@xopcai/xopc 0.0.389 → 0.0.390

package/dist/src/gateway/hono/lib/config-payload.js

@@ -14,6 +14,7 @@ import { buildChannelCatalogForConfig, buildChannelCatalogFromSnapshot, getChann
 import { resolveSessionConfigForWeb, resolveUpdateConfigForWeb } from "../../../config/web-patch.js";
 import { GENERIC_MASKED_SECRET, maskSecretLength } from "./mask-secret-length.js";
 import { safeToolsWebForGet } from "../../config-tools-web.js";
+import { voiceSettingsRevisions } from "./voice-settings-revisions.js";
 import { buildSafeProvidersConfigForWeb } from "./safe-providers-config.js";
 import { maskRealtimeVoiceConfigForWeb, maskSttConfigForWeb, maskTtsConfigForWeb } from "./safe-voice-config.js";
 //#region src/gateway/hono/lib/config-payload.ts
@@ -155,6 +156,7 @@ async function buildSafeWebConfigPayload(service, options = {}) {
 			transport: { tls: "broker_terminated" }
 		},
 		update: { ...resolveUpdateConfigForWeb(config) },
+		voiceSettingsRevisions: voiceSettingsRevisions(config),
 		stt: maskSttConfigForWeb(config.tools?.media?.audio),
 		tts: maskTtsConfigForWeb(config.messages?.tts),
 		voice: maskRealtimeVoiceConfigForWeb(config.voice),

package/dist/src/gateway/hono/lib/voice-settings-revisions.d.ts

@@ -0,0 +1,5 @@
+import type { Config } from '../../../config/schema.js';
+export declare const VOICE_SETTING_SCOPES: readonly ["transcription", "speech", "conversation"];
+export type VoiceSettingScope = typeof VOICE_SETTING_SCOPES[number];
+/** Opaque per-capability versions; credentials never appear in the public snapshot. */
+export declare function voiceSettingsRevisions(config: Config): Record<VoiceSettingScope, string>;

package/dist/src/gateway/hono/lib/voice-settings-revisions.js

@@ -0,0 +1,23 @@
+import { createHmac, randomBytes } from "node:crypto";
+//#region src/gateway/hono/lib/voice-settings-revisions.ts
+const revisionKey = randomBytes(32);
+const VOICE_SETTING_SCOPES = [
+	"transcription",
+	"speech",
+	"conversation"
+];
+/** Opaque per-capability versions; credentials never appear in the public snapshot. */
+function voiceSettingsRevisions(config) {
+	const values = {
+		transcription: {
+			stt: config.tools?.media?.audio,
+			language: config.voice?.language,
+			languageMode: config.voice?.languageMode
+		},
+		speech: config.messages?.tts,
+		conversation: config.voice?.realtime
+	};
+	return Object.fromEntries(VOICE_SETTING_SCOPES.map((scope) => [scope, createHmac("sha256", revisionKey).update(JSON.stringify(values[scope]) ?? "null").digest("hex")]));
+}
+//#endregion
+export { VOICE_SETTING_SCOPES, voiceSettingsRevisions };

package/dist/src/gateway/hono/routes/config-patch/misc.js

@@ -216,7 +216,14 @@ async function applyMiscPatch(config, body) {
 		config.messages.tts = mergeTtsConfigPatch(config.messages.tts, body.tts);
 	}
 	if (body.voice !== void 0) {
-		const parsed = VoiceConfigSchema.safeParse(mergeRealtimeVoiceConfigPatch(config.voice, body.voice));
+		if (!body.voice || typeof body.voice !== "object" || Array.isArray(body.voice)) return patchError("voice must be an object");
+		if (body.voice.realtime !== void 0 && (!body.voice.realtime || typeof body.voice.realtime !== "object" || Array.isArray(body.voice.realtime))) return patchError("voice.realtime must be an object");
+		if (body.voicePatchMode !== void 0 && body.voicePatchMode !== "merge") return patchError("Invalid voice patch mode");
+		const voicePatch = body.voicePatchMode === "merge" ? {
+			...config.voice,
+			...body.voice
+		} : body.voice;
+		const parsed = VoiceConfigSchema.safeParse(mergeRealtimeVoiceConfigPatch(config.voice, voicePatch));
 		if (!parsed.success) return patchError(parsed.error.issues.map((i) => i.message).join("; "));
 		config.voice = parsed.data;
 	}

