Name the modes for their form: terminal ⇄ reader

"Conversation" described the terminal just as well — both modes show the same
session; what differs is the form. Reader mode borrows the browser's idea: the
same content, laid out to be read. The toggle reads `terminal | reader`, the
type is `PaneMode = 'terminal' | 'reader'`, and the overlay is `.pane-reader`.

app-shots.mjs now resets the fleet before it runs — half its checks compare
pane counts, and previous runs left groups and spare shells behind.

docs/conversation-view.md

@@ -60,7 +60,7 @@ Four depths of one component:
 | **D0** brief | Overview tile | prompt (1 line) + state |
 | **D1** card | Overview card, collapsed | prompt, one meta line, answer, reply box |
 | **D2** open | Overview card, unfolded | + the steps between prompt and answer; `↑ show previous turn` walks back one exchange at a time |
-| **D3** full | Session pane, conversation mode | every exchange, windowed, each expandable to D2 |
+| **D3** full | Session pane, reader mode | every exchange, windowed, each expandable to D2 |
 
 The viewer stops being a different thing and becomes **a vertical stack of the card**. That is
 the whole design; the rest is consequences.
@@ -150,7 +150,7 @@ Rules that keep this from becoming the ugly viewer in a small box:
 - **Only the card you opened reads the trace.** Inline in the Overview list a card is a
   summary — one prompt, one clamped answer — and the digest already has that. Reading the trace
   per visible agent, and polling it every three seconds for the working ones, would turn the
-  Overview from one `/api/meta` poll into one transcript read per row. The window (and conversation mode)
+  Overview from one `/api/meta` poll into one transcript read per row. The window (and reader mode)
   is where the middle gets fetched.
 - **A card says nothing the surface around it already says.** No turn number (there is one turn),
   no clock (its header carries `·6m`), no model, no harness. In the viewer the same line's right
@@ -163,7 +163,7 @@ exchange per click**, never by "load everything":
 - A centred `↑ show previous turn` at the top of the body prepends the previous exchange,
   collapsed to prompt + answer (its steps fold behind the same `▸ n steps`). Centred, because it
   belongs to the whole card rather than to the column of text under it.
-- After the second one, `full history ↗` joins it, opening the pane in conversation mode at that
+- After the second one, `full history ↗` joins it, opening the pane in reader mode at that
   exchange. The card is for "what just happened"; archaeology has a bigger room, and the handoff
   is one click.
 - Cap the card at, say, 5 exchanges regardless — past that the card is the wrong tool and says so.
@@ -183,12 +183,12 @@ was a workaround for not having the middle; with the middle visible it has nothi
 `--vvh` is already maintained from `window.visualViewport` (`App.tsx:106-114`); the backdrop
 must use it, or an open keyboard covers the reply line it is there to serve.
 
-### 3.3 The session pane: terminal ⇄ conversation
+### 3.3 The session pane: terminal ⇄ reader
 
 The bottom bar gets a two-state control, next to the zoom:
 
 ```
-                                        [ terminal | conversation ]   − 100% +
+                                        [ terminal | reader ]   − 100% +
 ```
 
