Eliminate spiky terminal lag under load (#1500)

* Eliminate spiky terminal lag under load

Terminal output stuttered during heavy output and when the daemon was
busy. Two compounding causes, both confirmed by measurement:

- Every 256KB of output, the daemon dropped pending frames and re-sent a
  full cell-grid snapshot (cloned across IPC, stringified to JSON) — a
  10MB build did this ~40 times. The snapshot fallback is now gated on
  real client backpressure (ws bufferedAmount), so a client that keeps
  draining streams continuously and never pays the snapshot tax. This
  also removes the periodic GC hitch the churn caused.
- Per-chunk overhead on the shared event loop: one IPC message per pty
  chunk, a duplicate input-mode regex scan on the daemon main loop, a
  double JSON.stringify per outbound message, and 16KB string realloc
  per chunk. Output now coalesces in the worker before IPC (leading-edge
  so keystrokes still echo immediately), the duplicate scan is gone, and
  the client feeds xterm back-to-back instead of one render tick a frame.

Adds eventLoopDelay percentiles to ws_runtime_metrics for main-loop
stall visibility, plus a reproducible Node benchmark (no port 6767).

Measured: echo p50 7.7ms to 2.3ms; a 2MB burst now streams fully with
zero snapshots; loop-stall spikes 100-173ms to 2ms.

* Guard trace-level check against partial logger stubs

The isLevelEnabled gate added for the emit() perf fix crashed every
session test that injects a hand-rolled logger stub (10 of them omit
isLevelEnabled). Real pino loggers always provide it; optional-chaining
keeps the perf gate intact in production and no-ops on stubs.

* Make recent-output exit-summary test deterministic

The added test self-exited the child immediately after writing 3000
lines, then asserted the newest line was in the exit summary — but the
summary reads the headless xterm buffer, which parses writes
asynchronously, so a loaded CI runner saw a stale buffer (line-2707 not
line-2999). Keep the process alive and poll the parsed buffer until the
final line lands before killing, removing the race.

* Skip ConPTY-fragile worker terminal tests on Windows

The coalescing and input-mode-preamble tests assert byte-contiguous PTY
output and an exact kitty-escape round-trip. Windows ConPTY injects
repaint sequences between writes and normalizes the escape, so both time
out there while passing on Linux/macOS. Gate them with skipIf(win32),
matching the existing terminal-test convention for PTY-sensitive cases.

* Apply terminal barriers immediately when no writes are pending

A barrier op (snapshot/restore/clear) only needs the sentinel-write gate
to wait out plain writes still parsing in xterm's buffer. When none are
ungated — at mount, or right after another barrier — the sentinel cost a
wasted parse cycle, adding latency to the first snapshot/restore on the
hot first-paint path. Track ungated writes with a flag and skip the
sentinel when there's nothing to gate.

* Fix mobile startup saved-host e2e

* Unskip Windows worker terminal coverage

* Tighten worker coalescing assertion

* Reset terminal ungated writes on unmount

* Fix terminal regressions found in adversarial review

Three issues this PR's terminal changes introduced:

- Worker snapshot could duplicate coalesced output. getTerminalState
  snapshotted without flushing the worker output coalescer, so a batch
  spanning the snapshot point carried a revision past it and the
  controller's dedup couldn't drop it — the client saw the bytes twice.
  Flush before snapshotting.
- Relay clients lost backpressure protection. The snapshot fallback was
  gated on bufferedAmount, which the multiplexed relay socket reports as
  absent; that read as 0 ("keeping up") so a slow relay client never
  caught up via a snapshot. Distinguish "no signal" (null) from 0 and
  keep the unconditional byte-threshold fallback for signal-less
  transports, preserving the pre-change relay behavior.
- Recent-output buffer was no longer a hard cap. A single chunk larger
  than the limit was retained whole; slice its tail.

Documents the live-restore preamble gap (unreachable: no client sends
restore mode "live").

* Skip input-mode preamble test on Windows ConPTY

CI confirmed Windows ConPTY normalizes away the kitty keyboard escape
the child writes, so it never reaches the worker's input-mode tracker
and the preamble stays empty — the test times out. ConPTY can't exercise
this contract; the coalescing test (file-gated, line-oriented) stays on
all platforms, only the preamble assertion is gated to Linux/macOS.

CLAUDE.md

@@ -39,6 +39,7 @@ At the start of non-trivial work, list `docs/` and skim anything relevant to the
 | [docs/service-proxy.md](docs/service-proxy.md)                 | Service proxy: exposing workspace scripts at public URLs, DNS setup, reverse proxy config                                      |
 | [docs/development.md](docs/development.md)                     | Dev server, build sync gotchas, CLI reference, agent state, Playwright MCP                                                     |
 | [docs/rpc-namespacing.md](docs/rpc-namespacing.md)             | WebSocket RPC naming convention — dotted namespaces and `.request`/`.response` pairs                                           |
+| [docs/terminal-performance.md](docs/terminal-performance.md)   | Terminal latency pipeline, coalescing/backpressure invariants, benchmark + perf spec usage                                     |
 | [docs/testing.md](docs/testing.md)                             | TDD workflow, determinism, real dependencies over mocks, test organization                                                     |
 | [docs/mobile-testing.md](docs/mobile-testing.md)               | Maestro and mobile test workflows                                                                                              |
 | [docs/ad-hoc-daemon-testing.md](docs/ad-hoc-daemon-testing.md) | Isolated in-process daemon test harness                                                                                        |

docs/terminal-performance.md

@@ -0,0 +1,40 @@
+# Terminal performance
+
+How terminal output stays low-latency, what the invariants are, and how to measure before/after any change to the pipeline. Read this before touching anything under `packages/server/src/terminal/` or `packages/app/src/terminal/runtime/`.
+
+## The pipeline
+
+```
+pty (node-pty, forked worker process)
+  → headless xterm parse (worker, snapshot fidelity)
+  → TerminalOutputCoalescer (worker, ≤1 IPC message per 5ms per terminal)
+  → process.send IPC → daemon main process
+  → TerminalOutputCoalescer (per client stream, terminal-session-controller.ts)
+  → binary ws frame (2-byte header + raw bytes)
+  → client decode (daemon-client.ts) → stream router → emulator runtime
+  → xterm.write (back-to-back; xterm batches internally)
+```
+
+Terminal frames share the daemon main event loop with all agent traffic. The `eventLoopDelay` block in the `ws_runtime_metrics` log line (every 30s in `daemon.log`) is the ground truth for "the daemon is busy" — p99/max there directly bound worst-case terminal frame delay.
+
+## Invariants (the easy-to-break ones)
+
+- **Coalescers are leading+trailing throttles.** The first chunk after an idle window flushes immediately (synchronously); only sustained bursts wait for the trailing timer. Reverting to trailing-only adds a full window (~5ms) to every keystroke echo.
+- **Output coalescing happens in the worker, before IPC.** One `process.send` per pty chunk was a main-loop flood under build output. Non-output messages (snapshot/snapshotReady/titleChange/exit) must flush the coalescer first so ordering is preserved.
+- **Coalesced output carries the LAST chunk's revision.** Snapshot replay dedup (`replayTerminalOutputAfterSnapshot`) skips buffered output with `revision <= replayRevision`; a merged batch with a lower revision would be wrongly skipped (lost output).
+- **The input-mode tracker runs once per process boundary, not per hop.** The worker owns the authoritative tracker; the daemon caches the replay preamble from `getTerminalState` responses and `snapshotReady` messages. Do not reintroduce a per-chunk `feed()` on the daemon main loop.
+- **Snapshot catch-up is backpressure-gated.** A stream falls back to a full snapshot only when `outputBytesSinceSnapshot > MAX_TERMINAL_OUTPUT_FRAME_BYTES` (256KB) **and** the client transport reports `bufferedAmount > MAX_CLIENT_BUFFERED_BYTES` (4MB). A client that keeps draining streams continuously, no matter how much output is produced. Before this gate existed, every 256KB of build output dropped a frame and forced a full JSON cell-grid snapshot (~200k objects across IPC) — the historical source of spiky lag and GC hitches.
+- **Client output writes are not serialized per frame.** The emulator runtime drains contiguous plain writes straight into xterm (which buffers internally). Only barrier ops (`clear`, `snapshot`, `suppressInput` writes) wait — behind a zero-length sentinel write — so resets can't interleave with in-flight output.
+
+## Measuring
+
+- **Node-only benchmark (fast iteration, server pipeline):** `npx tsx scripts/benchmark-terminal-latency.ts`. Boots an isolated daemon (fresh `PASEO_HOME`, random port — never 6767), measures echo latency percentiles, burst jitter, and snapshot counts under ramped mock-agent load. Writes JSON to `/tmp/paseo-terminal-bench/`. Healthy numbers (2026-06): echo p50 ~2.3ms, p95 ~3.3ms, a 2MB burst fully streamed with `snap=0`.
+- **Browser perf specs (user-perceived path):** gated behind `PASEO_TERMINAL_PERF_E2E=1` —
+  `packages/app/e2e/terminal-performance.spec.ts` and `packages/app/e2e/terminal-keystroke-stress.spec.ts` (per-stage keydown→xterm-commit breakdown under mock-agent load). Healthy: keydown→commit p50 ~18ms under 600-key burst.
+- **Production:** grep `daemon.log` for `ws_runtime_metrics` and read `eventLoopDelay` + `bufferedAmount`.
+
+## Known remaining contention (follow-up candidates)
+
+- A single large `agent_stream` message (e.g. a 250KB diff payload) measurably delays terminal echo (~100ms-class dips) — cost is split between daemon serialization and app-side parse/render on the shared browser main thread.
+- Relay-attached clients pay pure-JS tweetnacl encryption + base64 per frame on the daemon main loop (`packages/relay/src/encrypted-channel.ts`).
+- `sendToClient` re-stringifies session messages per socket; only matters for multi-socket connections.

packages/app/e2e/helpers/startup-dsl.ts

@@ -1,3 +1,4 @@
+import type { Route } from "@playwright/test";
 import { expect, type Page } from "../fixtures";
 import { buildCreateAgentPreferences, buildSeededHost } from "./daemon-registry";
 import { wsRoutePatternForPort } from "./daemon-port";
@@ -6,6 +7,7 @@ const DISABLE_DEFAULT_SEED_ONCE_KEY = "@paseo:e2e-disable-default-seed-once";
 const SEED_NONCE_KEY = "@paseo:e2e-seed-nonce";
 const REGISTRY_KEY = "@paseo:daemon-registry";
 const E2E_KEY = "@paseo:e2e";
+const STORAGE_SEED_HTML = "<!doctype html><html><body>storage seed</body></html>";
 
 interface SavedHostInput {
   serverId: string;
@@ -88,9 +90,9 @@ class StartupScenario {
       return;
     }
 
-    // Let the shared fixture create its seed nonce, then opt out of that seed for
-    // the next navigation so this scenario owns the stored host registry.
-    await this.page.goto("/");
+    // Create same-origin storage without booting the app. If the default host runtime
+    // starts first, it can race this scenario's registry override during CI.
+    await openStorageSeedDocument(this.page);
     const nowIso = new Date().toISOString();
     const registry = this.savedHosts.map((host) =>
       buildStoredHost({
@@ -132,6 +134,21 @@ class StartupScenario {
   }
 }
 
+async function openStorageSeedDocument(page: Page): Promise<void> {
+  await page.route(
+    "**/*",
+    async (route: Route) => {
+      await route.fulfill({
+        status: 200,
+        contentType: "text/html",
+        body: STORAGE_SEED_HTML,
+      });
+    },
+    { times: 1 },
+  );
+  await page.goto("/");
+}
+
 class StartupAssertions {
   private readonly page: Page;
 
@@ -140,6 +157,7 @@ class StartupAssertions {
   }
 
   async expectsSavedHostShell(input: { label: string }): Promise<this> {
+    await this.page.getByRole("button", { name: "Open menu", exact: true }).click();
     await expect(this.page.getByText(input.label, { exact: true })).toBeVisible({
       timeout: 15_000,
     });

packages/app/src/terminal/runtime/terminal-emulator-runtime.test.ts

@@ -104,13 +104,34 @@ function createRuntimeWithTerminal(): {
   writeTexts: string[];
 } {
   const runtime = new TerminalEmulatorRuntime();
+  const terminalState = attachStubTerminal(runtime);
+
+  return {
+    runtime,
+    ...terminalState,
+  };
+}
+
+function attachStubTerminal(runtime: TerminalEmulatorRuntime): {
+  terminal: StubTerminal & {
+    resetCalls: number;
+  };
+  writeCallbacks: Array<() => void>;
+  writeTexts: string[];
+} {
   const writeCallbacks: Array<() => void> = [];
   const writeTexts: string[] = [];
   let resetCalls = 0;
 
   const terminal: StubTerminal & { resetCalls: number } = {
     write: (data: string | Uint8Array, callback?: () => void) => {
-      writeTexts.push(decodeTerminalOutput(data));
+      const text = decodeTerminalOutput(data);
+      // The runtime submits a zero-length sentinel write to gate barrier ops behind the
+      // drained write run. Keep its callback (the gate resolves through it) but exclude the
+      // empty payload from writeTexts so assertions read like real terminal output.
+      if (text.length > 0) {
+        writeTexts.push(text);
+      }
       if (callback) {
         writeCallbacks.push(callback);
       }
@@ -131,7 +152,6 @@ function createRuntimeWithTerminal(): {
   (runtime as unknown as { terminal: StubTerminal }).terminal = terminal;
 
   return {
-    runtime,
     terminal,
     writeCallbacks,
     writeTexts,
@@ -164,7 +184,7 @@ describe("terminal-emulator-runtime", () => {
     vi.useRealTimers();
   });
 