package/dist/src/gateway/hono/routes/config.js

@@ -1,5 +1,6 @@
 import { enumerateLanGatewayCandidates, init_host } from "../../host.js";
 import { getTunnelService } from "../../../tunnel/tunnel-service.js";
+import { VOICE_SETTING_SCOPES, voiceSettingsRevisions } from "../lib/voice-settings-revisions.js";
 import { buildSafeWebConfigPayload } from "../lib/config-payload.js";
 import { withGatewayConfigWriteLock } from "../lib/config-write-lock.js";
 import { applyChannelsPatch } from "./config-patch/channels.js";
@@ -39,6 +40,17 @@ function registerConfigRoutes(authenticated, deps) {
 		const body = await c.req.json();
 		const expectedConfig = structuredClone(service.currentConfig);
 		const config = structuredClone(expectedConfig);
+		if (body.expectedVoiceRevision !== void 0) {
+			const guard = body.expectedVoiceRevision;
+			if (!guard || typeof guard !== "object" || !VOICE_SETTING_SCOPES.includes(guard.capability) || typeof guard.revision !== "string" || !/^[a-f0-9]{64}$/.test(guard.revision)) return c.json({
+				ok: false,
+				error: "Invalid voice settings revision"
+			}, 400);
+			if (voiceSettingsRevisions(expectedConfig)[guard.capability] !== guard.revision) return c.json({
+				ok: false,
+				error: "Voice settings changed. Reload before saving to preserve the current model."
+			}, 409);
+		}
 		applyChannelsPatch(config, body);
 		const gatewayResult = applyGatewayPatch(config, body);
 		if (gatewayResult.ok === false) return c.json({

package/dist/src/gateway/hono/routes/discussions.js

@@ -79,10 +79,67 @@ function registerDiscussionRoutes(authenticated, deps) {
 		const detail = await service.discussions.getByNoteId(c.req.param("noteId"));
 		return detail ? c.json(detail) : c.json({ error: "Discussion not found" }, 404);
 	});
+	authenticated.get("/api/discussions/:id/review", async (c) => {
+		try {
+			return c.json(await service.discussions.review(c.req.param("id"), {
+				query: c.req.query("q"),
+				offset: Number(c.req.query("offset") ?? 0),
+				limit: Number(c.req.query("limit") ?? 20),
+				version: c.req.query("version")
+			}));
+		} catch (error) {
+			const response = errorResponse(error);
+			if (response) return c.json(response.body, response.status);
+			throw error;
+		}
+	});
 	authenticated.get("/api/discussions/:id", async (c) => {
 		const detail = await service.discussions.get(c.req.param("id"));
 		return detail ? c.json(detail) : c.json({ error: "Discussion not found" }, 404);
 	});
+	authenticated.get("/api/discussions/:id/audio-clip", mediaRateLimitMiddleware, async (c) => {
+		try {
+			const bytes = await service.discussions.audioClip(c.req.param("id"), Number(c.req.query("startMs")), Number(c.req.query("endMs")), c.req.raw.signal);
+			c.header("Cache-Control", "no-store");
+			c.header("Content-Type", "audio/wav");
+			return c.body(new Uint8Array(bytes));
+		} catch (error) {
+			const response = errorResponse(error);
+			if (response) return c.json(response.body, response.status);
+			throw error;
+		}
+	});
+	authenticated.get("/api/discussions/:id/annotations", async (c) => {
+		try {
+			return c.json(await service.discussions.annotations(c.req.param("id")));
+		} catch (error) {
+			const response = errorResponse(error);
+			if (response) return c.json(response.body, response.status);
+			throw error;
+		}
+	});
+	authenticated.put("/api/discussions/:id/annotations/:annotationId", strictRateLimitMiddleware, async (c) => {
+		const body = await c.req.json().catch(() => ({}));
+		try {
+			return c.json(service.discussions.saveAnnotation(c.req.param("id"), {
+				...body,
+				id: c.req.param("annotationId")
+			}));
+		} catch (error) {
+			const response = errorResponse(error);
+			if (response) return c.json(response.body, response.status);
+			throw error;
+		}
+	});
+	authenticated.get("/api/discussions/:id/evidence", async (c) => {
+		try {
+			return c.json(await service.discussions.evidence(c.req.param("id"), Number(c.req.query("revision")), (c.req.query("segments") ?? "").split(",").filter(Boolean).map(Number)));
+		} catch (error) {
+			const response = errorResponse(error);
+			if (response) return c.json(response.body, response.status);
+			throw error;
+		}
+	});
 	authenticated.get("/api/discussions/:id/transcript", (c) => {
 		const transcript = service.discussions.transcript(c.req.param("id"), c.req.query("revision") === void 0 ? void 0 : Number(c.req.query("revision")));
 		return transcript ? c.json(transcript) : c.json({ error: "Discussion not found" }, 404);
@@ -243,6 +300,8 @@ function registerDiscussionRoutes(authenticated, deps) {
 				chunkCount: Number(body.chunkCount),
 				mimeType: typeof body.mimeType === "string" ? body.mimeType : "",
 				fileName: typeof body.fileName === "string" ? body.fileName : "",
+				...body.events !== void 0 ? { events: body.events } : {},
+				...body.epochs !== void 0 ? { epochs: body.epochs } : {},
 				...body.containerMode === "independent_wav" ? { containerMode: "independent_wav" } : {}
 			}), 202);
 		} catch (error) {
@@ -252,8 +311,11 @@ function registerDiscussionRoutes(authenticated, deps) {
 		}
 	});
 	authenticated.post("/api/discussions/:id/retry", strictRateLimitMiddleware, async (c) => {
+		const body = await c.req.json().catch(() => ({}));
+		if (!body || typeof body !== "object" || Array.isArray(body)) return c.json({ error: "Invalid retry request" }, 400);
+		if (body.mode !== void 0 && body.mode !== "resume" && body.mode !== "from_audio") return c.json({ error: "Invalid retry mode" }, 400);
 		try {
-			const detail = await service.discussions.retry(c.req.param("id"));
+			const detail = await service.discussions.retry(c.req.param("id"), body.mode === "from_audio" ? "from_audio" : "resume");
 			return detail ? c.json(detail) : c.json({ error: "Discussion not found" }, 404);
 		} catch (error) {
 			const response = errorResponse(error);

