feat(metrics): consume contextSize — current context-usage readout

Backend context-size handoff: re-pin [email protected] / [email protected]
(+ re-mirror .dispatch reference snapshots). Thread the optional contextSize
through core/metrics (done fold + durable + selectCurrentContextSize: latest
turn's defined value, undefined=>unknown never 0, durable-wins-over-live).
Chat store exposes currentContextSize; ContextSizeBadge renders
"N tokens in context" / "context size unknown" above the composer.
GLOSSARY: add context size / context window. 533 tests green.

14 files changed, 241 insertions, 15 deletions

diff --git a/.dispatch/transport-contract.reference.md b/.dispatch/transport-contract.reference.md
index c2e2076..40ced1e 100644
--- a/.dispatch/transport-contract.reference.md
+++ b/.dispatch/transport-contract.reference.md
@@ -5,9 +5,16 @@
 > hangs on a permission prompt). Your CODE still imports `@dispatch/transport-contract` normally —
 > this file is for READING only.
 >
-> **Orchestrator:** SNAPSHOT of `[email protected]` (committed, backend `6db12ff`; the metrics
-> endpoint shipped + version-bumped + LIVE-VERIFIED). Depends on `@dispatch/[email protected]` (see
-> `wire.reference.md`) + `@dispatch/ui-contract` (see `ui-contract.reference.md`).
+> **Orchestrator:** SNAPSHOT of `[email protected]` (the metrics endpoint shipped +
+> version-bumped + LIVE-VERIFIED). Depends on `@dispatch/[email protected]` (see `wire.reference.md`) +
+> `@dispatch/[email protected]` (see `ui-contract.reference.md`).
+>
+> **2026-06-12 delta (context-size handoff — package bumped `0.5.0` → `0.6.0`, depends on
+> `[email protected]`):** no NEW transport shape — the optional `contextSize?: number` rides the
+> re-exported `TurnMetrics` (so `ConversationMetricsResponse.turns[].contextSize`) and, live, the
+> `TurnDoneEvent.contextSize` on the `done` AgentEvent (`chat.delta` WS / `/chat` NDJSON). On
+> (re)hydrate take the LAST `turns[]` element with a defined `contextSize`; live, update on `done`.
+> See the `wire.reference.md` context-size delta for the definition.
 >
 > **2026-06 delta (cache-warming handoff, additive — package still `0.4.0`):** adds
 > `POST /chat/warm` (`WarmRequest` → `WarmResponse`) for an on-demand prompt-cache warm, and the
diff --git a/.dispatch/wire.reference.md b/.dispatch/wire.reference.md
index ee5488c..cf1410a 100644
--- a/.dispatch/wire.reference.md
+++ b/.dispatch/wire.reference.md
@@ -4,8 +4,15 @@
 > types WITHOUT following the `file:` dep symlink out of this repo (which hangs on a permission
 > prompt). Your CODE still imports `@dispatch/wire` normally — this file is for READING only.
 >
-> **Orchestrator:** SNAPSHOT of `[email protected]` (committed, backend `6db12ff`; the metrics types below
-> shipped + version-bumped). Regenerate whenever `@dispatch/wire` changes.
+> **Orchestrator:** SNAPSHOT of `[email protected]` (the metrics types below shipped + version-bumped).
+> Regenerate whenever `@dispatch/wire` changes.
+>
+> **2026-06-12 delta (context-size handoff — package bumped `0.4.0` → `0.5.0`):** adds an OPTIONAL
+> `contextSize?: number` to BOTH `TurnDoneEvent` (live `done`) and `TurnMetrics` (persisted) — the
+> turn's FINAL step `inputTokens + outputTokens` (current context occupancy), NOT the aggregate
+> `usage` (which overcounts multi-step turns). The two carriers are equal for the same turn. Current
+> value = the LATEST turn's `contextSize`; `undefined` ⇒ render "unknown", never `0`. See the field
+> doc-comments on `TurnMetrics`/`TurnDoneEvent` below.
 >
 > **0.3.0 changes (token + timing metrics):**
 > - **Live per-step/per-turn telemetry on the event stream** (transient — NOT persisted):
@@ -221,6 +228,16 @@ export interface TurnMetrics {
 	readonly durationMs?: number;
 	/** Per-step metrics in step order. */
 	readonly steps: readonly StepMetrics[];
+	/**
+	 * **Context size** — tokens the conversation occupies as of this turn: the
+	 * turn's FINAL step `inputTokens + outputTokens` (the last entry of `steps`),
+	 * NOT the aggregate `usage` (which sums per-step prompts and overcounts a
+	 * multi-step turn). The persisted, replayable counterpart of
+	 * `TurnDoneEvent.contextSize` and equal to it for the same turn. A client
+	 * reopening a past conversation reads the LAST turn's `contextSize` as the
+	 * current context usage. Optional: absent when no per-step usage was available.
+	 */
+	readonly contextSize?: number;
 }
 