 **It is one setting for the whole app, like zoom** — not a per-pane toggle. Reading a fleet means
@@ -199,13 +199,15 @@ pane with nothing to render (a shell) simply stays a terminal.
   the files/trace panels are not sessions at all — same rule the Overview uses to decide what is
   an agent (`cli !== 'shell' && !isPassive(cli)`).
 - **terminal** — today's terminal, untouched.
-- **conversation** — the same session, read: the exchange renderer over `/api/trace/:id`, at D3.
+- **reader** — the same session, laid out: the exchange renderer over `/api/trace/:id`, at D3.
+  The two modes show the *same content*; what differs is the form, which is why the labels name
+  the form. "Conversation" would have described the terminal just as well.
 - The mode is a **view preference**, kept in `localStorage` for the app, not in the store.
-- The terminal element stays mounted and connected underneath; the conversation draws over it. Toggling
+- The terminal element stays mounted and connected underneath; the reader draws over it. Toggling
   must not detach tmux — a reattach costs a repaint and can trip the handoff path
   (`HANDOFF_CODE`, `TerminalPane.tsx`). **Verify** this before shipping: xterm needs layout to
   fit, so "cover, don't unmount" is the low-risk option, and a refit on return is required.
-- The conversation's toolbar is a second header row, not a squeeze into the first — on a phone the
+- The reader's toolbar is a second header row, not a squeeze into the first — on a phone the
   first row has no spare width. It carries only what is true of the whole session and said
   nowhere else: the model, `13 turns`, the token totals abbreviated (`2.2M↓ 654k↑`), `▲▼`, and
   the search box. The harness is the logo in the row above; the raw message count and the cached
@@ -219,7 +221,7 @@ pane with nothing to render (a shell) simply stays a terminal.
   the nearest positioned ancestor and lands a few rows off.
 - Vocabulary: a **turn** is one exchange, a **message** is a raw transcript row. Each turn's meta
   line says `turn 6/13`, so the bar counts the same things the reader does.
-- **The terminal must not keep the keyboard** while the conversation covers it. A mounted xterm with focus
+- **The terminal must not keep the keyboard** while reader mode covers it. A mounted xterm with focus
   swallows keystrokes into the agent's TTY, invisibly — and several paths grab focus back (the
   pane becoming active, the header, the key bar), some of them *after* the mode changes, so the
   guard belongs at each call rather than at the switch.
@@ -233,10 +235,10 @@ pane with nothing to render (a shell) simply stays a terminal.
   tool call is unusable while a task runs; one that never moves makes you chase it.
 - **The prompt band sticks to the top** while you read a long turn. What you want overhead deep
   in someone's 67-step answer is the question it is answering — not a row of numbers.
-- **The conversation fills the pane.** A fixed reading column left a gutter of nothing on each
+- **The reader fills the pane.** A fixed reading column left a gutter of nothing on each
   side while the prompt band still spanned the full width, so the two disagreed about where the
   conversation began. The pane is the measure: narrow the pane and the conversation narrows.
-- **The conversation can be replied to.** Reading a conversation and answering it are the same act — the
+- **Reader mode can be replied to.** Reading a conversation and answering it are the same act — the
   card has always known that, and a rendered session that could only be read would send you back
   to the terminal to type. It is the card's own composer (`.ov-live`), the same `sendInput`, and the
   same optimistic echo: your prompt appears at the bottom with a `working` line until the
@@ -250,11 +252,11 @@ pane with nothing to render (a shell) simply stays a terminal.
 
 | Today | Tomorrow |
 |---|---|
-| `Read this session's trace` on every agent row (`Sidebar.tsx:245`) | bottom bar → **conversation** |
+| `Read this session's trace` on every agent row (`Sidebar.tsx:245`) | bottom bar → **reader** |
 | `Share this session` on every agent row (`Sidebar.tsx:246`) | pane header → **share** |
 | A `trace` **session row per read** (`App.tsx:openTrace`) | **gone** — no duplicate rows |
 | Trace row's `Share` (`Sidebar.tsx:237`) | trace pane header (imported traces keep a pane) |
-| Trace row's `Handover` (`Sidebar.tsx:238`) | conversation / trace-pane footer |
+| Trace row's `Handover` (`Sidebar.tsx:238`) | reader / trace-pane footer |
 | Quick-add **Trace** = open a shared dataset (`Sidebar.tsx:482`) | **stays** |
 
 The last row is deliberate: an **imported** trace has no session behind it, so it is a genuine
@@ -376,7 +378,7 @@ the block model already supports.
 
 1. **How far back in the card?** Proposal: one exchange per click, hard cap 5, then hand off to
    the conversation. Alternative: one, then straight to the full conversation.
-2. **Does the app open in conversation mode**, or always start on the terminal?
+2. **Does the app open in reader mode**, or always start on the terminal?
 3. **Kill the `↑ ↓` turn stepper?** Proposal: yes — unfolding replaces it.
 4. **Unfolded card of a running agent**: live-refresh every 3 s, or freeze until it finishes
    (cheaper, less alive)?
@@ -421,7 +423,7 @@ Built (this branch):
    there is no transcript to read.
 3. `pageOf()` takes a negative offset — "the last N turns" without a round trip to learn `total`
    (`server/test/trace-tail.test.mjs`).
-4. The session pane gets terminal ⇄ conversation, from the bottom bar, app-wide. It draws over the terminal, which stays mounted and
+4. The session pane gets terminal ⇄ reader, from the bottom bar, app-wide. It draws over the terminal, which stays mounted and
    connected but loses the keyboard; the mode is a per-session view preference in `localStorage`
    (`web/src/lib/paneMode.ts`), and its event reaches a pane that is already open — which is what
    the card's "full history ↗" needs.
@@ -433,7 +435,7 @@ Not yet, in the order I would do it:
 6. `head.prompts[]` (§5) — index, first line and timestamp per prompt, so a surface can draw the
    skeleton and label "show previous turn" before fetching the page that holds it.
 7. The sidebar loses its trace buttons and `openTrace`; share moves into the pane header (§3.4).
-8. Windowing by exchange in conversation mode: a collapsed turn is 2–3 rows, so the DOM stays small,
+8. Windowing by exchange in reader mode: a collapsed turn is 2–3 rows, so the DOM stays small,
    `head.prompts[]` gives the skeleton up front, and only an opened turn needs its page. The
    measured-height machinery in `TraceView` is reused as-is — what changes is what a "row" means.
    Until then it reads the last 400 turns in one request.

