fix: make workspace panels self contained

docs/ARCHITECTURE.md

@@ -80,14 +80,14 @@ durability step. See decisions 0004 and 0005.
 
 ## Default workspace composition
 
-The default desktop workspace has three horizontal bands: the top 20% holds pane,
-agent, tool, context, and goal state; the middle 60% holds responsive vertical
-work lanes; and the bottom 20% is the answer, choice, chat, and approval shelf.
-The work region shows one lane in compact windows, three at normal desktop
-widths, and five on a 3440 px ultrawide. The selected lane alone owns keyboard
-input. Six surfaces continue updating so performance comparisons remain
-meaningful, with a sliding visible window keeping the selected pane onscreen.
-See decisions 0008 and 0009.
+The default desktop workspace is a responsive horizontal row of self-contained
+vertical panels: one panel in compact windows, three at normal desktop widths,
+and five on a 3440 px ultrawide. Inside every panel, the top 20% holds that pane's
+agent, tool, context, goal, target, and status; the middle 60% holds its terminal
+or work surface; and the bottom 20% holds that pane's answer, choices, chat, and
+approval boundary. The selected panel alone owns keyboard input. Six surfaces
+continue updating so performance comparisons remain meaningful, with a sliding
+visible window keeping the selected pane onscreen. See decisions 0008 and 0009.
 
 We should evaluate, not blindly copy, WezTerm, Zellij, RMUX, tmux, and cmux. The
 first spike must compare a reusable terminal crate with a small first-party layer.

docs/PRODUCT_SPEC.md

@@ -66,15 +66,14 @@ surface change never silently launches a process, moves the pane to another
 host, or changes which agent owns the session. Unsupported surfaces are shown as
 unavailable rather than simulated.
 
-The default workspace has three horizontal bands: the top 20% shows context,
-pane/agent state, tools, and goal; the middle 60% is a responsive set of vertical
-work lanes; and the bottom 20% is the decision shelf. The work band shows one
-lane in compact windows, three at normal desktop widths, and up to five on an
-ultrawide display. One lane owns keyboard input at a time, and selecting a lane
-never pauses the others. Explicit split trees and dashboard layouts remain
-supported, but they do not displace this terminal-first default.
+The default workspace is a responsive row of self-contained vertical panels:
+one panel in compact windows, three at normal desktop widths, and up to five on
+an ultrawide display. Every panel has its own 20% context region, 60% terminal or
+work surface, and 20% decision shelf. One panel owns keyboard input at a time,
+and selecting it never pauses the others. Explicit split trees and dashboard
+layouts remain supported, but they do not displace this cmux-style default.
 
-An optional decision shelf sits below the active work surface. It may show an
+An optional decision shelf sits below each pane's work surface. It may show an
 answer draft, two or three concrete choices, why each was suggested, and the
 capability each choice would need. Suggestions are inert data. Choosing one may
 prepare a typed workspace plan, but any command, file mutation, credential use,

docs/TESTING.md

@@ -118,8 +118,8 @@ who already have the final cargo-watch release installed.
   blocked OSC 52 behavior, key encoding, application-cursor mode, and resize.
 - `lumbridge-core` tests capability-gated split/select/rename/close commands,
   idempotent request IDs, close-tree promotion, and atomic agentic setup plans.
-- The GPUI slice tests its key-event adapter, responsive one/three/five-lane
-  geometry, selected-pane lane windowing, 20/60/20 geometry-to-PTY sizing, xterm
+- The GPUI slice tests its key-event adapter, responsive one/three/five-panel
+  geometry, selected-panel windowing, per-panel 20/60/20 PTY sizing, xterm
   256-color conversion, styled-run coalescing, cursor-run boundaries, and
   scrollback shortcut routing. It renders one real actor-owned VT session while
   five surfaces continue their deterministic background workload.

docs/UI_SPIKE_SCORECARD.md

@@ -24,8 +24,8 @@ from macOS and Linux and the hard gates pass.
 The current programs establish dependency, build, launch, and interaction
 baselines. GPUI now routes a real PTY through a VT engine and keyboard encoder,
 paints coalesced styled cell runs and cursor shapes, and derives PTY rows/columns
-from its responsive one/three/five-lane 60% terminal band when the window
-changes. Retained-history page/top/bottom navigation is wired. Text selection,
+from the 60% terminal region inside each responsive one/three/five-panel layout
+when the window changes. Retained-history page/top/bottom navigation is wired. Text selection,
 mouse modes, a native Markdown editor, and one isolated browser child remain.
 