 // ─── Outward events ─────────────────────────────────────────────────────────
@@ -393,6 +410,21 @@ export interface TurnDoneEvent {
 	 * provider reported no usage).
 	 */
 	readonly usage?: Usage;
+	/**
+	 * **Context size** — tokens the conversation occupies right now: the turn's
+	 * FINAL step `inputTokens + outputTokens` (the prompt sent into the last LLM
+	 * round-trip plus that round-trip's output). This is the "tokens in context"
+	 * figure a client renders as the chat's current context usage, and a client
+	 * treats the LATEST turn's value as the live total.
+	 *
+	 * Deliberately NOT the aggregate `usage` above: `usage` SUMS each step's
+	 * `inputTokens`, which overcounts a multi-step / tool-calling turn because every
+	 * step re-prefills the growing prompt — the final step's input already includes
+	 * all prior context, so its input+output is the true occupancy. Optional: absent
+	 * when no per-step usage was observed this turn (mirrors `usage`). A later field
+	 * will carry the model's max context-window LIMIT; this is only the current size.
+	 */
+	readonly contextSize?: number;
 }
 
 /**
diff --git a/GLOSSARY.md b/GLOSSARY.md
index 538ba7e..d632c8d 100644
--- a/GLOSSARY.md
+++ b/GLOSSARY.md
@@ -19,6 +19,8 @@
 | **step metrics** | The durable per-step metrics within a `TurnMetrics`: the step's `Usage` (tokens) + `ttftMs`/`decodeMs`/`genTotalMs` timing, keyed by `stepId` (`StepMetrics`). The persisted counterpart of the live `usage` + `step-complete` events. | step stats |
 | **TTFT** (time to first token) | Per-step latency: generation stream start → first content token (text or reasoning). One per step (each step re-prefills). On the wire as `step-complete.ttftMs` / `StepMetrics.ttftMs` (optional). | time-to-first-byte |
 | **decode time** | Per-step generation time after the first token (first token → stream end = `genTotalMs − ttftMs`). On the wire as `step-complete.decodeMs` / `StepMetrics.decodeMs` (optional). | — |
+| **context size** | The tokens a conversation currently occupies: the most recent turn's FINAL step `inputTokens + outputTokens` (NOT the aggregate per-turn `usage`, which sums per-step prompts and overcounts a multi-step turn). On the wire as `TurnDoneEvent.contextSize` (live `done`) + `TurnMetrics.contextSize` (persisted); the FE reads the LATEST turn's value as current usage, and treats `undefined` as "unknown" (renders a placeholder, never `0`). Mirrors the backend GLOSSARY. | context usage, context length, tokens used (and do NOT call it "context window" — that's the limit) |
+| **context window** | The model's MAXIMUM token capacity (the limit a **context size** is measured against). A FUTURE backend field — not on the wire yet; the FE shows context size alone (no `size / limit` denominator) until it ships. | max context, token limit (distinct from **context size**, the current usage) |
 
 ## Frontend-specific
 | Term | Meaning | Aliases to avoid |
diff --git a/backend-handoff.md b/backend-handoff.md
index 99c2964..e9b128a 100644
--- a/backend-handoff.md
+++ b/backend-handoff.md
@@ -5,24 +5,33 @@
 > **From:** dispatch-web orchestrator · **To:** arch-rewrite orchestrator · **Courier:** the user.
 > `lsp` does NOT span the repos (ORCHESTRATOR §5) — every cross-repo ask flows through here.
 