package/dist/src/gateway/hono/routes/global-defaults.js

@@ -6,8 +6,15 @@ init_agent_config();
 function parsePatchBody(raw) {
 	if (raw === null || typeof raw !== "object" || Array.isArray(raw)) return { error: "body must be an object" };
 	const body = raw;
+	if (body.scope !== void 0 && body.scope !== "models" && body.scope !== "agent") return { error: "scope must be models or agent" };
+	const expected = body.expectedModels === void 0 ? void 0 : AgentDefaultsSchema.shape.models.safeParse(body.expectedModels);
+	if (expected && !expected.success) return { error: "expectedModels is invalid" };
 	const parsed = AgentDefaultsSchema.safeParse(body.defaults);
-	return parsed.success ? { defaults: parsed.data } : { error: `defaults ${parsed.error.issues[0]?.message ?? "is invalid"}` };
+	return parsed.success ? {
+		defaults: parsed.data,
+		...body.scope ? { scope: body.scope } : {},
+		...expected?.success ? { expectedModels: expected.data } : {}
+	} : { error: `defaults ${parsed.error.issues[0]?.message ?? "is invalid"}` };
 }
 function isParseError(value) {
 	return typeof value === "object" && value !== null && "error" in value && typeof value.error === "string";

package/dist/src/gateway/hono/routes/models.js

@@ -1,3 +1,5 @@
+import { getProviderSetupProfile, init_profiles } from "../../../providers/setup/profiles.js";
+import { init_model_usage, withModelUsageOrigin } from "../../../providers/model-usage.js";
 import { resolveEffectiveAgentConfigForAgent } from "../../../config/agent-profile.js";
 import { init_resolve_config_value, testApiKeyResolution } from "../../../config/resolve-config-value.js";
 import { resolveModelsJsonPath } from "../../../config/paths.js";
@@ -20,11 +22,14 @@ import { applyImageGenerationCatalogUpdate, getAgentImageGenerationConfig, getDe
 import { discoverProviderModels, isProviderApiDiscoverable } from "../../../providers/model-discovery.js";
 import { getXopcCloudCatalogCoordinator } from "../../../providers/xopc-cloud-catalog-coordinator.js";
 import { respondStartupUnavailable } from "../lib/startup-unavailable.js";
-import { dismissXopcCloudOnboarding, getXopcCloudOnboardingStatus } from "../../xopc-cloud-onboarding.js";
 import { getDomesticProviderPreset, getDomesticProviderPresetIds } from "../../../providers/domestic-presets.js";
-import { auditModelReferences } from "../../../providers/model-reference-auditor.js";
 import { getOnboardingFeaturedProviders, getProviderHint, getRecommendedModelsForProvider, isRecommendedModel, sortModelsForPicker, sortProvidersForPicker } from "../../../providers/presentation.js";
+import { dismissXopcCloudOnboarding, getXopcCloudOnboardingStatus } from "../../xopc-cloud-onboarding.js";
+import { auditModelReferences } from "../../../providers/model-reference-auditor.js";
+import { registerProviderSetupRoutes } from "./provider-setup.js";
 //#region src/gateway/hono/routes/models.ts
+init_model_usage();
+init_profiles();
 init_model_policy();
 init_models_json();
 init_resolve_config_value();
@@ -83,6 +88,7 @@ function mapPluginModel(providerId, model, available) {
 	};
 }
 function registerModelsRoutes(authenticated, deps) {
+	registerProviderSetupRoutes(authenticated, deps);
 	const { service, strictRateLimitMiddleware } = deps;
 	const catalogSync = service.getModelCatalogSync();
 	const xopcCloudAccount = new XopcCloudAccountService();
@@ -518,14 +524,14 @@ function registerModelsRoutes(authenticated, deps) {
 		}, 400);
 		const prompt = typeof body.prompt === "string" && body.prompt.trim() ? body.prompt.trim() : "A simple blue circle centered on a white background.";
 		try {
-			const result = await provider.generateImage({
+			const result = await withModelUsageOrigin("interactive", () => provider.generateImage({
 				provider: providerId,
 				model: modelId,
 				prompt,
 				count: 1,
 				cfg: service.currentConfig,
 				timeoutMs: 12e4
-			});
+			}));
 			return c.json({
 				ok: true,
 				payload: {
@@ -744,6 +750,7 @@ function registerModelsRoutes(authenticated, deps) {
 			const registryModelCount = getAllModels().filter((m) => m.provider === provider).length;
 			return {
 				id: provider,
+				setupProfile: getProviderSetupProfile(provider, PROVIDER_META[provider]?.name ?? domesticPreset?.displayName ?? provider, PROVIDER_META[provider]?.supportsOAuth, resolveProviderApiBaseUrl(provider)),
 				name: plugin?.name ?? domesticPreset?.displayName ?? PROVIDER_META[provider]?.name ?? provider,
 				category: plugin ? "extension" : PROVIDER_META[provider]?.category || (domesticPreset ? "domestic" : "specialty"),
 				supportsOAuth: plugin ? false : PROVIDER_META[provider]?.supportsOAuth ?? false,
@@ -766,6 +773,7 @@ function registerModelsRoutes(authenticated, deps) {
 			const extensionId = resolveExtensionIdForProvider(service, plugin.id);
 			meta.push({
 				id: plugin.id,
+				setupProfile: getProviderSetupProfile(plugin.id, plugin.name),
 				name: plugin.name,
 				category: "extension",
 				supportsOAuth: false,