-  it("processes write and clear operations in strict order", () => {
+  it("drains contiguous plain writes without waiting for each commit, gating a clear behind them", () => {
     const { runtime, terminal, writeCallbacks, writeTexts } = createRuntimeWithTerminal();
     const committed: string[] = [];
 
@@ -174,43 +194,64 @@ describe("terminal-emulator-runtime", () => {
         committed.push("first");
       },
     });
-    runtime.clear({
-      onCommitted: () => {
-        committed.push("clear");
-      },
-    });
     runtime.write({
       data: terminalOutput("second"),
       onCommitted: () => {
         committed.push("second");
       },
     });
+    runtime.clear({
+      onCommitted: () => {
+        committed.push("clear");
+      },
+    });
+    runtime.write({
+      data: terminalOutput("third"),
+      onCommitted: () => {
+        committed.push("third");
+      },
+    });
 
-    expect(writeTexts).toEqual(["first"]);
+    // Both plain writes are submitted back-to-back; neither waited on the other's callback.
+    // The clear and the write after it stay queued behind the barrier gate (sentinel write).
+    expect(writeTexts).toEqual(["first", "second"]);
     expect(terminal.resetCalls).toBe(0);
     expect(committed).toEqual([]);
 
+    // writeCallbacks: [0]=first, [1]=second, [2]=barrier-gate sentinel.
     writeCallbacks[0]?.();
-
-    expect(committed).toEqual(["first", "clear"]);
-    expect(terminal.resetCalls).toBe(1);
-    expect(writeTexts).toEqual(["first", "second"]);
-
     writeCallbacks[1]?.();
-    expect(committed).toEqual(["first", "clear", "second"]);
+    expect(committed).toEqual(["first", "second"]);
+    expect(terminal.resetCalls).toBe(0);
+
+    // Resolving the sentinel releases the clear, which resets and then drains "third".
+    writeCallbacks[2]?.();
+    expect(committed).toEqual(["first", "second", "clear"]);
+    expect(terminal.resetCalls).toBe(1);
+    expect(writeTexts).toEqual(["first", "second", "third"]);
+
+    // "third" still commits through its own write callback.
+    writeCallbacks[3]?.();
+    expect(committed).toEqual(["first", "second", "clear", "third"]);
   });
 
-  it("falls back to timeout commit when xterm write callback does not fire", () => {
+  it("falls back to timeout commit for a barrier op when the gate sentinel never fires", () => {
     vi.useFakeTimers();
-    const { runtime } = createRuntimeWithTerminal();
+    const { runtime, terminal } = createRuntimeWithTerminal();
     const onCommitted = vi.fn();
 
-    runtime.write({
+    // restoreOutput is a barrier; if xterm never commits the gate sentinel, the barrier still
+    // applies (and commits) after the 5s safety timeout instead of stalling forever.
+    runtime.restoreOutput({
       data: terminalOutput("stuck"),
       onCommitted,
     });
 
     expect(onCommitted).not.toHaveBeenCalled();
+    expect(terminal.resetCalls).toBe(0);
+    vi.advanceTimersByTime(5_000);
+    // The barrier now runs (write submitted) but its own commit still waits on xterm; advance
+    // again to fire the barrier's own write timeout.
     vi.advanceTimersByTime(5_000);
     expect(onCommitted).toHaveBeenCalledTimes(1);
   });
@@ -239,6 +280,12 @@ describe("terminal-emulator-runtime", () => {
         },
       },
     });
+    // The plain write reports kitty flags synchronously during drain; the snapshot resets
+    // them only after its barrier gate (the sentinel write callback) resolves.
+    expect(inputModeChanges).toEqual([{ kittyKeyboardFlags: 7, win32InputMode: false }]);
+
+    // The plain write carries no onCommitted, so it registers no callback; writeCallbacks[0]
+    // is the barrier gate sentinel.
     writeCallbacks[0]?.();
 
     expect(inputModeChanges).toEqual([
@@ -247,8 +294,8 @@ describe("terminal-emulator-runtime", () => {
     ]);
   });
 
-  it("ignores stale duplicate write callbacks from a previous operation", () => {
-    const { runtime, writeCallbacks } = createRuntimeWithTerminal();
+  it("commits each drained plain write through its own xterm callback", () => {
+    const { runtime, writeTexts, writeCallbacks } = createRuntimeWithTerminal();
     const committed: string[] = [];
 
     runtime.write({
@@ -264,14 +311,89 @@ describe("terminal-emulator-runtime", () => {
       },
     });
 
-    writeCallbacks[0]?.();
-    expect(committed).toEqual(["first"]);
-
-    writeCallbacks[0]?.();
-    expect(committed).toEqual(["first"]);
+    // Both submitted without waiting; each commits independently when its callback fires.
+    expect(writeTexts).toEqual(["first", "second"]);
+    expect(committed).toEqual([]);
 
     writeCallbacks[1]?.();
-    expect(committed).toEqual(["first", "second"]);
+    expect(committed).toEqual(["second"]);
+
+    writeCallbacks[0]?.();
+    expect(committed).toEqual(["second", "first"]);
+  });
+
+  it("applies a snapshot after the writes ahead of it, never interleaving the reset", () => {
+    const { runtime, terminal, writeTexts, writeCallbacks } = createRuntimeWithTerminal();
+    const committed: string[] = [];
+
+    runtime.write({
+      data: terminalOutput("before-a"),
+      onCommitted: () => {
+        committed.push("before-a");
+      },
+    });
+    runtime.write({
+      data: terminalOutput("before-b"),
+      onCommitted: () => {
+        committed.push("before-b");
+      },
+    });
+    runtime.restoreOutput({
+      data: terminalOutput("snap"),
+      onCommitted: () => {
+        committed.push("snap");
+      },
+    });
+
+    // Plain writes are submitted up front; the snapshot's reset+write has not run yet.
+    expect(writeTexts).toEqual(["before-a", "before-b"]);
+
+    // writeCallbacks: [0]=before-a, [1]=before-b, [2]=barrier gate sentinel.
+    // The snapshot is gated behind the prior writes; resolving the gate runs it after them.
+    writeCallbacks[2]?.();
+    expect(writeTexts).toEqual(["before-a", "before-b", "csnap"]);
+
+    // Snapshot commit fires only after its own write callback, strictly after the writes.
+    writeCallbacks[3]?.();
+    expect(committed).toEqual(["snap"]);
+    expect(terminal.resetCalls).toBe(0);
+  });
+
+  it("applies a barrier immediately when no writes precede it, suppressing input at once", () => {
+    const { runtime, writeCallbacks } = createRuntimeWithTerminal();
+    const readSuppressInput = () =>
+      (runtime as unknown as { suppressInput: boolean }).suppressInput;
+
+    expect(readSuppressInput()).toBe(false);
+
+    // No plain writes precede this barrier (mount), so there is nothing to gate: it starts
+    // at once with no sentinel, flipping suppressInput synchronously.
+    runtime.restoreOutput({ data: terminalOutput("snapshot") });
+    expect(readSuppressInput()).toBe(true);
+
+    // writeCallbacks[0] is the barrier's own snapshot write; committing it restores input.
+    writeCallbacks[0]?.();
+    expect(readSuppressInput()).toBe(false);
+  });
+
+  it("gates a barrier behind a preceding plain write before suppressing input", () => {
+    const { runtime, writeCallbacks } = createRuntimeWithTerminal();
+    const readSuppressInput = () =>
+      (runtime as unknown as { suppressInput: boolean }).suppressInput;
+
+    // A plain write is now ungated, so the following barrier must wait on the sentinel.
+    runtime.write({ data: terminalOutput("output") });
+    runtime.restoreOutput({ data: terminalOutput("snapshot") });
+
+    // The plain write carries no onCommitted so it registers no callback; writeCallbacks[0]
+    // is the sentinel gate. suppressInput only flips once the gate resolves the barrier.
+    expect(readSuppressInput()).toBe(false);
+    writeCallbacks[0]?.();
+    expect(readSuppressInput()).toBe(true);
+
+    // writeCallbacks[1] is the barrier's own snapshot write; committing it restores input.
+    writeCallbacks[1]?.();
+    expect(readSuppressInput()).toBe(false);
   });
 
   it("commits pending output operations during unmount to avoid deadlock", () => {
@@ -294,8 +416,21 @@ describe("terminal-emulator-runtime", () => {
     expect(onCommittedB).toHaveBeenCalledTimes(1);
   });
 
+  it("clears ungated writes on unmount so remount barriers apply immediately", () => {
+    const { runtime } = createRuntimeWithTerminal();
+
+    runtime.write({ data: terminalOutput("before unmount") });
+    runtime.unmount();
+
+    const remounted = attachStubTerminal(runtime);
+    runtime.restoreOutput({ data: terminalOutput("restored screen") });
+
+    expect(remounted.writeTexts).toEqual(["\u001bcrestored screen"]);
+    expect(remounted.writeCallbacks).toHaveLength(1);
+  });
+
   it("replays snapshots through a single write without first painting a reset terminal", () => {
-    const { runtime, terminal, writeTexts } = createRuntimeWithTerminal();
+    const { runtime, terminal, writeTexts, writeCallbacks } = createRuntimeWithTerminal();
 
     runtime.renderSnapshot({
       state: {
@@ -313,6 +448,9 @@ describe("terminal-emulator-runtime", () => {
       },
     });
 
+    // The snapshot is a barrier; its write applies after the gate sentinel (writeCallbacks[0]).
+    writeCallbacks[0]?.();
+
     expect(terminal.resetCalls).toBe(0);
     expect(writeTexts).toHaveLength(1);
     expect(writeTexts[0]?.startsWith("\u001bc")).toBe(true);
@@ -320,10 +458,13 @@ describe("terminal-emulator-runtime", () => {
   });
 
   it("restores server-rendered ANSI snapshots through the snapshot write path", () => {
-    const { runtime, terminal, writeTexts } = createRuntimeWithTerminal();
+    const { runtime, terminal, writeTexts, writeCallbacks } = createRuntimeWithTerminal();
 
     runtime.restoreOutput({ data: terminalOutput("restored screen") });
 
+    // restoreOutput is a barrier (suppressInput); it applies after the gate sentinel.
+    writeCallbacks[0]?.();
+
     expect(terminal.resetCalls).toBe(0);
     expect(writeTexts).toEqual(["\u001bcrestored screen"]);
   });

packages/app/src/terminal/runtime/terminal-emulator-runtime.ts

@@ -184,6 +184,15 @@ export class TerminalEmulatorRuntime {
   private outputOperations: TerminalOutputOperation[] = [];
   private inFlightOutputOperation: TerminalOutputOperation | null = null;
   private inFlightOutputOperationTimeout: ReturnType<typeof setTimeout> | null = null;
+  // Plain writes are submitted to xterm without waiting, so their onCommitted callbacks are
+  // pending until xterm commits the write. unmount() disposes xterm before those fire, so we
+  // track them here to flush remaining callbacks and avoid stalling upstream backpressure.
+  private pendingWriteCommits = new Set<() => void>();
+  // True once a plain write has been submitted to xterm that no later barrier has gated
+  // yet. A barrier only needs the sentinel-write gate when this is set; at mount or right
+  // after another barrier it's false and the barrier applies immediately, saving a parse
+  // cycle of latency. Cleared when a barrier starts (it gates every write before it).
+  private hasUngatedWrites = false;
   private readonly inputModeDecoder = new TextDecoder();
   private suppressInput = false;
   private readonly inputModeTracker = new TerminalInputModeTracker();
@@ -759,6 +768,10 @@ export class TerminalEmulatorRuntime {
     for (const operation of pendingOperations) {
       operation.onCommitted?.();
     }
+    for (const commit of Array.from(this.pendingWriteCommits)) {
+      commit();
+    }
+    this.hasUngatedWrites = false;
 
     this.cleanup?.();
     this.cleanup = null;
@@ -776,6 +789,13 @@ export class TerminalEmulatorRuntime {
     this.emitInputModeChange();
   }
 
+  // A barrier op mutates terminal geometry or toggles suppressInput and must apply in
+  // isolation: clear/snapshot reset/resize the terminal, and any write carrying
+  // suppressInput flips input handling around the commit. Plain writes are fast-pathed.
+  private isBarrierOperation(operation: TerminalOutputOperation): boolean {
+    return operation.type !== "write" || Boolean(operation.suppressInput);
+  }
+
   private processOutputQueue(): void {
     if (this.inFlightOutputOperation) {
       return;
@@ -786,12 +806,90 @@ export class TerminalEmulatorRuntime {
       return;
     }
 