-_Last updated: 2026-06-11. **FE is current on `[email protected]`.** All handoffs to date are
-consumed: surfaces + WS, conversation transcript/metrics, tabs + model selector, cache-warming (incl.
-authoritative timer + retention + cache-rate fix), and **per-conversation cwd + LSP status** (new
-`workspace` feature — cwd field in the Model view + a "Language Servers" view; works for drafts too).
+_Last updated: 2026-06-12. **FE is current on `[email protected]` / `[email protected]`.** All handoffs
+to date are consumed: surfaces + WS, conversation transcript/metrics, tabs + model selector,
+cache-warming (incl. authoritative timer + retention + cache-rate fix), **per-conversation cwd + LSP
+status**, and **context size** (the `contextSize` field — `done` live + `TurnMetrics` persisted —
+rendered as a current-usage readout above the composer).
 **Open asks:** CR-1 (Loaded Extensions as a real table) + CR-2 (optional catalog `scope` flag) below.
 The cwd/LSP draft-path verification (`backend-handoff-cwd-lsp.md`) came back **all ✅ confirmed** by the
 backend (answers in their `frontend-lsp-cwd-handoff.md`) — see §2._
 
+**Context-size handoff (`frontend-context-size-handoff.md`) → CONSUMED ✅.** Re-pinned `[email protected]→0.5.0`
++ `[email protected]→0.6.0`; re-mirrored both `.dispatch/*.reference.md`; added "context size" +
+"context window" to FE `GLOSSARY.md`. `core/metrics` now threads `contextSize` through the `done` fold +
+durable metrics and exposes `selectCurrentContextSize` (LATEST turn's defined value, `undefined`⇒unknown,
+never `0`, durable-wins-over-live); the chat store exposes `currentContextSize`; `ContextSizeBadge`
+renders "N tokens in context" / "context size unknown" above the composer. 533 tests green. NO new
+backend ask — but the max-limit denominator is now a live FE need; see §3.
+
 ---
 
 ## 1. Pinned backend contracts (consumed by the FE)
 
-Pinned as `file:` deps: **`[email protected]`; `[email protected]`; `[email protected]`**.
+Pinned as `file:` deps: **`[email protected]`; `[email protected]`; `[email protected]`**.
 
 | Package | Used for |
 |---|---|
 | `@dispatch/ui-contract` | surfaces + surface WS protocol |
-| `@dispatch/wire` | `Chunk`/`StoredChunk`(+`seq`)/`ChatMessage`/`AgentEvent`/`TurnSealedEvent`/`Usage`/`StepId` + metrics: `StepMetrics`/`TurnMetrics`, `usage.stepId`, `step-complete`, `done.durationMs`/`done.usage`, `tool-result.durationMs` |
+| `@dispatch/wire` | `Chunk`/`StoredChunk`(+`seq`)/`ChatMessage`/`AgentEvent`/`TurnSealedEvent`/`Usage`/`StepId` + metrics: `StepMetrics`/`TurnMetrics`, `usage.stepId`, `step-complete`, `done.durationMs`/`done.usage`, `tool-result.durationMs`, **`done.contextSize`/`TurnMetrics.contextSize`** |
 | `@dispatch/transport-contract` | `ChatRequest`/`ModelsResponse`/`ConversationHistoryResponse`/`ConversationMetricsResponse` + `WarmRequest`/`WarmResponse` + `CwdResponse`/`SetCwdRequest` + LSP (`LspStatusResponse`/`LspServerInfo`/`LspServerState`) + WS chat ops + `WsClientMessage`/`WsServerMessage` |
 
 Endpoints in use (HTTP **24203**, WS **24205**, CORS `*` incl. `PUT`):
@@ -31,7 +40,7 @@ Endpoints in use (HTTP **24203**, WS **24205**, CORS `*` incl. `PUT`):
 `GET /conversations/:id/lsp` · `POST /chat/warm` · WS `chat.send`→`chat.delta`.
 
 Mirrored in-repo for headless agents: `.dispatch/{ui-contract,wire,transport-contract}.reference.md`
-(regenerate on any contract bump; all current as of `[email protected]`).
+(regenerate on any contract bump; all current as of `[email protected]` / `[email protected]`).
 