package/dist/src/gateway/hono/routes/provider-setup.d.ts

@@ -0,0 +1,3 @@
+import type { Hono } from 'hono';
+import type { AuthenticatedRouteDeps } from './deps.js';
+export declare function registerProviderSetupRoutes(app: Hono, deps: AuthenticatedRouteDeps): void;

package/dist/src/gateway/hono/routes/provider-setup.js

@@ -0,0 +1,106 @@
+import { listGlobalDefaults, prepareUpdateGlobalDefaults, updateGlobalDefaults } from "../../global-defaults-admin.js";
+import { applySetupVoice } from "../../../providers/setup/voice.js";
+import { loadProviderSetupSnapshot } from "../../provider-model-setup.js";
+import { SetupConflictError, applyProviderSetup, previewProviderSetup } from "../../../providers/setup/engine.js";
+import { withGatewayConfigWriteLock } from "../lib/config-write-lock.js";
+import { isDeepStrictEqual } from "node:util";
+import { z } from "zod";
+//#region src/gateway/hono/routes/provider-setup.ts
+const use = z.enum([
+	"chat",
+	"vision",
+	"image",
+	"background",
+	"speech",
+	"transcription",
+	"conversation"
+]);
+const requestSchema = z.object({
+	mode: z.enum(["fill", "replace"]).optional(),
+	selected: z.array(use).max(7).optional(),
+	overrides: z.partialRecord(use, z.string().min(1).max(512)).optional(),
+	modelIds: z.array(z.string().min(1).max(512)).max(500).optional()
+}).strict();
+const applySchema = requestSchema.extend({ revision: z.string().regex(/^[a-f0-9]{64}$/) });
+function registerProviderSetupRoutes(app, deps) {
+	app.get("/api/models/setup/:providerId", async (c) => {
+		try {
+			return c.json({
+				ok: true,
+				payload: previewProviderSetup(await loadProviderSetupSnapshot(c.req.param("providerId"), deps.service.currentConfig))
+			});
+		} catch (error) {
+			return c.json({
+				ok: false,
+				error: { message: error instanceof Error ? error.message : String(error) }
+			}, 400);
+		}
+	});
+	app.post("/api/models/setup/:providerId/preview", deps.strictRateLimitMiddleware, async (c) => {
+		const parsed = requestSchema.safeParse(await c.req.json().catch(() => null));
+		if (!parsed.success) return c.json({
+			ok: false,
+			error: { message: "Invalid setup preview request" }
+		}, 400);
+		try {
+			return c.json({
+				ok: true,
+				payload: previewProviderSetup(await loadProviderSetupSnapshot(c.req.param("providerId"), deps.service.currentConfig), parsed.data)
+			});
+		} catch (error) {
+			return c.json({
+				ok: false,
+				error: { message: error instanceof Error ? error.message : String(error) }
+			}, 400);
+		}
+	});
+	app.post("/api/models/setup/:providerId/apply", deps.strictRateLimitMiddleware, async (c) => withGatewayConfigWriteLock(deps.service, async () => {
+		const parsed = applySchema.safeParse(await c.req.json().catch(() => null));
+		if (!parsed.success) return c.json({
+			ok: false,
+			error: { message: "Invalid setup apply request" }
+		}, 400);
+		let voiceSaved = false;
+		try {
+			const expectedConfig = structuredClone(deps.service.currentConfig);
+			const snapshot = await loadProviderSetupSnapshot(c.req.param("providerId"), expectedConfig);
+			if (!isDeepStrictEqual(deps.service.currentConfig, expectedConfig)) throw new SetupConflictError("Configuration changed while preparing setup. Review and retry.");
+			const applied = applyProviderSetup(snapshot, parsed.data, parsed.data.revision);
+			const prepared = prepareUpdateGlobalDefaults({ defaults: applied.defaults });
+			if (prepared.ok === false) return c.json({
+				ok: false,
+				error: { message: prepared.error }
+			}, prepared.status ?? 400);
+			if (applied.voiceAssignments.length) {
+				const next = applySetupVoice(expectedConfig, applied.voiceAssignments);
+				const saved = await deps.service.saveConfig(next, expectedConfig);
+				if (!saved.saved) return c.json({
+					ok: false,
+					error: { message: `${saved.persisted ? "Voice settings persisted but runtime reload failed" : "Voice settings were not saved"}: ${saved.error ?? "Configuration write failed"}` }
+				}, saved.conflict ? 409 : 500);
+				voiceSaved = true;
+				if (!isDeepStrictEqual(listGlobalDefaults().defaults, snapshot.defaults)) return c.json({
+					ok: false,
+					error: { message: "Voice settings saved; model defaults changed concurrently. Review the refreshed configuration before applying model uses." }
+				}, 409);
+			}
+			const result = updateGlobalDefaults({ defaults: applied.defaults });
+			if (result.ok === false) return c.json({
+				ok: false,
+				error: { message: result.error }
+			}, result.status ?? 400);
+			if (result.data.changed) deps.service.refreshAgentCatalog();
+			return c.json({
+				ok: true,
+				payload: applied.preview
+			});
+		} catch (error) {
+			return c.json({
+				ok: false,
+				error: { message: (voiceSaved ? "Voice settings saved; model setup incomplete. " : "") + (error instanceof Error ? error.message : String(error)) }
+			}, error instanceof SetupConflictError ? 409 : 400);
+		}
+	}));
+}
+//#endregion
+export { registerProviderSetupRoutes };