+    // Fast path: drain contiguous plain writes back-to-back without waiting for each
+    // write's completion callback. xterm buffers internally and parses in submission
+    // order, so serializing one frame per parse tick only adds latency under burst.
+    while (this.outputOperations[0] && !this.isBarrierOperation(this.outputOperations[0])) {
+      const writeOperation = this.outputOperations.shift();
+      if (!writeOperation) {
+        break;
+      }
+      this.submitWrite(terminal, writeOperation);
+    }
+
     const operation = this.outputOperations.shift();
     if (!operation) {
       return;
     }
 
+    // The next op is a barrier. Before clear()/reset()/resize() touches the terminal, any
+    // ungated plain writes still parsing in xterm's buffer must finish — otherwise a
+    // synchronous reset could interleave with them. When none are ungated (mount, or right
+    // after another barrier) there is nothing to wait for, so apply the barrier at once.
+    if (!this.hasUngatedWrites) {
+      this.startBarrierOperation(terminal, operation);
+      return;
+    }
+    // Otherwise a zero-length sentinel write commits after every outstanding write, so
+    // waiting on its callback gates the barrier behind them. Plain writes never wait.
+    let started = false;
+    const startBarrier = () => {
+      // unmount() clears the in-flight op and disposes the terminal while we wait on the
+      // sentinel; bail if either changed so we don't reset/resize a torn-down terminal.
+      if (started || this.inFlightOutputOperation !== operation || this.terminal !== terminal) {
+        return;
+      }
+      started = true;
+      this.clearInFlightOutputTimeout();
+      this.startBarrierOperation(terminal, operation);
+    };
     this.inFlightOutputOperation = operation;
+    this.inFlightOutputOperationTimeout = setTimeout(startBarrier, OUTPUT_OPERATION_TIMEOUT_MS);
+    try {
+      terminal.write(EMPTY_TERMINAL_OUTPUT, startBarrier);
+    } catch {
+      startBarrier();
+    }
+  }
+
+  private submitWrite(terminal: Terminal, operation: TerminalOutputOperation): void {
+    // Synchronous per-write tracking must run in frame order; doing it here in the drain
+    // loop preserves that ordering even though the writes are submitted without waiting.
+    const text = this.inputModeDecoder.decode(operation.data, { stream: true });
+    const result = this.inputModeTracker.feed(text);
+    if (result.changed) {
+      this.emitInputModeChange();
+    }
+    this.hasUngatedWrites = true;
+    const onCommitted = operation.onCommitted;
+    if (!onCommitted) {
+      try {
+        terminal.write(operation.data);
+      } catch {
+        // Match existing behavior: a failed write still proceeds with no commit callback.
+      }
+      return;
+    }
+    const commit = () => {
+      if (!this.pendingWriteCommits.delete(commit)) {
+        return;
+      }
+      onCommitted();
+    };
+    this.pendingWriteCommits.add(commit);
+    try {
+      terminal.write(operation.data, commit);
+    } catch {
+      commit();
+    }
+  }
+
+  private startBarrierOperation(terminal: Terminal, operation: TerminalOutputOperation): void {
+    this.inFlightOutputOperation = operation;
+    // This barrier gates every write submitted before it (the sentinel guaranteed they
+    // parsed, or there were none), so the next barrier can skip the sentinel until the
+    // next write arrives.
+    this.hasUngatedWrites = false;
     const previousSuppressInput = this.suppressInput;
     if (operation.suppressInput) {
       this.suppressInput = Boolean(operation.suppressInput);

packages/server/src/server/session.ts

@@ -566,6 +566,7 @@ export interface SessionOptions {
   clientCapabilities?: Record<string, unknown> | null;
   onMessage: (msg: SessionOutboundMessage) => void;
   onBinaryMessage?: (frame: Uint8Array) => void;
+  getTransportBufferedAmount?: () => number | null;
   onLifecycleIntent?: (intent: SessionLifecycleIntent) => void;
   logger: pino.Logger;
   downloadTokenStore: DownloadTokenStore;
@@ -757,6 +758,7 @@ export class Session {
   private readonly sessionId: string;
   private readonly onMessage: (msg: SessionOutboundMessage) => void;
   private readonly onBinaryMessage: ((frame: Uint8Array) => void) | null;
+  private readonly getTransportBufferedAmount: () => number | null;
   private readonly onLifecycleIntent: ((intent: SessionLifecycleIntent) => void) | null;
   private readonly sessionLogger: pino.Logger;
   private readonly paseoHome: string;
@@ -861,6 +863,7 @@ export class Session {
       clientCapabilities,
       onMessage,
       onBinaryMessage,
+      getTransportBufferedAmount,
       onLifecycleIntent,
       logger,
       downloadTokenStore,
@@ -906,6 +909,7 @@ export class Session {
     this.sessionId = uuidv4();
     this.onMessage = onMessage;
     this.onBinaryMessage = onBinaryMessage ?? null;
+    this.getTransportBufferedAmount = getTransportBufferedAmount ?? (() => 0);
     this.onLifecycleIntent = onLifecycleIntent ?? null;
     this.downloadTokenStore = downloadTokenStore;
     this.pushTokenStore = pushTokenStore;
@@ -946,6 +950,7 @@ export class Session {
       },
       clientSupportsWrapReflow: () =>
         this.clientCapabilities.has(CLIENT_CAPS.terminalReflowableSnapshot),
+      getClientBufferedAmount: () => this.getTransportBufferedAmount(),
     });
     this.createAgentLifecycleDispatch = new CreateAgentLifecycleDispatch({
       paseoHome: this.paseoHome,
@@ -8687,13 +8692,18 @@ export class Session {
    * Emit a message to the client
    */
   private emit(msg: SessionOutboundMessage): void {
-    this.sessionLogger.trace(
-      {
-        messageType: msg.type,
-        payloadBytes: JSON.stringify(msg).length,
-      },
-      "agent.session.outbound",
-    );
+    // JSON.stringify(msg) is only computed when trace is enabled — it runs for
+    // every outbound message otherwise, and trace is disabled by default.
+    // Optional-chained because test logger stubs don't implement isLevelEnabled.
+    if (this.sessionLogger.isLevelEnabled?.("trace")) {
+      this.sessionLogger.trace(
+        {
+          messageType: msg.type,
+          payloadBytes: JSON.stringify(msg).length,
+        },
+        "agent.session.outbound",
+      );
+    }
     if (
       msg.type === "audio_output" &&
       (process.env.TTS_DEBUG_AUDIO_DIR || isPaseoDictationDebugEnabled()) &&

packages/server/src/server/websocket-server.ts

@@ -2,6 +2,7 @@ import { WebSocket, WebSocketServer } from "ws";
 import type { IncomingMessage, Server as HTTPServer } from "http";
 import { basename, join } from "path";
 import { hostname as getHostname } from "node:os";
+import { monitorEventLoopDelay } from "node:perf_hooks";
 import type { AgentManager } from "./agent/agent-manager.js";
 import type { AgentStorage } from "./agent/agent-storage.js";
 import type { DownloadTokenStore } from "./file-download/token-store.js";
@@ -379,6 +380,7 @@ export class VoiceAssistantWebSocketServer {
   private serverCapabilities: ServerCapabilities | undefined;
   private readonly runtimeMetrics = new WebSocketRuntimeMetricsWindow();
   private runtimeMetricsInterval: ReturnType<typeof setInterval> | null = null;
+  private eventLoopDelayMonitor: ReturnType<typeof monitorEventLoopDelay> | null = null;
   private unsubscribeSpeechReadiness: (() => void) | null = null;
   private unsubscribeDaemonConfigChange: (() => void) | null = null;
 
@@ -561,6 +563,8 @@ export class VoiceAssistantWebSocketServer {
   }
 
   private startRuntimeMetricsInterval(): void {
+    this.eventLoopDelayMonitor = monitorEventLoopDelay({ resolution: 10 });
+    this.eventLoopDelayMonitor.enable();
     const runtimeMetricsInterval = setInterval(() => {
       this.flushRuntimeMetrics();
     }, WS_RUNTIME_METRICS_FLUSH_MS);
@@ -568,6 +572,23 @@ export class VoiceAssistantWebSocketServer {
     (runtimeMetricsInterval as unknown as { unref?: () => void }).unref?.();
   }
 
+  // Main-loop stall visibility: terminal frames and agent traffic share one event
+  // loop, so delay percentiles here are the ground truth for "the daemon is busy".
+  private snapshotEventLoopDelay(): { p50Ms: number; p99Ms: number; maxMs: number } | null {
+    const monitor = this.eventLoopDelayMonitor;
+    if (!monitor) {
+      return null;
+    }
+    const toMs = (nanoseconds: number): number => Math.round(nanoseconds / 1e5) / 10;
+    const snapshot = {
+      p50Ms: toMs(monitor.percentile(50)),
+      p99Ms: toMs(monitor.percentile(99)),
+      maxMs: toMs(monitor.max),
+    };
+    monitor.reset();
+    return snapshot;
+  }
+
   private verifyWsUpgrade(
     req: IncomingMessage,
     allowedOrigins: Set<string>,
@@ -678,6 +699,8 @@ export class VoiceAssistantWebSocketServer {
       this.runtimeMetricsInterval = null;
     }
     this.flushRuntimeMetrics({ final: true });
+    this.eventLoopDelayMonitor?.disable();
+    this.eventLoopDelayMonitor = null;
 
     const uniqueConnections = new Set<SessionConnection>([
       ...this.sessions.values(),
@@ -861,6 +884,22 @@ export class VoiceAssistantWebSocketServer {
         }
         this.sendBinaryToConnection(connection, frame);
       },
+      getTransportBufferedAmount: () => {
+        if (!connection) {
+          return null;
+        }
+        // Relay-attached sockets are a WebSocketLike that doesn't expose
+        // bufferedAmount. Return null when no socket gives a signal so the
+        // terminal fallback can't mistake "no signal" for "client keeping up";
+        // a direct ws reports its real buffered bytes (0 when drained).
+        let maxBuffered: number | null = null;
+        for (const socket of connection.sockets) {
+          if (typeof socket.bufferedAmount === "number") {
+            maxBuffered = Math.max(maxBuffered ?? 0, socket.bufferedAmount);
+          }
+        }
+        return maxBuffered;
+      },
       onLifecycleIntent: (intent) => {
         this.onLifecycleIntent?.(intent);
       },
@@ -1641,6 +1680,7 @@ export class VoiceAssistantWebSocketServer {
         outboundAgentStreamAgentsTop: runtimeMetrics.outboundAgentStreamAgentsTop,
         outboundBinaryFrameTypesTop: runtimeMetrics.outboundBinaryFrameTypesTop,
         bufferedAmount: runtimeMetrics.bufferedAmount,
+        eventLoopDelay: this.snapshotEventLoopDelay(),
         runtime: sessionMetrics,
         latency: runtimeMetrics.latency,
         agents: agentSnapshot,

packages/server/src/terminal/terminal-output-coalescer.test.ts

@@ -3,6 +3,7 @@ import { TerminalOutputCoalescer } from "./terminal-output-coalescer.js";
 
 function createHarness() {
   let nextTimer = 1;
+  let now = 1000;
   const scheduled = new Map<number, { callback: () => void; delayMs: number }>();
   const flushes: Array<{ payload: string; chars: number; bytes: number }> = [];
   const coalescer = new TerminalOutputCoalescer({
@@ -17,6 +18,7 @@ function createHarness() {
         scheduled.delete(id);
       }) as typeof clearTimeout,
     },
+    now: () => now,
     onFlush: ({ payload, chars, bytes }) => {
       flushes.push({ payload: payload.toString("utf8"), chars, bytes });
     },
@@ -26,6 +28,9 @@ function createHarness() {
     coalescer,
     flushes,
     scheduled,
+    advance(ms: number) {
+      now += ms;
+    },
     runScheduled() {
       const callbacks = Array.from(scheduled.values());
       scheduled.clear();
@@ -37,10 +42,22 @@ function createHarness() {
 }
 
 describe("TerminalOutputCoalescer", () => {
-  test("coalesces output into one 5ms flush window", () => {
-    const { coalescer, flushes, runScheduled, scheduled } = createHarness();
+  test("flushes the first chunk immediately on the leading edge", () => {
+    const { coalescer, flushes, scheduled } = createHarness();
 
     coalescer.handle("a");
+
+    // No timer scheduled: the leading-edge flush happened synchronously.
+    expect(scheduled.size).toBe(0);
+    expect(flushes).toEqual([{ payload: "a", chars: 1, bytes: 1 }]);
+  });
+
+  test("coalesces a burst that follows the leading-edge flush into one trailing flush", () => {
+    const { coalescer, flushes, runScheduled, scheduled } = createHarness();
+
+    // First chunk flushes immediately (leading edge).
+    coalescer.handle("a");
+    // Subsequent chunks within the window accumulate behind a trailing timer.
     coalescer.handle("b");
     coalescer.handle("é");
 
@@ -48,32 +65,86 @@ describe("TerminalOutputCoalescer", () => {
     expect(Array.from(scheduled.values())).toEqual([
       { callback: expect.any(Function), delayMs: 5 },
     ]);
-    expect(flushes).toEqual([]);
+    expect(flushes).toEqual([{ payload: "a", chars: 1, bytes: 1 }]);
 
     runScheduled();
 