 ## 2. Open asks FOR THE BACKEND
 
@@ -101,6 +110,12 @@ harden `/chat` to treat blank as "not provided" if we ever want it — not neede
 
 ## 3. Likely NEXT backend asks (heads-up, not yet requested)
 
+- **Model max context-window LIMIT** (the denominator for context size) — the context-size handoff
+  flagged this as the separate, later field. The FE now shows current size alone (e.g. "34,102 tokens
+  in context"); once a per-model/per-turn `contextWindow` (max token capacity) ships, the FE can render
+  `contextSize / limit` (e.g. "34,102 / 200,000") + a usage bar. GLOSSARY term reserved: "context window"
+  = the limit (distinct from "context size" = current usage). **Likely the next ask** — raise when the
+  backend can source the model's advertised window.
 - `GET /conversations` — conversation list / sidebar (history explorer / switcher); could also expose a
   per-conversation "last model" so a reopened tab seeds its model from the server instead of localStorage.
 - `POST /conversations/:id/cancel` — "stop generating".
diff --git a/src/app/App.svelte b/src/app/App.svelte
index daab953..32db54f 100644
--- a/src/app/App.svelte
+++ b/src/app/App.svelte
@@ -6,7 +6,13 @@
 		manifest as cacheWarmingManifest,
 		type WarmFeedback,
 	} from "../features/cache-warming";
-	import { ChatView, Composer, manifest as chatManifest, ModelSelector } from "../features/chat";
+	import {
+		ChatView,
+		Composer,
+		manifest as chatManifest,
+		ContextSizeBadge,
+		ModelSelector,
+	} from "../features/chat";
 	import { manifest as conversationCacheManifest } from "../features/conversation-cache";
 	import { manifest as markdownManifest } from "../features/markdown";
 	import {
@@ -211,6 +217,7 @@
 			<ScrollToBottom show={smartScroll.showButton} onResume={() => smartScroll.resume()} />
 		</div>
 
+		<ContextSizeBadge contextSize={store.activeChat.currentContextSize} />
 		<Composer onSend={handleSend} />
 	</div>
 
diff --git a/src/core/metrics/format.test.ts b/src/core/metrics/format.test.ts
index 3eec93d..7c143d7 100644
--- a/src/core/metrics/format.test.ts
+++ b/src/core/metrics/format.test.ts
@@ -4,6 +4,7 @@ import {
 	computeCachePct,
 	computeExpectedCachePct,
 	computeTps,
+	formatContextSize,
 	viewCacheRate,
 	viewExpectedCache,
 	viewStepMetrics,
@@ -308,3 +309,17 @@ describe("viewExpectedCache", () => {
 		expect(v?.isHit).toBe(true);
 	});
 });
+
+describe("formatContextSize", () => {
+	it("formats a defined count with thousands separators", () => {
+		expect(formatContextSize(34102)).toBe("34,102 tokens in context");
+	});
+
+	it("renders a placeholder for undefined (never 0)", () => {
+		expect(formatContextSize(undefined)).toBe("context size unknown");
+	});
+
+	it("renders an explicit 0 as zero tokens (a real reported value)", () => {
+		expect(formatContextSize(0)).toBe("0 tokens in context");
+	});
+});
diff --git a/src/core/metrics/format.ts b/src/core/metrics/format.ts
index ee8db60..d8dd2cc 100644
--- a/src/core/metrics/format.ts
+++ b/src/core/metrics/format.ts
@@ -17,6 +17,17 @@ function formatTps(tps: number | null): string | null {
 	return `${Math.round(tps)} tok/s`;
 }
 
