contract closure evidence and remaining active repro verification).

Suggested issue comment body:

Stabilization progress receipt (2026-02-28) has been landed in docs with commit evidence and validation notes: design_docs/graphshell_docs/implementation_strategy/2026-02-28_stabilization_progress_receipt.md.

Newly landed commits on main: 001a121, 004fd13, 18c6ae9, 874e2a6, 441aded, d8983c9, 4708e55, f12e0cc, 37350e7, 4f1011f, e755f48, 57a55e6, 7782fd9, 1a6e8a8, 67a4ad9, 936073e, b6b931b, 0b837b4, b77fd1e, 2c1f3e1, 64cd66d.

Net result: camera/zoom command reliability, lasso boundary correctness, click-away selection determinism, and compositor pass-contract diagnostics/state isolation robustness improved; pane-open/pane-close focus activation race is substantially hardened with regression evidence. Lane remains partial pending closure of remaining bug-register items and final compositor pass-contract closure evidence.