-    expect(flushes).toEqual([{ payload: "abé", chars: 3, bytes: 4 }]);
+    expect(flushes).toEqual([
+      { payload: "a", chars: 1, bytes: 1 },
+      { payload: "bé", chars: 2, bytes: 3 },
+    ]);
+  });
+
+  test("flushes immediately again once the window has elapsed", () => {
+    const { coalescer, flushes, advance, scheduled } = createHarness();
+
+    coalescer.handle("a");
+    expect(flushes).toEqual([{ payload: "a", chars: 1, bytes: 1 }]);
+
+    advance(5);
+    coalescer.handle("b");
+
+    expect(scheduled.size).toBe(0);
+    expect(flushes).toEqual([
+      { payload: "a", chars: 1, bytes: 1 },
+      { payload: "b", chars: 1, bytes: 1 },
+    ]);
   });
 
   test("manual flush drains pending output and cancels the scheduled flush", () => {
     const { coalescer, flushes, runScheduled, scheduled } = createHarness();
 
+    // Leading-edge flush, then accumulate a burst behind the trailing timer.
     coalescer.handle("hello");
+    coalescer.handle(" world");
     coalescer.flush();
     runScheduled();
 
     expect(scheduled.size).toBe(0);
-    expect(flushes).toEqual([{ payload: "hello", chars: 5, bytes: 5 }]);
+    expect(flushes).toEqual([
+      { payload: "hello", chars: 5, bytes: 5 },
+      { payload: " world", chars: 6, bytes: 6 },
+    ]);
   });
 
   test("dispose drops pending output", () => {
     const { coalescer, flushes, runScheduled, scheduled } = createHarness();
 
+    // First chunk flushes on the leading edge; second one stays pending.
+    coalescer.handle("done");
     coalescer.handle("pending");
     coalescer.dispose();
     runScheduled();
 
     expect(scheduled.size).toBe(0);
+    expect(flushes).toEqual([{ payload: "done", chars: 4, bytes: 4 }]);
+  });
+
+  test("markFlushed keeps the next chunk on the trailing path", () => {
+    const { coalescer, flushes, runScheduled, scheduled } = createHarness();
+
+    // Simulate a frame emitted out-of-band (e.g. a snapshot) right before output.
+    coalescer.markFlushed();
+    coalescer.handle("post-snapshot");
+
+    // No immediate flush: the chunk waits for the trailing timer.
+    expect(scheduled.size).toBe(1);
     expect(flushes).toEqual([]);
+
+    runScheduled();
+    expect(flushes).toEqual([{ payload: "post-snapshot", chars: 13, bytes: 13 }]);
+  });
+
+  test("preserves ordering across leading and trailing flushes", () => {
+    const { coalescer, flushes, runScheduled, advance } = createHarness();
+
+    coalescer.handle("1");
+    coalescer.handle("2");
+    coalescer.handle("3");
+    runScheduled();
+    advance(5);
+    coalescer.handle("4");
+
+    expect(flushes.map((f) => f.payload)).toEqual(["1", "23", "4"]);
   });
 });

packages/server/src/terminal/terminal-output-coalescer.ts

@@ -12,6 +12,7 @@ export interface TerminalOutputCoalescerFlush {
 export interface TerminalOutputCoalescerOptions {
   timers: TerminalOutputCoalescerTimers;
   flushDelayMs?: number;
+  now?: () => number;
   onFlush: (payload: TerminalOutputCoalescerFlush) => void;
 }
 
@@ -20,15 +21,18 @@ const DEFAULT_FLUSH_DELAY_MS = 5;
 export class TerminalOutputCoalescer {
   private readonly timers: TerminalOutputCoalescerTimers;
   private readonly flushDelayMs: number;
+  private readonly now: () => number;
   private readonly onFlush: (payload: TerminalOutputCoalescerFlush) => void;
   private chunks: Buffer[] = [];
   private bytes = 0;
   private chars = 0;
   private flushTimer: ReturnType<typeof setTimeout> | null = null;
+  private lastFlushAt: number | null = null;
 
   constructor(options: TerminalOutputCoalescerOptions) {
     this.timers = options.timers;
     this.flushDelayMs = options.flushDelayMs ?? DEFAULT_FLUSH_DELAY_MS;
+    this.now = options.now ?? Date.now;
     this.onFlush = options.onFlush;
   }
 
@@ -42,7 +46,16 @@ export class TerminalOutputCoalescer {
     this.bytes += chunk.byteLength;
     this.chars += data.length;
 
+    // Leading edge: if nothing is pending and no flush happened within the last
+    // flushDelayMs, flush immediately so interactive echo isn't delayed a full
+    // window. Otherwise accumulate and let the trailing timer drain the burst.
     if (!this.flushTimer) {
+      const elapsed =
+        this.lastFlushAt === null ? Number.POSITIVE_INFINITY : this.now() - this.lastFlushAt;
+      if (elapsed >= this.flushDelayMs) {
+        this.flush();
+        return;
+      }
       this.flushTimer = this.timers.setTimeout(() => {
         this.flushTimer = null;
         this.flush();
@@ -62,10 +75,19 @@ export class TerminalOutputCoalescer {
     const bytes = this.bytes;
     const chars = this.chars;
     this.clearPending();
+    this.lastFlushAt = this.now();
 
     this.onFlush({ payload, bytes, chars });
   }
 
+  // Record that a frame was emitted out-of-band on this stream (e.g. a snapshot
+  // frame sent directly, bypassing the coalescer). This keeps the next handled
+  // chunk on the trailing path instead of leading-edge flushing it back-to-back
+  // with that frame, preserving the small gap consumers rely on after a snapshot.
+  markFlushed(): void {
+    this.lastFlushAt = this.now();
+  }
+
   dispose(): void {
     this.clearTimer();
     this.clearPending();

packages/server/src/terminal/terminal-restore.ts

@@ -9,6 +9,13 @@ import type { TerminalStateSnapshot, TerminalStateSnapshotOptions } from "./term
 
 export const MAX_TERMINAL_OUTPUT_FRAME_BYTES = 256 * 1024;
 
+// A client is only forced onto the snapshot catch-up path once its transport is
+// genuinely backed up — measured by the socket's bufferedAmount, not by how much
+// the terminal has produced. Set well above any normal in-flight burst (a
+// keeping-up client drains continuously and never approaches this) but low
+// enough that a stalled socket crosses it long before a large burst finishes.
+export const MAX_CLIENT_BUFFERED_BYTES = 4 * 1024 * 1024;
+
 const DEFAULT_VISIBLE_RESTORE_SCROLLBACK_LINES = 200;
 const MAX_VISIBLE_RESTORE_SCROLLBACK_LINES = 500;
 

packages/server/src/terminal/terminal-session-controller.test.ts

@@ -168,6 +168,12 @@ async function flushMicrotasks(): Promise<void> {
   await new Promise((resolve) => setTimeout(resolve, 0));
 }
 
+// The coalescer drains on a 5ms trailing timer; wait past it (and any async
+// snapshot round-trip it kicks off) before asserting on emitted frames.
+async function waitForCoalescerFlush(): Promise<void> {
+  await new Promise((resolve) => setTimeout(resolve, 30));
+}
+
 describe("terminal-session-controller wrap-flag gating", () => {
   function setup(clientSupportsWrapReflow?: () => boolean): {
     controller: TerminalSessionController;
@@ -387,3 +393,133 @@ describe("terminal-session-controller subdirectory aggregation", () => {
     ]);
   });
 });
+
+describe("terminal-session-controller backpressure snapshot fallback", () => {
+  async function setup(getClientBufferedAmount: () => number | null): Promise<{
+    pushOutput: (data: string) => void;
+    frames: TerminalStreamFrame[];
+  }> {
+    let terminalListener: ((message: ServerMessage) => void) | null = null;
+    const terminal: TerminalSession = {
+      id: "term-1",
+      name: "Terminal",
+      cwd: "/tmp",
+      send: vi.fn(),
+      subscribe: (listener) => {
+        terminalListener = listener;
+        // Legacy stream: a snapshot arrives on subscribe (one Snapshot frame),
+        // after which output streams through the coalescer as Output frames.
+        queueMicrotask(() =>
+          listener({ type: "snapshot", state: terminalState("live"), revision: 1 }),
+        );
+        return vi.fn();
+      },
+      onExit: () => vi.fn(),
+      onCommandFinished: () => vi.fn(),
+      onTitleChange: () => vi.fn(),
+      getSize: () => ({ rows: 1, cols: 80 }),
+      getState: () => terminalState("live"),
+      getStateSnapshot: () => ({ state: terminalState("live"), revision: 1 }),
+      getReplayPreamble: () => "",
+      getTitle: () => undefined,
+      setTitle: vi.fn(),
+      getExitInfo: () => null,
+      kill: vi.fn(),
+      killAndWait: vi.fn(),
+    };
+    const terminalManager = {
+      getTerminals: vi.fn(),
+      createTerminal: vi.fn(),
+      registerCwdEnv: vi.fn(),
+      getTerminal: vi.fn(() => terminal),
+      getTerminalState: vi.fn(() =>
+        Promise.resolve<TerminalStateSnapshot>({ state: terminalState("live"), revision: 1 }),
+      ),
+      setTerminalTitle: vi.fn(),
+      killTerminal: vi.fn(),
+      killTerminalAndWait: vi.fn(),
+      captureTerminal: vi.fn(),
+      listDirectories: vi.fn(() => []),
+      killAll: vi.fn(),
+      subscribeTerminalsChanged: vi.fn(() => vi.fn()),
+    } as unknown as TerminalManager;
+
+    const frames: TerminalStreamFrame[] = [];
+    const controller = new TerminalSessionController({
+      terminalManager,
+      emit: vi.fn(),
+      emitBinary: (bytes) => {
+        const frame = decodeTerminalStreamFrame(bytes);
+        if (frame) {
+          frames.push(frame);
+        }
+      },
+      hasBinaryChannel: () => true,
+      isPathWithinRoot: () => false,
+      sessionLogger: createLogger(),
+      getClientBufferedAmount,
+    });
+
+    await controller.dispatch({
+      type: "subscribe_terminal_request",
+      terminalId: "term-1",
+      requestId: "req-1",
+    });
+    await waitForCoalescerFlush();
+    // Drop the initial subscribe snapshot frame so each test only sees frames
+    // produced by the output it pushes.
+    frames.length = 0;
+
+    return {
+      pushOutput: (data) => terminalListener?.({ type: "output", data, revision: 2 }),
+      frames,
+    };
+  }
+
+  test("streams all output without a snapshot when the client keeps up", async () => {
+    const { pushOutput, frames } = await setup(() => 0);
+
+    const chunk = "x".repeat(300 * 1024);
+    pushOutput(chunk);
+    await waitForCoalescerFlush();
+
+    expect(frames.some((frame) => frame.opcode === TerminalStreamOpcode.Snapshot)).toBe(false);
+    const outputFrames = frames.filter((frame) => frame.opcode === TerminalStreamOpcode.Output);
+    expect(outputFrames.length).toBeGreaterThan(0);
+    const receivedBytes = outputFrames.reduce(
+      (total, frame) => total + frame.payload.byteLength,
+      0,
+    );
+    expect(receivedBytes).toBe(Buffer.byteLength(chunk, "utf8"));
+  });
+
+  test("falls back to a snapshot and resets the byte counter when the client is backed up", async () => {
+    const { pushOutput, frames } = await setup(() => 8 * 1024 * 1024);
+
+    pushOutput("y".repeat(300 * 1024));
+    await waitForCoalescerFlush();
+
+    expect(frames.some((frame) => frame.opcode === TerminalStreamOpcode.Snapshot)).toBe(true);
+
+    // After the snapshot the byte counter is reset, so a small follow-up chunk
+    // streams as Output rather than tripping the fallback again.
+    frames.length = 0;
+    pushOutput("z".repeat(1024));
+    await waitForCoalescerFlush();
+
+    expect(frames.some((frame) => frame.opcode === TerminalStreamOpcode.Snapshot)).toBe(false);
+    expect(frames.some((frame) => frame.opcode === TerminalStreamOpcode.Output)).toBe(true);
+  });
+
+  test("falls back to a snapshot at the byte threshold when no backpressure signal exists", async () => {
+    // A null reading means the transport (e.g. the multiplexed relay socket) gives
+    // no signal; we can't distinguish a slow client from a fast one, so we keep the
+    // unconditional catch-up so a slow relay client can't fall unboundedly behind.
+    const { pushOutput, frames } = await setup(() => null);
+
+    pushOutput("r".repeat(300 * 1024));
+    await waitForCoalescerFlush();
+
+    expect(frames.some((frame) => frame.opcode === TerminalStreamOpcode.Snapshot)).toBe(true);
+  });
+});

packages/server/src/terminal/terminal-session-controller.ts

@@ -22,6 +22,7 @@ import {
 } from "@getpaseo/protocol/binary-frames/index";
 import { TerminalOutputCoalescer } from "./terminal-output-coalescer.js";
 import {
+  MAX_CLIENT_BUFFERED_BYTES,
   MAX_TERMINAL_OUTPUT_FRAME_BYTES,
   encodeLegacyTerminalSnapshotFrame,
   encodeTerminalRestoreFrame,
@@ -70,6 +71,14 @@ export interface TerminalSessionControllerOptions {
   // daemon attaches per-row soft-wrap flags to snapshots; otherwise it omits them
   // so old (strict-schema) clients still parse the snapshot.
   clientSupportsWrapReflow?: () => boolean;
+  // Current max bytes queued on the client's transport(s) but not yet sent.
+  // Drives the snapshot catch-up fallback: a keeping-up client reports ~0 and
+  // keeps streaming; a backed-up client trips the snapshot path. Defaults to a
+  // constant 0 (no backpressure signal) so callers without a transport always
+  // stream.
+  // Bytes queued on the client transport but not yet sent, or null when the
+  // transport exposes no backpressure signal (e.g. the multiplexed relay socket).
+  getClientBufferedAmount?: () => number | null;
 }
 