+/**
+ * Format the current context size for display. A defined count renders as
+ * `"<n> tokens in context"` (thousands-separated); `undefined` ("unknown" — no
+ * per-step usage reported yet) renders the placeholder `"context size unknown"`.
+ * Never renders `0` for the unknown case.
+ */
+export function formatContextSize(n: number | undefined): string {
+	if (n === undefined) return "context size unknown";
+	return `${formatTokens(n)} tokens in context`;
+}
+
 /** Compute tokens-per-second. Returns null when elapsed time is absent or zero. */
 export function computeTps(outputTokens: number, elapsedMs: number | undefined): number | null {
 	if (elapsedMs === undefined || elapsedMs <= 0) return null;
diff --git a/src/core/metrics/index.ts b/src/core/metrics/index.ts
index 8822159..773d697 100644
--- a/src/core/metrics/index.ts
+++ b/src/core/metrics/index.ts
@@ -2,6 +2,7 @@ export {
 	computeCachePct,
 	computeExpectedCachePct,
 	computeTps,
+	formatContextSize,
 	viewCacheRate,
 	viewExpectedCache,
 	viewStepMetrics,
@@ -12,6 +13,7 @@ export {
 	applyDurableMetrics,
 	foldMetricsEvent,
 	initialMetricsState,
+	selectCurrentContextSize,
 	selectOrderedTurnMetrics,
 } from "./reducer";
 export type {
diff --git a/src/core/metrics/reducer.test.ts b/src/core/metrics/reducer.test.ts
index 16c88b3..cd9f673 100644
--- a/src/core/metrics/reducer.test.ts
+++ b/src/core/metrics/reducer.test.ts
@@ -4,6 +4,7 @@ import {
 	applyDurableMetrics,
 	foldMetricsEvent,
 	initialMetricsState,
+	selectCurrentContextSize,
 	selectOrderedTurnMetrics,
 } from "./reducer";
 
@@ -39,7 +40,11 @@ const stepCompleteEvent = (
 
 const doneEvent = (
 	turnId: string,
-	extra: { durationMs?: number; usage?: { inputTokens: number; outputTokens: number } } = {},
+	extra: {
+		durationMs?: number;
+		usage?: { inputTokens: number; outputTokens: number };
+		contextSize?: number;
+	} = {},
 ): TurnDoneEvent => ({
 	type: "done",
 	conversationId: "c1",
@@ -366,3 +371,72 @@ describe("applyDurableMetrics", () => {
 		expect(s.durable.get("t1")?.usage.inputTokens).toBe(99);
 	});
 });
+
+describe("contextSize / selectCurrentContextSize", () => {
+	it("live done carries contextSize onto the turn total", () => {
+		let s = initialMetricsState();
+		s = foldMetricsEvent(s, usageEvent("t1", 100, 50, "s1"));
+		s = foldMetricsEvent(s, stepCompleteEvent("t1", "s1"));
+		s = foldMetricsEvent(s, doneEvent("t1", { contextSize: 1234 }));
+
+		const ordered = selectOrderedTurnMetrics(s);
+		expect(ordered[0]?.total?.contextSize).toBe(1234);
+		expect(selectCurrentContextSize(s)).toBe(1234);
+	});
+
+	it("contextSize is NOT the aggregate usage sum (multi-step turn)", () => {
+		let s = initialMetricsState();
+		// Two steps: usage sums to 300 in / 130 out = 430, but contextSize is the
+		// backend-stamped final-step occupancy, independent of the sum.
+		s = foldMetricsEvent(s, usageEvent("t1", 100, 50, "s1"));
+		s = foldMetricsEvent(s, stepCompleteEvent("t1", "s1"));
+		s = foldMetricsEvent(s, usageEvent("t1", 200, 80, "s2"));
+		s = foldMetricsEvent(s, stepCompleteEvent("t1", "s2"));
+		s = foldMetricsEvent(s, doneEvent("t1", { contextSize: 250 }));
+
+		const ordered = selectOrderedTurnMetrics(s);
+		expect(ordered[0]?.total?.usage).toEqual({ inputTokens: 300, outputTokens: 130 });
+		expect(ordered[0]?.total?.contextSize).toBe(250);
+		expect(selectCurrentContextSize(s)).toBe(250);
+	});
+
+	it("persisted (durable) contextSize is preserved and selected", () => {
+		let s = initialMetricsState();
+		s = applyDurableMetrics(s, [
+			{ turnId: "t1", usage: { inputTokens: 10, outputTokens: 5 }, steps: [], contextSize: 4096 },
+		]);
+		expect(s.durable.get("t1")?.contextSize).toBe(4096);
+		expect(selectCurrentContextSize(s)).toBe(4096);
+	});
+
+	it("selectCurrentContextSize returns the LATEST turn's value", () => {
+		let s = initialMetricsState();
+		s = foldMetricsEvent(s, doneEvent("t1", { contextSize: 100 }));
+		s = foldMetricsEvent(s, doneEvent("t2", { contextSize: 900 }));
+		expect(selectCurrentContextSize(s)).toBe(900);
+	});
+
+	it("selectCurrentContextSize skips a later turn that lacks contextSize", () => {
+		let s = initialMetricsState();
+		s = foldMetricsEvent(s, doneEvent("t1", { contextSize: 700 }));
+		// t2 finishes but the provider reported no per-step usage → no contextSize.
+		s = foldMetricsEvent(s, doneEvent("t2"));
+		expect(selectCurrentContextSize(s)).toBe(700);
+	});
+
+	it("selectCurrentContextSize is undefined (not 0) when nothing reported", () => {
+		let s = initialMetricsState();
+		expect(selectCurrentContextSize(s)).toBeUndefined();
+		s = foldMetricsEvent(s, doneEvent("t1"));
+		expect(selectCurrentContextSize(s)).toBeUndefined();
+	});
+
+	it("durable contextSize wins over live for a shared turnId", () => {
+		let s = initialMetricsState();
+		s = foldMetricsEvent(s, doneEvent("t1", { contextSize: 111 }));
+		s = applyDurableMetrics(s, [
+			{ turnId: "t1", usage: { inputTokens: 1, outputTokens: 1 }, steps: [], contextSize: 222 },
+		]);
+		expect(selectCurrentContextSize(s)).toBe(222);
+	});
+});
diff --git a/src/core/metrics/reducer.ts b/src/core/metrics/reducer.ts
index d36dba1..1e66cc8 100644
--- a/src/core/metrics/reducer.ts
+++ b/src/core/metrics/reducer.ts
@@ -62,6 +62,9 @@ function liveTurnToMetrics(lt: LiveTurn): TurnMetrics {
 	if (lt.durationMs !== undefined) {
 		(base as { durationMs?: number }).durationMs = lt.durationMs;
 	}
+	if (lt.doneContextSize !== undefined) {
+		(base as { contextSize?: number }).contextSize = lt.doneContextSize;
+	}
 	return base;
 }
 