web/src/App.tsx

@@ -79,11 +79,11 @@ export default function App() {
   // stored — flipping the setting instantly (un)archives.
   const [showArchived, setShowArchived] = useState(false);
   const [archiveAfter, setArchiveAfter] = useState<'week' | 'month' | 'never'>('month');
-  // How every pane is read — the terminal itself, or the conversation in it.
-  // App-wide, like zoom, and remembered the same way.
+  // How every pane is read — the terminal itself, or reader mode over the same
+  // session. App-wide, like zoom, and remembered the same way.
   const [paneMode, setPaneMode] = useState(readPaneMode);
   useEffect(() => onPaneMode(setPaneMode), []);
-  const showPaneMode = (m: 'terminal' | 'conversation') => { setPaneMode(m); writePaneMode(m); };
+  const showPaneMode = (m: 'terminal' | 'reader') => { setPaneMode(m); writePaneMode(m); };
   const [zoom, setZoom] = useState<number>(() => {
     const z = parseInt(localStorage.getItem('am-zoom') || '100', 10);
     return Number.isFinite(z) ? z : 100;
@@ -986,14 +986,14 @@ export default function App() {
               </span>
             )}
             <span className="spacer" />
-            {/* Reading mode sits with zoom because it is the same kind of
-                setting: how you are looking at everything, not what a
-                particular pane is. */}
+            {/* Reader mode sits with zoom because it is the same kind of
+                setting: how you are looking at everything, not what any one
+                pane is. The content is identical either way — this is form. */}
             <span className="seg modebar">
               <button className={paneMode === 'terminal' ? 'on' : ''} title="The terminal itself"
                 onClick={() => showPaneMode('terminal')}>terminal</button>
-              <button className={paneMode === 'conversation' ? 'on' : ''} title="The conversation in it, rendered"
-                onClick={() => showPaneMode('conversation')}>conversation</button>
+              <button className={paneMode === 'reader' ? 'on' : ''} title="Reader mode — the same session, laid out"
+                onClick={() => showPaneMode('reader')}>reader</button>
             </span>
             <button className="zbtn" title="Zoom out" onClick={() => setZoom((z) => Math.max(50, z - 10))}>−</button>
             <button className="zlvl" title="Reset to 100%" onClick={() => setZoom(100)}>{zoom}%</button>

web/src/components/Overview.tsx

@@ -72,9 +72,9 @@ function useConversationTail(id: string, on: boolean, live: boolean) {
   return { turns, missing };
 }
 
-/** Opening this session's pane, reading the conversation rather than the TTY. */
+/** Opening this session's pane in reader mode rather than on the TTY. */
 const openRendered = (id: string, onOpen: (sid: string) => void) => {
-  writePaneMode('conversation');   // app-wide, and it reaches open panes too
+  writePaneMode('reader');   // app-wide, and it reaches open panes too
   onOpen(id);
 };
 