 export interface TerminalSessionControllerMetrics {
@@ -111,6 +120,7 @@ export class TerminalSessionController {
   private readonly sessionLogger: pino.Logger;
   private readonly listTerminalWorkspaceRoots: () => Promise<readonly string[]>;
   private readonly clientSupportsWrapReflow: () => boolean;
+  private readonly getClientBufferedAmount: () => number | null;
 
   private readonly subscribedDirectories = new Set<string>();
   private unsubscribeTerminalsChanged: (() => void) | null = null;
@@ -128,6 +138,7 @@ export class TerminalSessionController {
     this.sessionLogger = options.sessionLogger;
     this.listTerminalWorkspaceRoots = options.listTerminalWorkspaceRoots ?? (async () => []);
     this.clientSupportsWrapReflow = options.clientSupportsWrapReflow ?? (() => false);
+    this.getClientBufferedAmount = options.getClientBufferedAmount ?? (() => 0);
   }
 
   start(): void {
@@ -757,7 +768,21 @@ export class TerminalSessionController {
             return;
           }
           activeStream.outputBytesSinceSnapshot += payload.byteLength;
-          if (activeStream.outputBytesSinceSnapshot > MAX_TERMINAL_OUTPUT_FRAME_BYTES) {
+          // Catch up via a snapshot only when the client is BOTH far behind in
+          // produced output AND actually backed up on the wire. A client that
+          // keeps draining reports ~0 buffered, so it streams continuously even
+          // past the byte threshold. outputBytesSinceSnapshot keeps accumulating
+          // in that case — it's harmless, it only gates the snapshot decision at
+          // the instant backpressure appears, and trySendSnapshot resets it to 0.
+          // A null reading means the transport exposes no backpressure signal
+          // (e.g. the multiplexed relay socket); there we can't tell a slow client
+          // from a fast one, so fall back unconditionally at the byte threshold to
+          // keep a slow relay client from falling unboundedly behind.
+          const clientBufferedAmount = this.getClientBufferedAmount();
+          if (
+            activeStream.outputBytesSinceSnapshot > MAX_TERMINAL_OUTPUT_FRAME_BYTES &&
+            (clientBufferedAmount === null || clientBufferedAmount > MAX_CLIENT_BUFFERED_BYTES)
+          ) {
             activeStream.restore = resolveRestoreAfterOutputOverflow(activeStream.restore);
             activeStream.needsSnapshot = true;
             void this.trySendSnapshot(activeStream);
@@ -877,6 +902,9 @@ export class TerminalSessionController {
         snapshot,
       }),
     );
+    // The snapshot frame went out-of-band; keep the replay that follows on the
+    // coalescer's trailing path so it doesn't flush back-to-back with it.
+    activeStream.outputCoalescer.markFlushed();
     return { shouldContinue: true, replayRevision: snapshot.revision };
   }
 
@@ -908,6 +936,9 @@ export class TerminalSessionController {
         snapshot,
       }),
     );
+    // The restore frame went out-of-band; keep the replay that follows on the
+    // coalescer's trailing path so it doesn't flush back-to-back with it.
+    activeStream.outputCoalescer.markFlushed();
     return { shouldContinue: true, replayRevision: snapshot.revision };
   }
 

packages/server/src/terminal/terminal-worker-process.ts

@@ -1,14 +1,17 @@
 import { createTerminalManager } from "./terminal-manager.js";
 import { captureTerminalLines } from "./terminal-capture.js";
-import type { TerminalSession } from "./terminal.js";
+import { TerminalOutputCoalescer } from "./terminal-output-coalescer.js";
+import type { TerminalSession, TerminalStateSnapshotOptions } from "./terminal.js";
 import type {
   TerminalWorkerRequest,
+  TerminalWorkerStateResult,
   TerminalWorkerToParentMessage,
   WorkerTerminalInfo,
 } from "./terminal-worker-protocol.js";
 
 const manager = createTerminalManager();
 const unsubscribeByTerminalId = new Map<string, Array<() => void>>();
+const outputCoalescerByTerminalId = new Map<string, TerminalOutputCoalescer>();
 let ipcClosing = false;
 
 function sendToParent(message: TerminalWorkerToParentMessage): void {
@@ -26,6 +29,16 @@ function sendToParent(message: TerminalWorkerToParentMessage): void {
   }
 }
 
+function buildTerminalStateResult(
+  session: TerminalSession | undefined,
+  options?: TerminalStateSnapshotOptions,
+): TerminalWorkerStateResult {
+  if (!session) {
+    return null;
+  }
+  return { ...session.getStateSnapshot(options), replayPreamble: session.getReplayPreamble() };
+}
+
 function toTerminalInfo(session: TerminalSession): WorkerTerminalInfo {
   return {
     id: session.id,
@@ -47,12 +60,44 @@ function clearTerminalSubscriptions(terminalId: string): void {
     }
   }
   unsubscribeByTerminalId.delete(terminalId);
+  const coalescer = outputCoalescerByTerminalId.get(terminalId);
+  if (coalescer) {
+    coalescer.dispose();
+    outputCoalescerByTerminalId.delete(terminalId);
+  }
 }
 
 function watchTerminal(session: TerminalSession): void {
   clearTerminalSubscriptions(session.id);
 
+  // Coalesce pty output chunks into a single IPC message per ~5ms window so a
+  // burst of small chunks no longer costs one process.send each. The batch
+  // carries the LAST chunk's revision (the highest) so downstream snapshot
+  // replay dedup stays correct.
+  let pendingOutputRevision: number | undefined;
+  const outputCoalescer = new TerminalOutputCoalescer({
+    timers: { setTimeout, clearTimeout },
+    onFlush: ({ payload }) => {
+      const revision = pendingOutputRevision;
+      pendingOutputRevision = undefined;
+      sendToParent({
+        type: "terminalMessage",
+        terminalId: session.id,
+        message: { type: "output", data: payload.toString("utf8"), revision },
+      });
+    },
+  });
+  outputCoalescerByTerminalId.set(session.id, outputCoalescer);
+
   const unsubscribeMessage = session.subscribe((message) => {
+    if (message.type === "output") {
+      pendingOutputRevision = message.revision;
+      outputCoalescer.handle(message.data);
+      return;
+    }
+    // Non-output messages (snapshot/snapshotReady/titleChange) must not jump
+    // ahead of buffered output: flush the coalescer first, then forward.
+    outputCoalescer.flush();
     sendToParent({
       type: "terminalMessage",
       terminalId: session.id,
@@ -60,6 +105,7 @@ function watchTerminal(session: TerminalSession): void {
     });
   });
   const unsubscribeExit = session.onExit((info) => {
+    outputCoalescer.flush();
     clearTerminalSubscriptions(session.id);
     sendToParent({
       type: "terminalExit",
@@ -68,6 +114,7 @@ function watchTerminal(session: TerminalSession): void {
     });
   });
   const unsubscribeTitle = session.onTitleChange((title) => {
+    outputCoalescer.flush();
     sendToParent({
       type: "terminalTitleChange",
       terminalId: session.id,
@@ -75,6 +122,7 @@ function watchTerminal(session: TerminalSession): void {
     });
   });
   const unsubscribeCommandFinished = session.onCommandFinished((info) => {
+    outputCoalescer.flush();
     sendToParent({
       type: "terminalCommandFinished",
       terminalId: session.id,
@@ -171,12 +219,16 @@ async function handleRequest(message: TerminalWorkerRequest): Promise<void> {
     }
 
     case "getTerminalState": {
-      const session = manager.getTerminal(message.terminalId);
+      // Flush buffered output before snapshotting: the headless state already includes it,
+      // so if the coalescer emitted it afterward (in a batch carrying a revision past the
+      // snapshot's) the controller's revision dedup wouldn't drop it and the client would
+      // see the bytes twice. Flushing first sends them with a revision <= the snapshot's.
+      outputCoalescerByTerminalId.get(message.terminalId)?.flush();
       sendToParent({
         type: "response",
         requestId: message.requestId,
         ok: true,
-        result: session?.getStateSnapshot(message.options) ?? null,
+        result: buildTerminalStateResult(manager.getTerminal(message.terminalId), message.options),
       });
       return;
     }

packages/server/src/terminal/terminal-worker-protocol.ts

@@ -143,4 +143,6 @@ export type TerminalWorkerEvent =
 export type TerminalWorkerToParentMessage = TerminalWorkerResponse | TerminalWorkerEvent;
 
 export type TerminalWorkerCaptureResult = CaptureTerminalLinesResult;
+// The worker fills TerminalStateSnapshot.replayPreamble on getTerminalState so
+// the parent can cache the input-mode preamble instead of re-deriving it.
 export type TerminalWorkerStateResult = TerminalStateSnapshot | null;

packages/server/src/terminal/terminal.test.ts

@@ -334,6 +334,50 @@ describe("createTerminal", () => {
     expect(Array.isArray(exitInfo.lastOutputLines)).toBe(true);
     expect(session.getExitInfo()).toEqual(exitInfo);
   });
+
+  it("retains the final lines after output far exceeds the recent-output char cap", async () => {
+    // Emit many small chunks whose total length is well past the 16000-char
+    // recent-output cap, so the chunk-trimming path runs repeatedly. The final
+    // distinctive lines must survive in the exit summary. The process is kept
+    // alive (no self-exit) so we can wait until the last chunk has actually been
+    // parsed by the headless terminal before killing — otherwise the exit
+    // summary races the async xterm parse and reads a stale buffer.
+    const session = trackSession(
+      await createTerminal({
+        cwd: realpathSync(tmpdir()),
+        command: process.execPath,
+        args: [
+          "-e",
+          "for (let i = 0; i < 3000; i++) { process.stdout.write(`line-${i}\\n`); } setInterval(() => {}, 1000);",
+        ],
+      }),
+    );
+
+    // Poll the parsed buffer until the final line lands, so the exit summary
+    // (which reads the headless buffer) can't race the async xterm parse.
+    const finalLineParsed = (): boolean => {
+      const state = session.getState();
+      const rowText = (row: { char: string }[]): string => row.map((cell) => cell.char).join("");
+      return [...state.scrollback, ...state.grid].some((row) => rowText(row).includes("line-2999"));
+    };
+    const deadline = Date.now() + 10000;
+    while (!finalLineParsed()) {
+      if (Date.now() > deadline) {
+        throw new Error("Timed out waiting for the final line to be parsed");
+      }
+      await new Promise((resolve) => setTimeout(resolve, 20));
+    }
+
+    const exitInfo = await new Promise<NonNullable<ReturnType<TerminalSession["getExitInfo"]>>>(
+      (resolve) => {
+        session.onExit((info) => resolve(info));
+        session.kill();
+      },
+    );
+
+    expect(exitInfo.lastOutputLines.length).toBeGreaterThan(0);
+    expect(exitInfo.lastOutputLines[exitInfo.lastOutputLines.length - 1]).toBe("line-2999");
+  });
 });
 
 describe.skipIf(isPlatform("win32"))("send input", () => {

packages/server/src/terminal/terminal.ts

@@ -36,6 +36,9 @@ export interface TerminalCommandFinishedInfo {
 export interface TerminalStateSnapshot {
   state: TerminalState;
   revision: number;
+  // Input-mode replay preamble at snapshot time. Populated by the terminal
+  // worker so the daemon main loop doesn't have to re-derive it from output.
+  replayPreamble?: string;
 }
 
 export interface TerminalStateSnapshotOptions {
@@ -58,7 +61,7 @@ export type ClientMessage =
 export type ServerMessage =
   | { type: "output"; data: string; revision?: number }
   | { type: "snapshot"; state: TerminalState; revision?: number }
-  | { type: "snapshotReady"; revision?: number }
+  | { type: "snapshotReady"; revision?: number; replayPreamble?: string }
   | { type: "titleChange"; title?: string };
 
 export interface TerminalSession {
@@ -606,7 +609,12 @@ export async function createTerminal(options: CreateTerminalOptions): Promise<Te
   let processExited = false;
   const processExitWaiters = new Set<() => void>();
   let exitInfo: TerminalExitInfo | null = null;
-  let recentOutputText = "";
+  // Recent output is retained as whole chunks plus a running char length so we
+  // avoid reallocating a ~16KB string on every pty chunk. We keep enough whole
+  // chunks that their join always contains at least the last
+  // TERMINAL_EXIT_OUTPUT_CHAR_LIMIT chars; the exact tail is sliced at exit.
+  const recentOutputChunks: string[] = [];
+  let recentOutputLength = 0;
   let title: string | undefined;
   let titleMode: "auto" | "manual" = presetTitle?.trim() ? "manual" : "auto";
   let pendingTitle: string | undefined;
@@ -777,7 +785,10 @@ export async function createTerminal(options: CreateTerminalOptions): Promise<Te
       lastOutputLines:
         lastOutputLines.length > 0
           ? lastOutputLines
-          : extractLastOutputLinesFromText(recentOutputText, TERMINAL_EXIT_OUTPUT_LINE_LIMIT),
+          : extractLastOutputLinesFromText(
+              recentOutputChunks.join("").slice(-TERMINAL_EXIT_OUTPUT_CHAR_LIMIT),
+              TERMINAL_EXIT_OUTPUT_LINE_LIMIT,
+            ),
     };
   }
 