package/dist/src/gateway/hono/routes/status.js

@@ -3,6 +3,7 @@ import { isSTTAvailable } from "../../../voice/stt/availability.js";
 import { mergeSttConfigFromAppConfig } from "../../../voice/stt/config.js";
 import "../../../voice/stt/index.js";
 import { getGatewayAgentEffectiveConfig } from "../../agents-admin.js";
+import { resolveDictationRoute } from "../../../voice/dictation.js";
 import { createHash } from "node:crypto";
 //#region src/gateway/hono/routes/status.ts
 init_providers();
@@ -82,6 +83,7 @@ function buildStatusPayload(service) {
 			sttAvailable: sttConfig ? isSTTAvailable(sttConfig) : false,
 			sttEnabled: sttConfig?.enabled === true,
 			sttProvider: sttConfig?.provider ?? null,
+			dictation: config ? resolveDictationRoute(config) : { mode: "unavailable" },
 			refineAvailable: isRefineAvailable(config)
 		}
 	};

package/dist/src/gateway/hono/routes/voice.js

@@ -1,3 +1,4 @@
+import { init_model_usage, withModelUsageOrigin } from "../../../providers/model-usage.js";
 import { voiceSelectionSchema } from "../../../../packages/realtime-protocol/src/voice-catalog.js";
 import { VOICE_TIMING_METRICS, createVoiceSessionRequestSchema, init_voice } from "../../../../packages/realtime-protocol/src/voice.js";
 import { getDefaultModelSync, init_providers, resolveModel } from "../../../providers/index.js";
@@ -5,6 +6,7 @@ import { completeWithResolvedCredentials } from "../../../providers/model-call.j
 import { isPersonalConversation } from "../../../personal-agent/repository.js";
 import { PlatformVoiceError } from "../../../voice/platform-streams.js";
 import { resolveSttProviderConfigSlice } from "../../../voice/stt/config-slice.js";
+import { resolveSTTProviderChain } from "../../../voice/stt/factory.js";
 import { STTTranscriptionError, transcribe } from "../../../voice/stt/transcribe-core.js";
 import { isSTTAvailable } from "../../../voice/stt/availability.js";
 import { mergeSttConfigFromAppConfig } from "../../../voice/stt/config.js";
@@ -21,10 +23,11 @@ import { resolveOmniRoute } from "../../../voice/realtime/omniRoute.js";
 import { VoiceSessionCreationError, resolveStreamingStt, resolveStreamingTts } from "../../../voice/realtime/runtime.js";
 import { getGatewayPrincipal } from "../../security/gateway-principal.js";
 import { createGatewayRouteLogger } from "../lib/route-logger.js";
-import { selectPlatformVoice, voiceSettingsCatalog } from "../../../voice/platform-settings.js";
 import { listTtsProvidersForApi } from "../../../voice/tts/list-providers.js";
+import { selectPlatformVoice, voiceSettingsCatalog } from "../../../voice/platform-settings.js";
 import { z } from "zod";
 //#region src/gateway/hono/routes/voice.ts
+init_model_usage();
 init_voice();
 init_providers();
 const log = createGatewayRouteLogger("Voice");
@@ -104,7 +107,10 @@ async function refineTranscript(raw, config, signal) {
 			maxTokens: Math.min(raw.length * 3, 4096),
 			temperature: .2,
 			signal: mergedSignal
-		}, void 0, { operation: "voice.select" });
+		}, void 0, {
+			operation: "voice.select",
+			trigger: "user"
+		});
 		let out = "";
 		if (Array.isArray(result.content)) {
 			for (const c of result.content) if (c && typeof c === "object" && c.type === "text") out += String(c.text || "");
@@ -246,12 +252,13 @@ function registerVoiceRoutes(authenticated, deps) {
 		let result;
 		let reader;
 		try {
-			result = await speakStream(config.voice?.language === "en" ? "Hello, I am your assistant. You can talk to me naturally." : "你好，我是你的助手。现在可以和我自然对话了。", previewConfig, {
+			const text = config.voice?.language === "en" ? "Hello, I am your assistant. You can talk to me naturally." : "你好，我是你的助手。现在可以和我自然对话了。";
+			result = await withModelUsageOrigin("interactive", () => speakStream(text, previewConfig, {
 				signal,
 				appConfig: config,