 ## Ubuntu baseline — metal, 2026-08-31
@@ -87,9 +87,10 @@ fallback.
 ## Measurement semantics
 
 Both renderers retain the same all-deterministic six-surface action stream for
-comparison. GPUI now presents a 20/60/20 workspace whose middle band shows one,
-three, or five vertical work lanes while one terminal fixture is replaced with a
-real actor-owned VT session and five deterministic surfaces keep running. Counters
+comparison. GPUI now presents one, three, or five vertical panels, each with its
+own 20/60/20 context/work/decision composition, while one terminal fixture is
+replaced with a real actor-owned VT session and five deterministic surfaces keep
+running. Counters
 separate external PTY batches/lines from total model updates. The GPUI footer
 reports dispatch-to-element-build p50/p95 over a bounded 256-sample window. It
 is deliberately not called key-to-present or frame-present latency: neither

docs/UX_VERTICAL_SLICE.md

@@ -10,14 +10,14 @@ framework decision gate, not a decorative dashboard.
 ## Current-run audit
 
 The first 2026-08-31 GPUI baseline proved six-surface density but gave every
-pane equal visual priority. The current slice replaces that dashboard with three
-horizontal bands: 20% context and pane selection, a 60% responsive work region,
-and a 20% decision shelf. The work region shows one, three, or five vertical
-lanes according to available width. Keyboard focus, command-palette input, live
-PTY input, styled VT cells, cursor shapes, geometry-driven resize, and retained-
-history navigation are wired. Remaining visible gaps are terminal text selection
-and mouse modes, real decision-shelf actions, usage-detail provenance, and end-to-
-end platform accessibility/IME.
+pane equal visual priority. The current slice replaces that dashboard with one,
+three, or five cmux-style vertical panels according to available width. Every
+panel owns its own 20% context region, 60% work surface, and 20% decision shelf.
+Keyboard focus, command-palette input, live PTY input, styled VT cells, cursor
+shapes, geometry-driven resize, and retained-history navigation are wired.
+Remaining visible gaps are terminal text selection and mouse modes, real
+decision-shelf actions, usage-detail provenance, and end-to-end platform
+accessibility/IME.
 
 Screenshots for the audit are stored outside Git under
 `~/shots/2026-08/lumbridge-ui-audit/`. Accessibility and IME correctness cannot
@@ -42,8 +42,8 @@ research, not source assets for Lumbridge. We adapt these interaction patterns:
   cards;
 - a quiet top tab strip for mixed terminal, Markdown, browser, and review
   surfaces;
-- a primary work region with a small responsive number of equal vertical lanes,
-  instead of either one ultrawide terminal or a dense equal-card dashboard;
+- a small responsive number of self-contained vertical panels, instead of
+  either one ultrawide terminal or a dense equal-card dashboard;
 - narrow contextual tools, such as files, review, or Buzz, that can collapse
   when the terminal needs the space;
 - usage and agent state at the edge of the workspace rather than in modal
@@ -60,20 +60,21 @@ surface rather than a hosted Lumbridge control plane.
    palette shortcut.
 2. The sidebar starts with a compact `ATTENTION` group, followed by workspaces
    and remote hosts. It does not become a notification feed.
-3. A compact top 20% context band keeps all pane identities and states visible;
-   the selected pane receives the accent treatment.
-4. A waiting pane uses an amber semantic label in both the context band and
-   attention sidebar. The label always includes words such as `NEEDS INPUT`.
-5. The middle 60% gives one, three, or five terminal/work lanes uninterrupted
-   height and retains target, harness state, dimensions, and surface tabs. The
-   selected lane alone owns keyboard input.
+3. The top 20% of every visible panel keeps that pane's identity, surfaces,
+   execution target, context, goal, and state visible; the selected panel
+   receives the accent treatment.
+4. A waiting pane uses an amber semantic label in both its panel context region
+   and the attention sidebar. The label always includes words such as
+   `NEEDS INPUT`.
+5. The middle 60% of every panel gives its terminal or work surface uninterrupted
+   height. The selected panel alone owns keyboard input.
 6. The footer groups connection state, selected-harness identity, usage-window
    provenance, and burn forecast into readable regions.
 7. The selected pane has a quiet surface strip for Terminal, Browser, Tools,
    Context, Goal, and Review. Switching surfaces keeps one pane identity.
-8. A compact decision shelf may show local-analyst choices below the work
-   surface. It labels its model/provider, trace scope, and whether suggestions
-   are inert; no choice is visually conflated with an approved action.
+8. Every panel has a compact decision shelf below its work surface. It labels
+   whether suggestions are inert or review is required; no choice is visually
+   conflated with an approved action.
 