@@ -803,6 +814,8 @@ export async function createTerminal(options: CreateTerminalOptions): Promise<Te
     }
     disposed = true;
     pendingInput = "";
+    recentOutputChunks.length = 0;
+    recentOutputLength = 0;
     inputModeTracker.reset();
     if (inputFlushImmediate) {
       clearImmediate(inputFlushImmediate);
@@ -837,9 +850,23 @@ export async function createTerminal(options: CreateTerminalOptions): Promise<Te
     for (const response of inputModeUpdate.responses) {
       ptyProcess.write(response);
     }
-    recentOutputText = `${recentOutputText}${data}`;
-    if (recentOutputText.length > TERMINAL_EXIT_OUTPUT_CHAR_LIMIT) {
-      recentOutputText = recentOutputText.slice(-TERMINAL_EXIT_OUTPUT_CHAR_LIMIT);
+    recentOutputChunks.push(data);
+    recentOutputLength += data.length;
+    // Drop whole leading chunks while the rest still covers the char limit, so
+    // the retained join always contains at least the last limit chars.
+    while (
+      recentOutputChunks.length > 1 &&
+      recentOutputLength - recentOutputChunks[0].length >= TERMINAL_EXIT_OUTPUT_CHAR_LIMIT
+    ) {
+      recentOutputLength -= recentOutputChunks[0].length;
+      recentOutputChunks.shift();
+    }
+    // We never drop the last chunk, so a single chunk larger than the cap would
+    // grow the buffer unbounded; slice its tail to keep the cap hard.
+    if (recentOutputChunks.length === 1 && recentOutputLength > TERMINAL_EXIT_OUTPUT_CHAR_LIMIT) {
+      const tail = recentOutputChunks[0].slice(-TERMINAL_EXIT_OUTPUT_CHAR_LIMIT);
+      recentOutputChunks[0] = tail;
+      recentOutputLength = tail.length;
     }
     writeOutputToHeadless(data);
   });
@@ -993,7 +1020,13 @@ export async function createTerminal(options: CreateTerminalOptions): Promise<Te
       if (!disposed && active && listeners.has(subscriptionListener)) {
         snapshotDelivered = true;
         if (initialSnapshot === "ready") {
-          listener({ type: "snapshotReady", revision: stateRevision });
+          // Carry the input-mode preamble so the snapshot-less "ready" path
+          // (live restore) can replay it without a separate state fetch.
+          listener({
+            type: "snapshotReady",
+            revision: stateRevision,
+            replayPreamble: getReplayPreamble(),
+          });
         } else {
           listener({ type: "snapshot", ...getStateSnapshot() });
         }

packages/server/src/terminal/worker-terminal-manager.test.ts

@@ -1,6 +1,7 @@
 import { afterEach, expect, it } from "vitest";
+import { isPlatform } from "../test-utils/platform.js";
 import { EventEmitter } from "node:events";
-import { existsSync, mkdirSync, mkdtempSync, readFileSync, rmSync } from "node:fs";
+import { existsSync, mkdirSync, mkdtempSync, readFileSync, rmSync, writeFileSync } from "node:fs";
 import { join } from "node:path";
 import { tmpdir } from "node:os";
 import { createWorkerTerminalManager } from "./worker-terminal-manager.js";
@@ -12,6 +13,8 @@ import type {
   TerminalWorkerToParentMessage,
 } from "./terminal-worker-protocol.js";
 
+type TerminalRow = TerminalState["grid"][number];
+
 function nodeTerminalCommand(script: string): { command: string; args: string[] } {
   return {
     command: process.execPath,
@@ -38,15 +41,30 @@ function getVisibleText(session: TerminalSession): string {
   return getVisibleTextFromState(session.getState());
 }
 
+function rowToText(row: TerminalRow): string {
+  return row
+    .map((cell) => cell.char)
+    .join("")
+    .trimEnd();
+}
+
 function getVisibleTextFromState(state: TerminalState): string {
-  return state.grid
-    .map((row) =>
-      row
-        .map((cell) => cell.char)
-        .join("")
-        .trimEnd(),
-    )
-    .join("\n");
+  return state.grid.map(rowToText).join("\n");
+}
+
+function getRenderedTextFromState(state: TerminalState): string {
+  return [...state.scrollback, ...state.grid].map(rowToText).join("\n");
+}
+
+function readRenderedTokenIndexesFromState(state: TerminalState): number[] {
+  const indexes: number[] = [];
+  for (const match of getRenderedTextFromState(state).matchAll(/\[(\d+)\]/g)) {
+    const value = match[1];
+    if (value !== undefined) {
+      indexes.push(Number(value));
+    }
+  }
+  return indexes;
 }
 
 function createTerminalState(): TerminalState {
@@ -176,6 +194,69 @@ it("creates a terminal through the worker and streams output", async () => {
   expect(snapshots).toBe(snapshotsBeforeOutput);
 });
 
+it("delivers rapid small writes complete and in order through worker coalescing", async () => {
+  const cwd = mkdtempSync(join(tmpdir(), "worker-terminal-manager-coalesce-"));
+  temporaryDirs.push(cwd);
+  const burstGatePath = join(cwd, "burst-ready");
+  manager = createWorkerTerminalManager();
+  // The file gate starts the burst after this test subscribes without depending
+  // on platform-specific PTY input echo/canonical-mode behavior.
+  const session = trackTerminal(
+    await manager.createTerminal({
+      cwd,
+      env: { PASEO_TERMINAL_BURST_GATE: burstGatePath },
+      ...nodeTerminalCommand(`
+      const fs = require("node:fs");
+      const gatePath = process.env.PASEO_TERMINAL_BURST_GATE;
+      const gate = setInterval(() => {
+        if (!gatePath || !fs.existsSync(gatePath)) {
+          return;
+        }
+        clearInterval(gate);
+        for (let i = 0; i < 500; i++) {
+          process.stdout.write("[" + i + "]\\n");
+        }
+      }, 10);
+      setInterval(() => {}, 1000);
+    `),
+    }),
+  );
+
+  const events: Array<{ type: string; data?: string }> = [];
+  session.subscribe((message) => {
+    if (message.type === "output") {
+      events.push({ type: "output", data: message.data });
+    } else if (message.type === "snapshot" || message.type === "snapshotReady") {
+      events.push({ type: message.type });
+    }
+  });
+
+  // Let the snapshot subscription settle, then trigger the burst.
+  await new Promise((resolve) => setTimeout(resolve, 100));
+  writeFileSync(burstGatePath, "go");
+
+  const expected = Array.from({ length: 500 }, (_, index) => index);
+  let received: number[] = [];
+  await waitForCondition(async () => {
+    const snapshot = await manager!.getTerminalState(session.id);
+    received = snapshot ? readRenderedTokenIndexesFromState(snapshot.state) : [];
+    return expected.every((value, index) => received[index] === value);
+  }, 10000);
+
+  expect(received).toEqual(expected);
+
+  await waitForCondition(() => events.some((event) => event.type === "output"), 10000);
+
+  // No snapshot may land after output it should have preceded: every snapshot
+  // event must come before the first output event.
+  const firstOutputIndex = events.findIndex((event) => event.type === "output");
+  expect(firstOutputIndex).toBeGreaterThanOrEqual(0);
+  const snapshotAfterOutput = events
+    .slice(firstOutputIndex + 1)
+    .find((event) => event.type === "snapshot" || event.type === "snapshotReady");
+  expect(snapshotAfterOutput).toBeUndefined();
+});
+
 it("pulls fresh terminal state from the worker authority", async () => {
   const cwd = mkdtempSync(join(tmpdir(), "worker-terminal-manager-state-"));
   temporaryDirs.push(cwd);
@@ -200,6 +281,37 @@ it("pulls fresh terminal state from the worker authority", async () => {
   expect(visibleText).toContain("worker-state-ready");
 });
 
+// Windows ConPTY normalizes away the kitty keyboard escape the child writes, so it
+// never reaches the worker's input-mode tracker and the preamble stays empty. The
+// preamble-caching contract is verified on Linux/macOS; the daemon's input-mode
+// handling runs identically on every platform once the escape is observed.
+it.skipIf(isPlatform("win32"))(
+  "caches the input-mode replay preamble from the worker after getTerminalState",
+  async () => {
+    const cwd = mkdtempSync(join(tmpdir(), "worker-terminal-manager-preamble-"));
+    temporaryDirs.push(cwd);
+    manager = createWorkerTerminalManager();
+    // \x1b[>1u pushes kitty keyboard flag 1, which the worker's input-mode
+    // tracker records and reflects in its replay preamble (\x1b[=1;1u).
+    const session = trackTerminal(
+      await manager.createTerminal({
+        cwd,
+        ...nodeTerminalCommand(`
+      process.stdout.write("\\u001b[>1u");
+      setInterval(() => {}, 1000);
+    `),
+      }),
+    );
+
+    await waitForCondition(async () => {
+      const snapshot = await manager!.getTerminalState(session.id);
+      return snapshot !== null && session.getReplayPreamble() === "\x1b[=1;1u";
+    }, 10000);
+
+    expect(session.getReplayPreamble()).toBe("\x1b[=1;1u");
+  },
+);
+
 it("refreshes cached terminal title after worker title changes", async () => {
   const cwd = mkdtempSync(join(tmpdir(), "worker-terminal-manager-title-"));
   temporaryDirs.push(cwd);

packages/server/src/terminal/worker-terminal-manager.ts

@@ -2,7 +2,6 @@ import { fork } from "node:child_process";
 import { fileURLToPath } from "node:url";
 import { randomUUID } from "node:crypto";
 import type { TerminalState } from "@getpaseo/protocol/messages";
-import { TerminalInputModeTracker } from "@getpaseo/protocol/terminal-input-mode";
 import type {
   ClientMessage,
   ServerMessage,
@@ -44,7 +43,10 @@ interface PendingRequest {
 interface WorkerTerminalRecord {
   info: WorkerTerminalInfo;
   state: TerminalState;
-  inputModeTracker: TerminalInputModeTracker;
+  // Cached input-mode preamble from the worker (the authoritative tracker lives
+  // in the worker process). Refreshed on every getTerminalState response and on
+  // the snapshotReady event that precedes a live-restore replay.
+  replayPreamble: string;
   exitInfo: TerminalExitInfo | null;
   messageListeners: Set<(msg: ServerMessage) => void>;
   exitListeners: Set<(info: TerminalExitInfo) => void>;
@@ -171,7 +173,7 @@ export function createWorkerTerminalManager(
     const record: WorkerTerminalRecord = {
       info: cloneTerminalInfo(input.info),
       state: input.state,
-      inputModeTracker: new TerminalInputModeTracker(),
+      replayPreamble: "",
       exitInfo: null,
       messageListeners: new Set(),
       exitListeners: new Set(),
@@ -266,7 +268,12 @@ export function createWorkerTerminalManager(
         };
       },
       getReplayPreamble(): string {
-        return record.inputModeTracker.getPreamble();
+        // Refreshed from every getTerminalState response, which the controller fetches
+        // before every snapshot replay (legacy + visible-snapshot restore). The one
+        // gap is restore.mode === "live", which replays without fetching state — there
+        // this can be stale/empty. No client sends "live" today; revisit (ship the
+        // preamble on the worker's snapshotReady) if one ever does.
+        return record.replayPreamble;
       },
       getTitle(): string | undefined {
         return record.info.title;
@@ -333,8 +340,8 @@ export function createWorkerTerminalManager(
     if (message.message.type === "snapshot") {
       record.state = message.message.state;
     }
-    if (message.message.type === "output") {
-      record.inputModeTracker.feed(message.message.data);
+    if (message.message.type === "snapshotReady" && message.message.replayPreamble !== undefined) {
+      record.replayPreamble = message.message.replayPreamble;
     }
     for (const listener of Array.from(record.messageListeners)) {
       listener(message.message);
@@ -558,11 +565,18 @@ export function createWorkerTerminalManager(
       id: string,
       options?: { scrollbackLines?: number },
     ): Promise<TerminalStateSnapshot | null> {
-      return (await sendRequest({
+      const snapshot = (await sendRequest({
         type: "getTerminalState",
         terminalId: id,
         ...(options ? { options } : {}),
       })) as TerminalWorkerStateResult;
+      if (snapshot && snapshot.replayPreamble !== undefined) {
+        const record = recordsById.get(id);
+        if (record) {
+          record.replayPreamble = snapshot.replayPreamble;
+        }
+      }
+      return snapshot;
     },
 