 				allowFallback: false,
 				parseDirectives: false
-			});
+			}));
 			if (result.outputFormat !== "pcm") throw new Error("Realtime preview requires PCM audio");
 			reader = result.audioStream.getReader();
 			const chunks = [];
@@ -551,12 +558,12 @@ function registerVoiceRoutes(authenticated, deps) {
 			error: { message: `TTS provider \"${provider}\" is not configured.` }
 		}, 503);
 		try {
-			const result = await speak(text, ttsConfig, {
+			const result = await withModelUsageOrigin("interactive", () => speak(text, ttsConfig, {
 				appConfig: config,
 				parseDirectives: false,
 				signal: c.req.raw.signal,
 				tts: voice ? { voice } : void 0
-			});
+			}));
 			if (!result.audio.length) throw new Error(`TTS provider \"${result.provider}\" returned empty audio`);
 			return new Response(new Uint8Array(result.audio), { headers: {
 				"Content-Type": speechMimeType(result.format),
@@ -630,14 +637,14 @@ function registerVoiceRoutes(authenticated, deps) {
 		}, 503);
 		try {
 			const startedAt = Date.now();
-			const result = await speak(text, ttsConfig, {
+			const result = await withModelUsageOrigin("interactive", () => speak(text, ttsConfig, {
 				appConfig: config,
 				parseDirectives: false,
 				tts: {
 					...typeof body.model === "string" && body.model.trim() ? { model: body.model.trim() } : {},
 					...typeof body.voice === "string" && body.voice.trim() ? { voice: body.voice.trim() } : {}
 				}
-			});
+			}));
 			if (!result.audio.length) throw new Error(`TTS provider "${result.provider}" returned empty audio`);
 			const mimeType = result.format === "opus" || result.format === "ogg" ? "audio/ogg" : result.format === "mp3" || result.format === "mpeg" ? "audio/mpeg" : result.format === "wav" ? "audio/wav" : `audio/${result.format}`;
 			return c.json({
@@ -678,18 +685,22 @@ function registerVoiceRoutes(authenticated, deps) {
 		}, 400);
 		const config = service.currentConfig;
 		const sttConfig = mergeSttConfigFromAppConfig(config.tools?.media?.audio, config.tools?.media);
+		if (input.expectedProvider && input.expectedProvider !== resolveSTTProviderChain(sttConfig)[0]?.id) return c.json({
+			ok: false,
+			error: { message: "Speech recognition provider changed during recording. Restore the previous model or start a new recording." }
+		}, 409);
 		if (!isSTTAvailable(sttConfig)) return c.json({
 			ok: false,
 			error: { message: "STT is not configured. Enable STT in gateway config (tools.media.audio)." }
 		}, 503);
 		const startedAt = Date.now();
 		try {
-			const result = await transcribe(input.audioBuffer, sttConfig, {
+			const result = await withModelUsageOrigin("interactive", () => transcribe(input.audioBuffer, sttConfig, {
 				language: input.language,
 				mime: input.mimeType,
 				fileName: input.fileName,
 				signal: c.req.raw.signal
-			});
+			}));
 			const detectedLanguage = result.language ?? input.language;
 			const refinementAvailable = (config.voice?.input?.refinement?.mode ?? "off") !== "off";
 			return c.json({
@@ -769,12 +780,14 @@ function registerVoiceRoutes(authenticated, deps) {
 			error: { message: "Missing required file field: audio" }
 		}, 400);
 		const languageValue = form.get("language");
+		const expectedProvider = form.get("expectedProvider");
 		const language = typeof languageValue === "string" && languageValue.trim() ? languageValue.trim() : void 0;
 		const fileName = "name" in audio && typeof audio.name === "string" && audio.name.trim() ? audio.name.trim() : `audio-${Date.now()}`;
 		return runTranscription(c, {
 			audioBuffer: Buffer.from(await audio.arrayBuffer()),
 			mimeType: audio.type || "audio/webm",
 			fileName,
+			...typeof expectedProvider === "string" && expectedProvider ? { expectedProvider } : {},
 			...language ? { language } : {}
 		});
 	});