 ## Keyboard task flow
 

docs/decisions/0008-primary-terminal-3x1-workspace.md

@@ -1,22 +1,23 @@
-# 0008: The default workspace is a primary-terminal 3×1 composition
+# 0008: Every workspace panel uses a terminal-first 3×1 composition
 
 Status: accepted for the native vertical slice.
 
-Lumbridge defaults to three horizontal bands in one column. The top 20% exposes
-pane, agent, tools, context, goal, target, and attention state. The middle 60%
-is the terminal-first work region. The bottom 20% holds answers, choices, chat,
-approvals, and the local-analyst boundary.
+Every Lumbridge workspace panel defaults to three horizontal regions in one
+column. The top 20% exposes that pane's agent, tools, context, goal, target, and
+attention state. The middle 60% is that pane's terminal-first work surface. The
+bottom 20% is that pane's answer, choice, chat, approval, and local-analyst
+boundary.
 
-This composition favors the region where engineers type and read for most of the
-day. It keeps context and decisions visible without making every surface a small
-dashboard card. The middle region may contain a responsive number of vertical
-work lanes under decision 0009; the selected lane alone owns keyboard input.
+This composition keeps the information needed to understand and steer a process
+attached to the same visual object as its terminal. The workspace may contain a
+responsive number of these panels under decision 0009; the selected panel alone
+owns keyboard input.
 
-The framework spike retains six updating surfaces even when five are offscreen.
-That preserves the earlier streaming and reducer workload for performance
-comparison. Hidden surfaces do not receive keyboard input merely because they
-continue to update.
+The framework spike retains six updating surfaces even when one is outside the
+five-panel ultrawide window. That preserves the earlier streaming and reducer
+workload for performance comparison. Hidden panels do not receive keyboard input
+merely because they continue to update.
 
-Window and lane geometry determine each terminal's rows and columns. The UI
+Window and panel geometry determine each terminal's rows and columns. The UI
 resizes the terminal engine and PTY through the bounded runtime actor; it does
 not resize only the text view or infer terminal wrapping after the fact.

docs/decisions/0009-responsive-ultrawide-work-lanes.md

@@ -1,24 +1,26 @@
-# 0009: The primary work region uses responsive vertical lanes
+# 0009: The workspace uses responsive self-contained vertical panels
 
 Status: accepted for the native vertical slice.
 
-The middle 60% work region shows one vertical lane in compact windows, three at
-normal desktop widths, and five when at least 2800 px remains after the sidebar.
-A 3440 px ultrawide therefore presents five lanes. A sliding window over the six
+The workspace shows one vertical panel in compact windows, three at normal
+desktop widths, and five when at least 2800 px remains after the sidebar. A
+3440 px ultrawide therefore presents five panels. A sliding window over the six
 comparison panes always keeps the selected pane visible.
 
-This is not a return to the original 2×3 dashboard. Every lane receives the full
-work-region height, the surrounding context and decision bands remain stable,
-and only the selected lane owns keyboard input. Clicking a lane selects it. Pane
-processes and output continue independently of selection.
+This is not a return to the original 2×3 dashboard and it is not one shared
+20/60/20 frame sliced only through the middle. Every panel receives the full
+workspace height and independently contains its own 20% context region, 60%
+terminal/work surface, and 20% decision shelf. Only the selected panel owns
+keyboard input. Clicking a panel selects it. Pane processes and output continue
+independently of selection.
 
-Terminal geometry is computed from the actual width of one lane, including gaps
-and insets. The selected real PTY is resized through the runtime actor when the
-responsive lane count or window bounds change. On the 3440×1440 reference
-display the current fixed-cell spike derives approximately 71 columns by 40
-rows per lane.
+Terminal geometry is computed from the width of one panel and the height of its
+middle 60% region, including gaps and content insets. The selected real PTY is
+resized through the runtime actor when the responsive panel count or window
+bounds change. On the 3440×1440 reference display the current fixed-cell spike
+derives approximately 71 columns by 42 rows per panel.
 
 The first actor slice still owns one real PTY and five deterministic comparison
 surfaces. Promoting the runtime boundary from one actor to a pane-indexed session
-registry is required before all five visible terminal lanes can own independent
+registry is required before all five visible terminal panels can own independent
 real processes.