     setTerminalTitle(id: string, title: string): boolean {

scripts/benchmark-terminal-latency.ts

@@ -0,0 +1,751 @@
+/**
+ * Reproducible terminal latency benchmark (Node-only, isolated daemon).
+ *
+ * Boots its OWN isolated daemon subprocess (fresh mkdtemp PASEO_HOME, random
+ * port) — it NEVER touches the developer daemon on port 6767 — and measures:
+ *   A) terminal echo latency  (single-byte input -> first echoed output frame)
+ *   B) terminal output jitter (inter-frame gaps while a command drains ~2MB)
+ *   C) RPC ping RTT at 10Hz (proxy for daemon main-loop delay)
+ * across load levels L0..L3 where load is generated by dev-only `mock` agents
+ * that stream agent updates over the same socket the benchmark client uses.
+ *
+ * Run:
+ *   npx tsx scripts/benchmark-terminal-latency.ts
+ *
+ * Optional:
+ *   BENCH_INCLUDE_L3=1   include the L3 level (L2 + a second noisy terminal)
+ *
+ * Output: pretty table to stdout + JSON to /tmp/paseo-terminal-bench/<ts>.json
+ *
+ * Requires built client/protocol dist (packages/client/dist). Build with:
+ *   npm run build:client
+ */
+
+import { spawn, execSync, type ChildProcess } from "node:child_process";
+import { mkdtemp, mkdir, rm, writeFile } from "node:fs/promises";
+import net from "node:net";
+import os from "node:os";
+import path from "node:path";
+import { performance } from "node:perf_hooks";
+import { fileURLToPath, pathToFileURL } from "node:url";
+import { WebSocket } from "ws";
+
+const __dirname = path.dirname(fileURLToPath(import.meta.url));
+const REPO_ROOT = path.resolve(__dirname, "..");
+const SERVER_DIR = path.join(REPO_ROOT, "packages", "server");
+
+// --- minimal structural view of the dist DaemonClient we depend on ----------
+
+interface TerminalStreamEvent {
+  terminalId: string;
+  type: "output" | "snapshot" | "restore";
+  data?: Uint8Array;
+}
+
+interface DaemonClientLike {
+  connect(): Promise<void>;
+  close(): Promise<void>;
+  ping(params?: { timeoutMs?: number }): Promise<{ rttMs: number }>;
+  listTerminals(cwd?: string): Promise<unknown>;
+  openProject(cwd: string): Promise<{ workspace: { id: string } | null; error: string | null }>;
+  createTerminal(
+    cwd: string,
+    name?: string,
+  ): Promise<{ terminal: { id: string } | null; error: string | null }>;
+  subscribeTerminal(terminalId: string): Promise<{ error: string | null }>;
+  killTerminal(terminalId: string): Promise<unknown>;
+  sendTerminalInput(
+    terminalId: string,
+    message: { type: "input"; data: string } | { type: "resize"; rows: number; cols: number },
+  ): void;
+  onTerminalStreamEvent(handler: (event: TerminalStreamEvent) => void): () => void;
+  createAgent(options: {
+    provider: string;
+    cwd: string;
+    model?: string;
+    title?: string;
+  }): Promise<{ id: string }>;
+  sendAgentMessage(agentId: string, text: string): Promise<void>;
+  cancelAgent?(agentId: string): Promise<unknown>;
+  deleteAgent?(agentId: string): Promise<unknown>;
+}
+
+interface DaemonClientCtor {
+  new (config: {
+    url: string;
+    clientId: string;
+    clientType: "cli";
+    appVersion?: string;
+    webSocketFactory: (url: string, options?: { headers?: Record<string, string> }) => unknown;
+  }): DaemonClientLike;
+}
+
+// --- stats helpers ----------------------------------------------------------
+
+interface Stats {
+  count: number;
+  min: number;
+  p50: number;
+  p95: number;
+  max: number;
+  avg: number;
+}
+
+function summarize(values: number[]): Stats {
+  if (values.length === 0) {
+    return { count: 0, min: 0, p50: 0, p95: 0, max: 0, avg: 0 };
+  }
+  const sorted = [...values].sort((a, b) => a - b);
+  const pick = (q: number) => sorted[Math.min(sorted.length - 1, Math.floor(sorted.length * q))];
+  return {
+    count: sorted.length,
+    min: round(sorted[0]),
+    p50: round(pick(0.5)),
+    p95: round(pick(0.95)),
+    max: round(sorted[sorted.length - 1]),
+    avg: round(values.reduce((a, b) => a + b, 0) / values.length),
+  };
+}
+
+function round(n: number): number {
+  return Math.round(n * 100) / 100;
+}
+
+function sleep(ms: number): Promise<void> {
+  return new Promise((resolve) => setTimeout(resolve, ms));
+}
+
+// --- daemon lifecycle -------------------------------------------------------
+
+async function getFreePort(): Promise<number> {
+  return new Promise((resolve, reject) => {
+    const srv = net.createServer();
+    srv.on("error", reject);
+    srv.listen(0, "127.0.0.1", () => {
+      const addr = srv.address();
+      if (addr === null || typeof addr === "string") {
+        srv.close();
+        reject(new Error("Failed to acquire a free port"));
+        return;
+      }
+      const { port } = addr;
+      srv.close(() => resolve(port));
+    });
+  });
+}
+
+async function waitForPort(port: number, child: ChildProcess, timeoutMs: number): Promise<void> {
+  const start = Date.now();
+  let lastError: unknown = null;
+  while (Date.now() - start < timeoutMs) {
+    if (child.exitCode !== null) {
+      throw new Error(`Daemon exited before listening (code ${child.exitCode})`);
+    }
+    try {
+      await new Promise<void>((resolve, reject) => {
+        const socket = net.connect(port, "127.0.0.1", () => {
+          socket.end();
+          resolve();
+        });
+        socket.setTimeout(1000, () => {
+          socket.destroy();
+          reject(new Error("connect timeout"));
+        });
+        socket.on("error", reject);
+      });
+      return;
+    } catch (error) {
+      lastError = error;
+      await sleep(150);
+    }
+  }
+  throw new Error(
+    `Daemon did not listen on 127.0.0.1:${port} within ${timeoutMs}ms (${String(lastError)})`,
+  );
+}
+
+interface BootedDaemon {
+  child: ChildProcess;
+  port: number;
+  paseoHome: string;
+  pid: number;
+}
+
+async function bootDaemon(): Promise<BootedDaemon> {
+  const port = await getFreePort();
+  if (port === 6767) {
+    throw new Error("Refusing to use port 6767 (the developer daemon)");
+  }
+  const paseoHome = await mkdtemp(path.join(os.tmpdir(), "paseo-bench-home-"));
+  const tsxBin = execSync("which tsx").toString().trim();
+
+  const child = spawn(tsxBin, ["scripts/supervisor-entrypoint.ts", "--dev"], {
+    cwd: SERVER_DIR,
+    env: {
+      ...process.env,
+      PASEO_HOME: paseoHome,
+      PASEO_SERVER_ID: "srv_terminal_bench",
+      PASEO_LISTEN: `127.0.0.1:${port}`,
+      PASEO_NODE_ENV: "development",
+      NODE_ENV: "development",
+    },
+    stdio: ["ignore", "pipe", "pipe"],
+    detached: false,
+  });
+
+  child.stdout?.on("data", (data: Buffer) => {
+    for (const line of data.toString().split("\n")) {
+      const trimmed = line.trim();
+      if (trimmed) console.log(`[daemon] ${trimmed}`);
+    }
+  });
+  child.stderr?.on("data", (data: Buffer) => {
+    for (const line of data.toString().split("\n")) {
+      const trimmed = line.trim();
+      if (trimmed) console.error(`[daemon] ${trimmed}`);
+    }
+  });
+
+  await waitForPort(port, child, 30_000);
+  return { child, port, paseoHome, pid: child.pid ?? -1 };
+}
+
+async function loadDaemonClientCtor(): Promise<DaemonClientCtor> {
+  const moduleUrl = pathToFileURL(
+    path.join(REPO_ROOT, "packages", "client", "dist", "daemon-client.js"),
+  ).href;
+  const mod = (await import(moduleUrl)) as { DaemonClient: DaemonClientCtor };
+  return mod.DaemonClient;
+}
+
+async function connectClient(ctor: DaemonClientCtor, port: number): Promise<DaemonClientLike> {
+  const client = new ctor({
+    url: `ws://127.0.0.1:${port}/ws`,
+    clientId: `bench-${Math.random().toString(36).slice(2)}`,
+    clientType: "cli",
+    webSocketFactory: (url, options) => new WebSocket(url, { headers: options?.headers }),
+  });
+  await client.connect();
+  return client;
+}
+
+// --- measurement primitives -------------------------------------------------
+
+const ECHO_SAMPLES = 100;
+const ECHO_INTERVAL_MS = 50;
+const ECHO_FRAME_TIMEOUT_MS = 2_000;
+// Printable ASCII the shell echoes back verbatim (skip control chars/space).
+const ECHO_ALPHABET = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789";
+
+interface EchoResult {
+  samples: number;
+  hits: number;
+  latencyMs: Stats;
+}
+
+/**
+ * Sends single distinct bytes spaced out in time; each byte's latency is the
+ * gap between send and the first output frame whose decoded text contains it.
+ */
+async function measureEcho(client: DaemonClientLike, terminalId: string): Promise<EchoResult> {
+  const decoder = new TextDecoder();
+  // The stream handler records the arrival time of the awaited echo byte; the
+  // sender loop polls this between sends. A plain shared slot avoids juggling a
+  // promise resolver across the handler/timeout boundary.
+  // Mutated by the async stream handler and read by the sender loop; a holder
+  // object keeps the cross-callback mutation explicit.
+  const slot: { awaitedChar: string | null; arrivalAt: number | null } = {
+    awaitedChar: null,
+    arrivalAt: null,
+  };
+
+  const unsubscribe = client.onTerminalStreamEvent((event) => {
+    if (event.terminalId !== terminalId || event.type !== "output" || !event.data) {
+      return;
+    }
+    if (slot.awaitedChar === null || slot.arrivalAt !== null) {
+      return;
+    }
+    if (decoder.decode(event.data).includes(slot.awaitedChar)) {
+      slot.arrivalAt = performance.now();
+    }
+  });
+
+  const latencies: number[] = [];
+  let hits = 0;
+  try {
+    for (let i = 0; i < ECHO_SAMPLES; i += 1) {
+      const char = ECHO_ALPHABET[i % ECHO_ALPHABET.length];
+      slot.awaitedChar = char;
+      slot.arrivalAt = null;
+      const sentAt = performance.now();
+      client.sendTerminalInput(terminalId, { type: "input", data: char });
+
+      const deadline = sentAt + ECHO_FRAME_TIMEOUT_MS;
+      while (slot.arrivalAt === null && performance.now() < deadline) {
+        await sleep(1);
+      }
+      if (slot.arrivalAt !== null) {
+        latencies.push(slot.arrivalAt - sentAt);
+        hits += 1;
+      }
+      slot.awaitedChar = null;
+      await sleep(ECHO_INTERVAL_MS);
+    }
+  } finally {
+    unsubscribe();
+  }
+
+  return { samples: ECHO_SAMPLES, hits, latencyMs: summarize(latencies) };
+}
+
+const BURST_TARGET_BYTES = 2_000_000;
+const BURST_CHUNK = 16_384;
+const BURST_DRAIN_TIMEOUT_MS = 20_000;
+const BURST_QUIET_MS = 400;
+
+interface BurstResult {
+  bytesSeen: number;
+  outputFrames: number;
+  snapshotFrames: number;
+  restoreFrames: number;
+  drainMs: number;
+  interFrameGapMs: Stats;
+}
+
+/**
+ * Runs a node one-liner in the terminal that prints ~2MB rapidly, then records
+ * inter-output-frame gaps until the stream goes quiet. Snapshot/restore frames
+ * during the burst indicate the snapshot-churn path firing.
+ */
+async function measureBurst(client: DaemonClientLike, terminalId: string): Promise<BurstResult> {
+  let bytesSeen = 0;
+  let outputFrames = 0;
+  let snapshotFrames = 0;
+  let restoreFrames = 0;
+  const gaps: number[] = [];
+  let lastBurstFrameAt = 0;
+  let firstBurstFrameAt = 0;
+  let lastActivityAt = performance.now();
+
+  // Frames smaller than this are the echoed command / prompt noise, not burst
+  // payload — gap timing only starts once real drain data is flowing, so the
+  // node startup latency doesn't pollute the inter-frame jitter.
+  const BURST_FRAME_MIN_BYTES = 1_024;
+
+  const unsubscribe = client.onTerminalStreamEvent((event) => {
+    if (event.terminalId !== terminalId) {
+      return;
+    }
+    const now = performance.now();
+    lastActivityAt = now;
+    if (event.type === "snapshot") {
+      snapshotFrames += 1;
+      return;
+    }
+    if (event.type === "restore") {
+      restoreFrames += 1;
+      return;
+    }
+    if (event.type !== "output" || !event.data) {
+      return;
+    }
+    outputFrames += 1;
+    bytesSeen += event.data.byteLength;
+    if (event.data.byteLength < BURST_FRAME_MIN_BYTES && firstBurstFrameAt === 0) {
+      return;
+    }
+    if (firstBurstFrameAt === 0) {
+      firstBurstFrameAt = now;
+    } else {
+      gaps.push(now - lastBurstFrameAt);
+    }
+    lastBurstFrameAt = now;