package/dist/src/gateway/provider-model-setup.d.ts

@@ -0,0 +1,4 @@
+import type { SetupSnapshot } from '../providers/setup/engine.js';
+import type { Config } from '../config/schema.js';
+export declare function loadProviderSetupSnapshot(provider: string, config?: Config): Promise<SetupSnapshot>;
+export declare function readProviderSetupSnapshot(provider: string, configured: boolean, config?: Config): SetupSnapshot;

package/dist/src/gateway/provider-model-setup.js

@@ -0,0 +1,56 @@
+import { getProviderSetupProfile, init_profiles, isInteractiveOnlyProvider } from "../providers/setup/profiles.js";
+import { getModelCatalogStore, init_model_catalog_store } from "../providers/model-catalog-store.js";
+import { getAllModels, getAllProviders, getModelsByProvider, getProviderDisplayName, init_providers, isProviderConfigured, isProviderConfiguredSync, providerSupportsOAuth } from "../providers/index.js";
+import { getImageGenerationProvider } from "../agent/image/generation/provider-registry.js";
+import { listSttProvidersForApi } from "../voice/stt/list-providers.js";
+import { listGlobalDefaults } from "./global-defaults-admin.js";
+import { getRecommendedModelsForProvider } from "../providers/presentation.js";
+import { listTtsProvidersForApi } from "../voice/tts/list-providers.js";
+import { readSetupVoiceState } from "../providers/setup/voice.js";
+//#region src/gateway/provider-model-setup.ts
+init_providers();
+init_model_catalog_store();
+init_profiles();
+async function loadProviderSetupSnapshot(provider, config) {
+	const languageProvider = getAllProviders().includes(provider);
+	const mediaProvider = getImageGenerationProvider(provider) || listSttProvidersForApi(config).providers.some((entry) => entry.id === provider) || listTtsProvidersForApi(config).providers.some((entry) => entry.id === provider);
+	if (!languageProvider && !mediaProvider) throw new Error(`Unknown model provider: ${provider}`);
+	return readProviderSetupSnapshot(provider, languageProvider && await isProviderConfigured(provider), config);
+}
+function readProviderSetupSnapshot(provider, configured, config) {
+	const languageProvider = getAllProviders().includes(provider);
+	const models = (languageProvider ? getModelsByProvider(provider) : []).filter((model) => !("computerUse" in model && model.computerUse));
+	const recommended = new Set((languageProvider ? getRecommendedModelsForProvider(provider) : []).map((model) => model.ref));
+	const source = getModelCatalogStore().getSource(provider);
+	const image = getImageGenerationProvider(provider);
+	const images = image?.models.filter((id) => provider !== "xopc-cloud" || source?.models.some((model) => model.id === id && model.availability === "available")) ?? [];
+	const configuredProviders = new Set(getAllProviders().filter(isProviderConfiguredSync));
+	const backgroundRestrictedModels = new Set(getAllModels().filter((model) => isInteractiveOnlyProvider(model.provider, model.baseUrl)).map((model) => `${model.provider}/${model.id}`));
+	for (const model of models) if (/^https?:\/\/(localhost|127\.0\.0\.1|\[::1\])(?=[:/]|$)/.test(model.baseUrl)) configuredProviders.add(provider);
+	const voice = config ? readSetupVoiceState(provider, config) : void 0;
+	const imageConfigured = image?.isConfigured({ cfg: config }) ?? false;
+	const nativeVoiceConfigured = Boolean(!["qwen-token-plan-cn", "xopc-cloud"].includes(provider) && voice && Object.values(voice.choices).some((choices) => choices.length));
+	return {
+		...config ? { voice } : {},
+		configured: configured || configuredProviders.has(provider) || imageConfigured || nativeVoiceConfigured,
+		configuredProviders,
+		backgroundRestrictedModels,
+		defaults: listGlobalDefaults().defaults,
+		profile: getProviderSetupProfile(provider, getProviderDisplayName(provider), providerSupportsOAuth(provider), models[0]?.baseUrl),
+		models: (!configured && !configuredProviders.has(provider) && (nativeVoiceConfigured || imageConfigured) ? [] : models).map((model) => ({
+			ref: `${provider}/${model.id}`,
+			name: model.name,
+			vision: model.input.includes("image"),
+			reasoning: model.reasoning,
+			baseUrl: model.baseUrl,
+			api: model.api,
+			recommended: recommended.has(`${provider}/${model.id}`)
+		})),
+		images: (nativeVoiceConfigured && !imageConfigured && !configured ? [] : images).map((id) => ({
+			ref: `${provider}/${id}`,
+			name: id
+		}))
+	};
+}
+//#endregion
+export { loadProviderSetupSnapshot, readProviderSetupSnapshot };