@@ -74,6 +77,7 @@ function ensureLiveTurn(state: MetricsState, turnId: string): [MetricsState, Liv
 		done: false,
 		durationMs: undefined,
 		doneUsage: undefined,
+		doneContextSize: undefined,
 		stepMap: new Map(),
 		stepOrder: [],
 	};
@@ -127,7 +131,7 @@ export function initialMetricsState(): MetricsState {
  * - `usage` with `stepId`: upsert that step's usage.
  * - `usage` without `stepId`: ignored.
  * - `step-complete`: upsert that step's timing; default usage to zeros if absent.
- * - `done`: set turn's `durationMs` and optional aggregate `usage`.
+ * - `done`: set turn's `durationMs`, optional aggregate `usage`, and optional `contextSize`.
  * - All other event types: return state unchanged.
  */
 export function foldMetricsEvent(state: MetricsState, event: AgentEvent): MetricsState {
@@ -161,6 +165,7 @@ export function foldMetricsEvent(state: MetricsState, event: AgentEvent): Metric
 				done: true,
 				durationMs: event.durationMs ?? lt.durationMs,
 				doneUsage: event.usage ?? lt.doneUsage,
+				doneContextSize: event.contextSize ?? lt.doneContextSize,
 			};
 			const newLive = new Map(s1.live);
 			newLive.set(event.turnId, updated);
@@ -237,3 +242,22 @@ export function selectOrderedTurnMetrics(state: MetricsState): readonly TurnMetr
 
