Render file change diffs and index session activity from item timestamps

AGENTS.md

@@ -48,3 +48,4 @@ Per L2-DES-APP-008 and L2-DES-CONV-002 (both Approved):
 - Session settings changes go through canonical `session/metadata/update` with `SessionSettingsPatch` (partial semantics: only present fields change). Do not add per-concern settings methods.
 - Settings writes are persist-first: field-level rollout lines (`InternalRecordV2::SessionSettings`) written synchronously by the handler, which never waits on the session actor; actor notification is best-effort (`SessionHandle::notify_*`). Replay prefers field lines over whole-record `SessionMeta` values.
 - Mid-turn effect rides the turn-inline override (`TurnInlineState.live_turn_settings`, `sandbox_profile_live`). Every live setting must declare its decision point and mid-turn semantics in the DD-6 promise matrix of L2-DES-CONV-002 before implementation.
+- Session actor vs turn isolation: see `L2-DES-SERVER-002` and `crates/server/AGENTS.md`. Actor mailbox commands must stay short; turns run on a spawned task with `TurnWorkingSet` and re-enter via `MergeTurn`.

apps/desktop/package.json

@@ -61,6 +61,7 @@
 		"@devo/configconv": "file:./packages/configconv",
 		"@devo/ui": "file:./packages/ui",
 		"@fontsource-variable/inter": "^5.2.8",
+		"@fontsource-variable/noto-sans-sc": "^5.3.0",
 		"@fontsource/ibm-plex-mono": "^5.2.7",
 		"@libsql/client": "^0.17.0",
 		"@pierre/diffs": "^1.0.10",

apps/desktop/packages/devo-ai-sdk/src/v2/client-config-options.test.ts

@@ -48,10 +48,32 @@ const initializeResult = {
 const modelPreferences = {
 	model: "test-openai",
 	availableModels: [
-		{ value: "test-openai", label: "Test OpenAI", description: "OpenAI: test-model" },
-		{ value: "alt-openai", label: "Alt OpenAI", description: "OpenAI: alt-model" },
+		{
+			value: "test-openai",
+			label: "Test OpenAI",
+			description: "OpenAI: test-model",
+			availableEfforts: [
+				{ value: "low", label: "Low" },
+				{ value: "medium", label: "Medium" },
+				{ value: "high", label: "High" },
+			],
+		},
+		{
+			value: "alt-openai",
+			label: "Alt OpenAI",
+			description: "OpenAI: alt-model",
+			availableEfforts: [
+				{ value: "high", label: "High" },
+				{ value: "max", label: "Max" },
+			],
+		},
+	],
+	availableEfforts: [
+		{ value: "low", label: "Low" },
+		{ value: "medium", label: "Medium" },
+		{ value: "high", label: "High" },
 	],
-	availableEfforts: [],
+	reasoningEffort: "medium",
 }
 
 const configOptions = [
@@ -131,6 +153,17 @@ describe("Native desktop SDK config option cache", () => {
 		const config = await client.config.get()
 
 		expect(providers.data.default).toEqual({ session: "test-openai" })
+		expect(Object.keys(providers.data.providers[0].models["test-openai"].variants)).toEqual([
+			"low",
+			"medium",
+			"high",
+		])
+		expect(Object.keys(providers.data.providers[0].models["alt-openai"].variants)).toEqual([
+			"high",
+			"max",
+		])
+		expect(providers.data.providers[0].models["test-openai"].currentVariant).toBe("medium")
+		expect(providers.data.providers[0].models["alt-openai"].currentVariant).toBeUndefined()
 		expect(config.data).toEqual({ model: "session/test-openai" })
 		expect(transport.requests.map((request) => request.method)).toEqual([
 			"initialize",