package/dist/src/gateway/service/config-coordinator.d.ts

@@ -612,12 +612,12 @@ export declare class GatewayConfigCoordinator {
                     model: string;
                 };
                 tts?: {
-                    provider: "alibaba" | "xopc-cloud";
+                    provider: "xopc-cloud" | "alibaba";
                     model?: string;
                     voice?: string;
                 };
                 omni?: {
-                    provider: "alibaba" | "xopc-cloud";
+                    provider: "qwen-token-plan-cn" | "xopc-cloud" | "alibaba";
                     model: string;
                     voice: string;
                     instructions: string;

package/dist/src/gateway/service.js

@@ -81,9 +81,6 @@ import "../infra/bus/index.js";
 import { collectMediaUrisFromMessages, deleteMediaUris } from "../media/session-references.js";
 import { AgentService } from "../agent/service.js";
 import { bootstrapApplicationStateSync } from "../bootstrap/application-state.js";
-import { setPairingBroadcastSink } from "../channels/pairing/pairing-events.js";
-import { registerClarificationChannelRuntime } from "./clarify-runtime.js";
-import { ChannelManager } from "../channels/manager.js";
 import { ChatPreviewService } from "../chat-previews/service.js";
 import "../chat-previews/index.js";
 import "../config/index.js";
@@ -102,8 +99,13 @@ import { DiscussionSealer } from "../discussions/sealer.js";
 import "../discussions/index.js";
 import { EndpointToolRuntime } from "../endpoint-tools/runtime.js";
 import "../endpoint-tools/index.js";
+import { HomeCapabilityPreflightService } from "../home-intelligence/capability-preflight.js";
+import { HomeAdviceGenerator } from "../home-intelligence/generator.js";
+import { HomeIntelligenceHost } from "../home-intelligence/host.js";
+import { HomeSnapshotBuilder } from "../home-intelligence/snapshot.js";
 import { assertGatewayAuthConfigured, extractToken, resolveGatewayAuth, validateToken } from "./auth.js";
 import { prepareAutomationAgentSession } from "./automation-agent-session.js";
+import { registerClarificationChannelRuntime } from "./clarify-runtime.js";
 import { stopTailscaleExposure } from "./tailscale-lifecycle.js";
 import { getGatewayFileSpaceService } from "./file-space-service.js";
 import { reconcileHomeIntelligenceAutomation } from "./home-intelligence-automation-reconciler.js";
@@ -117,6 +119,8 @@ import { WorkDiscoveryService } from "../work-discovery/service.js";
 import { buildVoiceMemoryContext } from "../voice/realtime/memory-context.js";
 import { buildVoicePersonaContext } from "../voice/realtime/persona-context.js";
 import { voiceMemoryBudget, voicePersonaBudget } from "../voice/realtime/conversation-context.js";
+import { setPairingBroadcastSink } from "../channels/pairing/pairing-events.js";
+import { ChannelManager } from "../channels/manager.js";
 import "../notes/index.js";
 import { buildWorkflowChildTools } from "../agent/workflow/workflow-child-tools.js";
 import { WorkflowRunService } from "../workflows/service/workflow-run-service.js";
@@ -162,10 +166,6 @@ import { TaskMainUpdateDecisionService } from "../tasks/task-main-update-decisio
 import { selectTaskMainUpdateAttempt, submitAndConfirmTaskMainUpdate } from "../tasks/task-main-update-input.js";
 import { checkAppContextAccess, prepareAppContext } from "./service/app-context-access.js";
 import { applyAutomaticVoiceLanguage, inferProductLanguageFromEnvironment, initializeVoiceDefaults } from "../voice/language-profile.js";
-import { HomeAdviceGenerator } from "../home-intelligence/generator.js";
-import { HomeIntelligenceHost } from "../home-intelligence/host.js";
-import { HomeSnapshotBuilder } from "../home-intelligence/snapshot.js";
-import { HomeCapabilityPreflightService } from "../home-intelligence/capability-preflight.js";
 import crypto from "node:crypto";
 //#region src/gateway/service.ts
 init_session_repository();