 	return result;
 }
+
+/**
+ * Select the conversation's CURRENT context size — the tokens it occupies right
+ * now. Per the wire contract a client reads the LATEST turn's `contextSize`; we
+ * scan the merged ordered turns NEWEST → OLDEST and return the first DEFINED
+ * `contextSize` (a finalized turn whose provider reported per-step usage).
+ *
+ * Returns `undefined` ("unknown") when no finalized turn carries a context size —
+ * the caller renders a placeholder, NEVER `0`. Durable (sealed) data wins over
+ * live for a shared `turnId` (it is the persisted, authoritative value).
+ */
+export function selectCurrentContextSize(state: MetricsState): number | undefined {
+	const ordered = selectOrderedTurnMetrics(state);
+	for (let i = ordered.length - 1; i >= 0; i--) {
+		const total = ordered[i]?.total;
+		if (total?.contextSize !== undefined) return total.contextSize;
+	}
+	return undefined;
+}
diff --git a/src/core/metrics/types.ts b/src/core/metrics/types.ts
index f5557f7..c22fd9f 100644
--- a/src/core/metrics/types.ts
+++ b/src/core/metrics/types.ts
@@ -19,6 +19,12 @@ export interface LiveTurn {
 	readonly done: boolean;
 	readonly durationMs: number | undefined;
 	readonly doneUsage: Usage | undefined;
+	/**
+	 * Context size carried on the turn's `done` event (the turn's FINAL step
+	 * `inputTokens + outputTokens` — current context occupancy). `undefined` when
+	 * the provider reported no per-step usage; never coerced to `0`.
+	 */
+	readonly doneContextSize: number | undefined;
 	readonly stepMap: ReadonlyMap<string, BuildingStep>;
 	readonly stepOrder: readonly string[];
 }
diff --git a/src/features/chat/index.ts b/src/features/chat/index.ts
index 18ed693..adfb670 100644
--- a/src/features/chat/index.ts
+++ b/src/features/chat/index.ts
@@ -6,6 +6,7 @@ export type { ChatStore, ChatStoreDependencies } from "./store.svelte";
 export { createChatStore } from "./store.svelte";
 export { default as ChatView } from "./ui/ChatView.svelte";
 export { default as Composer } from "./ui/Composer.svelte";
+export { default as ContextSizeBadge } from "./ui/ContextSizeBadge.svelte";
 export { default as ModelSelector } from "./ui/ModelSelector.svelte";
 
 /** Public module manifest — aggregated by the shell's "Loaded Modules" view. */
diff --git a/src/features/chat/store.svelte.ts b/src/features/chat/store.svelte.ts
index f4ad07b..6344aec 100644
--- a/src/features/chat/store.svelte.ts
+++ b/src/features/chat/store.svelte.ts
@@ -18,6 +18,7 @@ import {
 	applyDurableMetrics,
 	foldMetricsEvent,
 	initialMetricsState,
+	selectCurrentContextSize,
 	selectOrderedTurnMetrics,
 } from "../../core/metrics";
 import type { ConversationCache } from "../conversation-cache";
@@ -36,6 +37,12 @@ export interface ChatStore {
 	readonly messages: readonly ChatMessage[];
 	readonly chunks: readonly RenderedChunk[];
 	readonly turnMetrics: readonly TurnMetricsEntry[];
+	/**
+	 * The conversation's current context size (tokens occupied) — the latest
+	 * finalized turn's `contextSize`, or `undefined` ("unknown") when none is
+	 * known yet. Never `0` for the unknown case.
+	 */
+	readonly currentContextSize: number | undefined;
 	readonly pendingSync: boolean;
 	readonly error: string | null;
 	readonly model: string | undefined;
@@ -91,6 +98,9 @@ export function createChatStore(deps: ChatStoreDependencies): ChatStore {
 		get turnMetrics(): readonly TurnMetricsEntry[] {
 			return selectOrderedTurnMetrics(metrics);
 		},
+		get currentContextSize(): number | undefined {
+			return selectCurrentContextSize(metrics);
+		},
 		get pendingSync(): boolean {
 			return _pendingSync;
 		},
diff --git a/src/features/chat/ui/ContextSizeBadge.svelte b/src/features/chat/ui/ContextSizeBadge.svelte
new file mode 100644
index 0000000..475d54f
--- /dev/null
+++ b/src/features/chat/ui/ContextSizeBadge.svelte
@@ -0,0 +1,20 @@
+<script lang="ts">
+	import { formatContextSize } from "../../../core/metrics";
+
+	let {
+		contextSize,
+	}: {
+		// The conversation's current context size (tokens occupied), or `undefined`
+		// ("unknown") when no finalized turn has reported one yet. Never `0` for the
+		// unknown case — `formatContextSize` renders a placeholder instead.
+		contextSize: number | undefined;
+	} = $props();
+
+	const label = $derived(formatContextSize(contextSize));
+</script>
+
+<div class="px-4 pb-1 text-xs opacity-60" aria-live="polite">
+	<span title="The model's max context window is not reported yet — current usage only.">
+		{label}
+	</span>
+</div>