web/src/components/TerminalPane.tsx

@@ -202,7 +202,7 @@ export default function TerminalPane({
   // it — becoming active, the header, the key bar — and some fire after the mode
   // changes, so the guard lives with the call rather than with the switch.
   const modeRef = useRef<PaneMode>('terminal');
-  const focusTerm = () => { if (modeRef.current !== 'conversation') termRef.current?.focus(); };
+  const focusTerm = () => { if (modeRef.current !== 'reader') termRef.current?.focus(); };
   const frameRef = useRef<HTMLDivElement>(null);
   const termRef = useRef<Terminal | null>(null);
   const resyncRef = useRef<() => void>(() => {});
@@ -213,11 +213,11 @@ export default function TerminalPane({
   const previousZoomRef = useRef(zoom);
   const [preview] = useState<TerminalPreview | null>(() => loadTerminalPreview(session.id));
   // The mode is app-wide (the bottom bar owns it, like zoom), but only an agent
-  // has a conversation to show: a shell is a shell, and files/trace panels are
+  // has a conversation to read: a shell is a shell, and files/trace panels are
   // not this component's business at all.
   const canRender = session.cli !== 'shell' && !isPassive(session.cli);
-  const reading = mode === 'conversation' && canRender;
-  modeRef.current = reading ? 'conversation' : 'terminal';
+  const reading = mode === 'reader' && canRender;
+  modeRef.current = reading ? 'reader' : 'terminal';
   // Send a raw byte string to the PTY (for the mobile key-bar: arrows, Esc…).
   const sendKeyRef = useRef<(d: string) => void>(() => {});
   const [conn, setConn] = useState<ConnState>('connecting');
@@ -805,7 +805,7 @@ export default function TerminalPane({
     return () => clearTimeout(t);
   }, [active]);
 
-  // Under RENDER the terminal is covered but still mounted — and a mounted xterm
+  // In reader mode the terminal is covered but still mounted — and a mounted xterm
   // with focus swallows every keystroke into the agent's TTY, invisibly. Hand
   // focus back when the terminal is on top again.
   useEffect(() => {
@@ -862,11 +862,11 @@ export default function TerminalPane({
       </div>
       <div className="term-host" ref={frameRef}>
         <div className="term-fill" ref={hostRef} />
-        {/* The conversation draws OVER the terminal rather than replacing it: xterm needs
+        {/* Reader mode draws OVER the terminal rather than replacing it: xterm needs
             layout to fit, and detaching tmux costs a repaint and can trip the
             handoff path. The terminal stays mounted and connected underneath. */}
         {reading && (
-          <div className="pane-render" onMouseDown={(e) => e.stopPropagation()}>
+          <div className="pane-reader" onMouseDown={(e) => e.stopPropagation()}>
             <ConversationView session={session} paused={visible === false} isMobile={isMobile} />
           </div>
         )}

web/src/conversation.css

@@ -187,10 +187,10 @@ mark.cx-hit.on { background: var(--accent); color: var(--panel); }
   .cs-detail { font-size: 12px; }
 }
 
-/* ---------- conversation mode inside a session pane ---------- */
+/* ---------- reader mode inside a session pane ---------- */
 /* Drawn over the live terminal, which stays mounted underneath. */
-.pane-render { position: absolute; inset: 0; z-index: 3; display: flex; background: var(--panel); }
-.pane-render > .cxv { flex: 1; min-width: 0; }
+.pane-reader { position: absolute; inset: 0; z-index: 3; display: flex; background: var(--panel); }
+.pane-reader > .cxv { flex: 1; min-width: 0; }
 .cxv-empty { margin: auto; padding: 24px; font-size: 11.5px; color: var(--muted); text-align: center; }
 .ph-modes { flex: none; }
 .ph-modes button { padding: 1px 7px; font-size: 9.5px; letter-spacing: 0.06em; }

web/src/lib/paneMode.ts

@@ -1,17 +1,17 @@
-// How every session pane is being read: the terminal, or the conversation.
+// How every session pane is being read: the terminal itself, or reader mode.
 // (docs/conversation-view.md §3.3)
 //
 // One setting for the whole app, like zoom — not per session. Reading a fleet
 // means reading it the same way; flipping panes one at a time was a preference
 // nobody wanted to manage. Kept in localStorage so a reload does not undo it,
 // and announced so an already-mounted pane hears about it.
-export type PaneMode = 'terminal' | 'conversation';
+export type PaneMode = 'terminal' | 'reader';
 
 const KEY = 'am-pane-mode';
 const EVENT = 'am:pane-mode';
 
 export function readPaneMode(): PaneMode {
-  try { return localStorage.getItem(KEY) === 'conversation' ? 'conversation' : 'terminal'; } catch { return 'terminal'; }
+  try { return localStorage.getItem(KEY) === 'reader' ? 'reader' : 'terminal'; } catch { return 'terminal'; }
 }
 
 export function writePaneMode(mode: PaneMode): void {