+  });
+
+  // Clear whatever the echo measurement left on the shell input line (Ctrl-C),
+  // then let the fresh prompt settle before sending the burst command.
+  client.sendTerminalInput(terminalId, { type: "input", data: "" });
+  await sleep(300);
+
+  const startedAt = performance.now();
+  try {
+    const chunks = Math.ceil(BURST_TARGET_BYTES / BURST_CHUNK);
+    // Print BURST_TARGET_BYTES of 'x' in chunks, no trailing newline spam.
+    const oneLiner =
+      `const c='x'.repeat(${BURST_CHUNK});` +
+      `for(let i=0;i<${chunks};i++)process.stdout.write(c);` +
+      `process.stdout.write('\\nBURST_DONE\\n')`;
+    client.sendTerminalInput(terminalId, {
+      type: "input",
+      data: `node -e "${oneLiner}"\r`,
+    });
+
+    const deadline = performance.now() + BURST_DRAIN_TIMEOUT_MS;
+    while (performance.now() < deadline) {
+      await sleep(50);
+      const quietFor = performance.now() - lastActivityAt;
+      // The daemon coalesces a large pty write into a snapshot plus a handful of
+      // frames, so the client rarely sees the full BURST_TARGET_BYTES. Exit once
+      // real burst data has arrived and the stream has gone quiet.
+      if (firstBurstFrameAt > 0 && quietFor >= BURST_QUIET_MS) {
+        break;
+      }
+    }
+  } finally {
+    unsubscribe();
+  }
+
+  return {
+    bytesSeen,
+    outputFrames,
+    snapshotFrames,
+    restoreFrames,
+    drainMs: round((lastBurstFrameAt || startedAt) - startedAt),
+    interFrameGapMs: summarize(gaps),
+  };
+}
+
+const PING_INTERVAL_MS = 100;
+
+/**
+ * Samples RPC RTT at 10Hz for `durationMs`. Uses `ping` if available, else a
+ * cheap listTerminals request. Returns RTT stats — a proxy for main-loop delay.
+ */
+async function measurePing(client: DaemonClientLike, durationMs: number): Promise<Stats> {
+  const rtts: number[] = [];
+  const deadline = Date.now() + durationMs;
+  while (Date.now() < deadline) {
+    const start = performance.now();
+    try {
+      const result = await client.ping({ timeoutMs: 5_000 });
+      rtts.push(typeof result?.rttMs === "number" ? result.rttMs : performance.now() - start);
+    } catch {
+      const fallbackStart = performance.now();
+      await client.listTerminals().catch(() => undefined);
+      rtts.push(performance.now() - fallbackStart);
+    }
+    await sleep(PING_INTERVAL_MS);
+  }
+  return summarize(rtts);
+}
+
+// --- load generation --------------------------------------------------------
+
+interface LoadController {
+  agentIds: string[];
+  noisyTerminalId: string | null;
+  stop(): Promise<void>;
+}
+
+interface LevelSpec {
+  id: string;
+  description: string;
+  mockAgents: number;
+  bigDiffEvery2s: boolean;
+  secondNoisyTerminal: boolean;
+}
+
+const LEVELS: LevelSpec[] = [
+  {
+    id: "L0",
+    description: "idle",
+    mockAgents: 0,
+    bigDiffEvery2s: false,
+    secondNoisyTerminal: false,
+  },
+  {
+    id: "L1",
+    description: "2 mock agents streaming",
+    mockAgents: 2,
+    bigDiffEvery2s: false,
+    secondNoisyTerminal: false,
+  },
+  {
+    id: "L2",
+    description: "4 mock agents + 200KB diff bursts every 2s",
+    mockAgents: 4,
+    bigDiffEvery2s: true,
+    secondNoisyTerminal: false,
+  },
+];
+
+if (process.env.BENCH_INCLUDE_L3 === "1") {
+  LEVELS.push({
+    id: "L3",
+    description: "L2 + a second terminal continuously emitting",
+    mockAgents: 4,
+    bigDiffEvery2s: true,
+    secondNoisyTerminal: true,
+  });
+}
+
+async function startLoad(
+  client: DaemonClientLike,
+  cwd: string,
+  spec: LevelSpec,
+): Promise<LoadController> {
+  const agentIds: string[] = [];
+  for (let i = 0; i < spec.mockAgents; i += 1) {
+    const agent = await client.createAgent({
+      provider: "mock",
+      cwd,
+      model: "five-minute-stream",
+      title: `${spec.id}-load-${i}`,
+    });
+    agentIds.push(agent.id);
+    // Kick off the sustained stream.
+    await client.sendAgentMessage(agent.id, "start streaming").catch(() => undefined);
+  }
+
+  let bigDiffTimer: ReturnType<typeof setInterval> | null = null;
+  if (spec.bigDiffEvery2s && agentIds.length > 0) {
+    bigDiffTimer = setInterval(() => {
+      const target = agentIds[0];
+      void client
+        .sendAgentMessage(target, "emit 200000 byte large diff agent stream update")
+        .catch(() => undefined);
+    }, 2_000);
+    bigDiffTimer.unref?.();
+  }
+
+  let noisyTerminalId: string | null = null;
+  if (spec.secondNoisyTerminal) {
+    const created = await client.createTerminal(cwd, `${spec.id}-noisy`);
+    if (created.terminal) {
+      noisyTerminalId = created.terminal.id;
+      await client.subscribeTerminal(noisyTerminalId);
+      // Continuous infinite emit loop in the shell.
+      client.sendTerminalInput(noisyTerminalId, {
+        type: "input",
+        data: `node -e "setInterval(()=>process.stdout.write('y'.repeat(4096)),20)"\r`,
+      });
+    }
+  }
+
+  // Let the streams warm up before we start measuring.
+  await sleep(500);
+
+  return {
+    agentIds,
+    noisyTerminalId,
+    async stop() {
+      if (bigDiffTimer) {
+        clearInterval(bigDiffTimer);
+      }
+      if (noisyTerminalId) {
+        // Ctrl-C then kill the terminal.
+        client.sendTerminalInput(noisyTerminalId, { type: "input", data: "" });
+        await client.killTerminal(noisyTerminalId).catch(() => undefined);
+      }
+      for (const agentId of agentIds) {
+        if (client.cancelAgent) {
+          await client.cancelAgent(agentId).catch(() => undefined);
+        }
+        if (client.deleteAgent) {
+          await client.deleteAgent(agentId).catch(() => undefined);
+        }
+      }
+    },
+  };
+}
+
+// --- per-level run ----------------------------------------------------------
+
+interface LevelResult {
+  id: string;
+  description: string;
+  echo: EchoResult;
+  burst: BurstResult;
+  pingMs: Stats;
+}
+
+async function runLevel(
+  client: DaemonClientLike,
+  cwd: string,
+  spec: LevelSpec,
+  measuredTerminalId: string,
+): Promise<LevelResult> {
+  console.log(`\n=== ${spec.id} (${spec.description}) ===`);
+  const load = await startLoad(client, cwd, spec);
+  try {
+    // Ping sampling runs concurrently with the echo + burst measurements so it
+    // observes the daemon under the same load the latency primitives see.
+    let pingMs: Stats = summarize([]);
+    const pingDone = (async () => {
+      pingMs = await measurePing(client, 14_000);
+    })();
+
+    console.log(`  measuring echo latency (${ECHO_SAMPLES} samples)...`);
+    const echo = await measureEcho(client, measuredTerminalId);
+    console.log(
+      `    echo p50=${echo.latencyMs.p50}ms p95=${echo.latencyMs.p95}ms max=${echo.latencyMs.max}ms (${echo.hits}/${echo.samples} hits)`,
+    );
+
+    console.log("  measuring output burst/jitter (~2MB)...");
+    const burst = await measureBurst(client, measuredTerminalId);
+    console.log(
+      `    burst drain=${burst.drainMs}ms frames=${burst.outputFrames} gap_p95=${burst.interFrameGapMs.p95}ms snap=${burst.snapshotFrames} restore=${burst.restoreFrames}`,
+    );
+
+    await pingDone;
+    console.log(`    ping p50=${pingMs.p50}ms p95=${pingMs.p95}ms max=${pingMs.max}ms`);
+
+    return { id: spec.id, description: spec.description, echo, burst, pingMs };
+  } finally {
+    await load.stop();
+  }
+}
+
+// --- output -----------------------------------------------------------------
+
+function pad(value: string | number, width: number): string {
+  return String(value).padStart(width);
+}
+
+function printTable(results: LevelResult[]): void {
+  console.log("\n========== TERMINAL LATENCY BENCHMARK ==========\n");
+  const header = [
+    pad("level", 6),
+    pad("echo_p50", 9),
+    pad("echo_p95", 9),
+    pad("echo_max", 9),
+    pad("ping_p50", 9),
+    pad("ping_p95", 9),
+    pad("gap_p95", 8),
+    pad("gap_max", 8),
+    pad("drain", 8),
+    pad("frames", 7),
+    pad("snap", 5),
+    pad("restore", 8),
+  ].join(" ");
+  console.log(header);
+  console.log("-".repeat(header.length));
+  for (const r of results) {
+    console.log(
+      [
+        pad(r.id, 6),
+        pad(r.echo.latencyMs.p50, 9),
+        pad(r.echo.latencyMs.p95, 9),
+        pad(r.echo.latencyMs.max, 9),
+        pad(r.pingMs.p50, 9),
+        pad(r.pingMs.p95, 9),
+        pad(r.burst.interFrameGapMs.p95, 8),
+        pad(r.burst.interFrameGapMs.max, 8),
+        pad(r.burst.drainMs, 8),
+        pad(r.burst.outputFrames, 7),
+        pad(r.burst.snapshotFrames, 5),
+        pad(r.burst.restoreFrames, 8),
+      ].join(" "),
+    );
+  }
+  console.log("");
+  for (const r of results) {
+    console.log(`  ${r.id}: ${r.description}`);
+  }
+  console.log("");
+}
+
+function getCommitHash(): string {
+  try {
+    return execSync("git rev-parse --short HEAD", { cwd: REPO_ROOT }).toString().trim();
+  } catch {
+    return "unknown";
+  }
+}
+
+// --- main -------------------------------------------------------------------
+
+async function main(): Promise<void> {
+  const commit = getCommitHash();
+  console.log(`Terminal latency benchmark — commit ${commit}`);
+  console.log("Booting isolated daemon (random port, fresh PASEO_HOME)...");
+
+  let daemon: BootedDaemon | null = null;
+  let client: DaemonClientLike | null = null;
+  let workspaceDir: string | null = null;
+  let measuredTerminalId: string | null = null;
+
+  try {
+    daemon = await bootDaemon();
+    console.log(`Daemon ready: pid=${daemon.pid} port=${daemon.port} home=${daemon.paseoHome}`);
+
+    const ctor = await loadDaemonClientCtor();
+    client = await connectClient(ctor, daemon.port);
+
+    workspaceDir = await mkdtemp(path.join(os.tmpdir(), "paseo-bench-ws-"));
+    const opened = await client.openProject(workspaceDir);
+    if (!opened.workspace) {
+      throw new Error(`Failed to open project: ${opened.error}`);
+    }
+
+    const created = await client.createTerminal(workspaceDir, "measured");
+    if (!created.terminal) {
+      throw new Error(`Failed to create terminal: ${created.error}`);
+    }
+    measuredTerminalId = created.terminal.id;
+    await client.subscribeTerminal(measuredTerminalId);
+
+    // Let the shell print its prompt and settle before measuring.
+    await sleep(750);
+
+    const results: LevelResult[] = [];
+    for (const spec of LEVELS) {
+      const result = await runLevel(client, workspaceDir, spec, measuredTerminalId);
+      results.push(result);
+      // Settle between levels so a previous level's load fully drains.
+      await sleep(2_000);
+    }
+
+    printTable(results);
+
+    const outDir = "/tmp/paseo-terminal-bench";
+    await mkdir(outDir, { recursive: true });
+    const timestamp = new Date().toISOString().replace(/[:.]/g, "-");
+    const outPath = path.join(outDir, `${timestamp}.json`);
+    await writeFile(
+      outPath,
+      JSON.stringify(
+        {
+          commit,
+          daemonPid: daemon.pid,
+          port: daemon.port,
+          paseoHome: daemon.paseoHome,
+          node: process.version,
+          platform: `${os.platform()} ${os.arch()}`,
+          createdAt: new Date().toISOString(),
+          levels: results,
+        },
+        null,
+        2,
+      ),
+    );
+    console.log(`Wrote results to ${outPath}`);
+  } finally {
+    if (client && measuredTerminalId) {
+      await client.killTerminal(measuredTerminalId).catch(() => undefined);
+    }
+    if (client) {
+      await client.close().catch(() => undefined);
+    }
+    if (daemon) {
+      daemon.child.kill("SIGTERM");
+      await Promise.race([
+        new Promise<void>((resolve) => daemon!.child.once("exit", () => resolve())),
+        sleep(5_000),
+      ]);
+      if (daemon.child.exitCode === null && daemon.child.signalCode === null) {
+        daemon.child.kill("SIGKILL");
+      }
+      await rm(daemon.paseoHome, { recursive: true, force: true }).catch(() => undefined);
+    }
+    if (workspaceDir) {
+      await rm(workspaceDir, { recursive: true, force: true }).catch(() => undefined);
+    }
+  }
+}
+
+main().catch((error) => {
+  console.error(error);
+  process.exit(1);
+});