@alfe.ai/gateway 0.9.13 → 0.9.14

package/dist/health.js

@@ -4533,6 +4533,7 @@ _enum(MOONSHOT_MODELS);
 _enum(QWEN_MODELS);
 string().trim().min(1);
 AnthropicModel.Opus5, AnthropicModel.Opus48, AnthropicModel.Opus47, AnthropicModel.Opus46, AnthropicModel.Sonnet5, AnthropicModel.Sonnet46, AnthropicModel.Haiku45, OpenAIModel.GPT4o, OpenAIModel.GPT4oMini, OpenAIModel.O3, OpenAIModel.GPT41, OpenAIModel.GPT41Mini, OpenAIModel.GPT41Nano, OpenAIModel.GPT54, OpenAIModel.GPT54Mini, OpenAIModel.GPT54Nano, OpenAIModel.GPT54Pro, OpenAIModel.GPT55, OpenAIModel.GPT55Pro, OpenAIModel.GPT56Sol, OpenAIModel.GPT56Terra, OpenAIModel.GPT56Luna, OpenAIModel.O3Mini, OpenAIModel.O4Mini, OpenAIModel.TextEmbedding3Small, OpenAIModel.TextEmbedding3Large, DeepSeekModel.Chat, DeepSeekModel.Reasoner, DeepSeekModel.V4Flash, DeepSeekModel.V4Pro, GoogleModel.Gemini35Flash, GoogleModel.Gemini31Pro, GoogleModel.Gemini31FlashLite, GoogleModel.Gemini25Pro, GoogleModel.Gemini25Flash, GoogleModel.Gemini25FlashLite, GoogleModel.Gemini20Flash, MiniMaxModel.M3, MiniMaxModel.M27, MiniMaxModel.M27HighSpeed, MiniMaxModel.M25, MiniMaxModel.M21, MiniMaxModel.M2, MistralModel.Large, MistralModel.Medium, MistralModel.Small, MistralModel.Codestral, MistralModel.Ministral8b, MistralModel.Ministral3b, MistralModel.MagistralMedium, MistralModel.MagistralSmall, MistralModel.DevstralMedium, XAIModel.Grok45, XAIModel.Grok43, XAIModel.Grok4, XAIModel.Grok41Fast, ZhipuModel.GLM52, ZhipuModel.GLM51, ZhipuModel.GLM46, ZhipuModel.GLM45, ZhipuModel.GLM45Air, MoonshotModel.K3, MoonshotModel.K26, MoonshotModel.K27Code, MoonshotModel.K27CodeHighSpeed, QwenModel.Qwen37Max, QwenModel.Qwen37Plus, QwenModel.Qwen36Flash, QwenModel.Qwen35Flash;
+({ Qwen3_8B: "alfe-qwen3-8b" }).Qwen3_8B;
 AnthropicModel.Opus5, AnthropicModel.Opus48, AnthropicModel.Opus47, AnthropicModel.Opus46, AnthropicModel.Sonnet5, AnthropicModel.Sonnet46, AnthropicModel.Haiku45, OpenAIModel.GPT4o, OpenAIModel.GPT4oMini, OpenAIModel.O3, OpenAIModel.GPT41, OpenAIModel.GPT41Mini, OpenAIModel.GPT41Nano, OpenAIModel.GPT54, OpenAIModel.GPT54Mini, OpenAIModel.GPT54Nano, OpenAIModel.GPT54Pro, OpenAIModel.GPT55, OpenAIModel.GPT55Pro, OpenAIModel.GPT56Sol, OpenAIModel.GPT56Terra, OpenAIModel.GPT56Luna, OpenAIModel.O3Mini, OpenAIModel.O4Mini, OpenAIModel.TextEmbedding3Small, OpenAIModel.TextEmbedding3Large, DeepSeekModel.Chat, DeepSeekModel.Reasoner, DeepSeekModel.V4Flash, DeepSeekModel.V4Pro, GoogleModel.Gemini35Flash, GoogleModel.Gemini31Pro, GoogleModel.Gemini31FlashLite, GoogleModel.Gemini25Pro, GoogleModel.Gemini25Flash, GoogleModel.Gemini25FlashLite, GoogleModel.Gemini20Flash, MiniMaxModel.M3, MiniMaxModel.M27, MiniMaxModel.M27HighSpeed, MiniMaxModel.M25, MiniMaxModel.M21, MiniMaxModel.M2, MistralModel.Large, MistralModel.Medium, MistralModel.Small, MistralModel.Codestral, MistralModel.Ministral8b, MistralModel.Ministral3b, MistralModel.MagistralMedium, MistralModel.MagistralSmall, MistralModel.DevstralMedium, XAIModel.Grok45, XAIModel.Grok43, XAIModel.Grok4, XAIModel.Grok41Fast, ZhipuModel.GLM52, ZhipuModel.GLM51, ZhipuModel.GLM46, ZhipuModel.GLM45, ZhipuModel.GLM45Air, MoonshotModel.K3, MoonshotModel.K26, MoonshotModel.K27Code, MoonshotModel.K27CodeHighSpeed, QwenModel.Qwen37Max, QwenModel.Qwen37Plus, QwenModel.Qwen36Flash, QwenModel.Qwen35Flash;
 enumValues({
 	Standard: "standard",
@@ -5397,6 +5398,24 @@ function deriveGatewayWsUrl(apiEndpoint) {
 		return "wss://gateway.alfe.ai/ws";
 	}
 }
+function validatePrivateManagedWsUrl(value, service) {
+	let url;
+	try {
+		url = new URL(value);
+	} catch {
+		throw new Error(`${service} WebSocket URL must be a valid regional service URL`);
+	}
+	if (url.protocol !== "ws:" || !url.hostname.endsWith(".local") && !url.hostname.endsWith(".sst") || url.pathname !== "/ws" || url.username !== "" || url.password !== "" || url.search !== "" || url.hash !== "") throw new Error(`${service} WebSocket URL must be a canonical VPC-local regional service URL`);
+	return url.toString();
+}
+function _resolveManagedGatewayWsUrl(apiEndpoint) {
+	const override = process.env.ALFE_GATEWAY_WS_URL?.trim();
+	if (process.env.ALFE_PRIVATE_DATA_PLANE === "true") {
+		if (!override) throw new Error("ALFE_GATEWAY_WS_URL is required in AWS-private managed mode");
+		return validatePrivateManagedWsUrl(override, "Gateway");
+	}
+	return override && override.length > 0 ? override : deriveGatewayWsUrl(apiEndpoint);
+}
 /**
 * Parse [runtimes.*] sections from config.toml.
 * Returns a map of runtime name → RuntimeConfig.
@@ -5471,7 +5490,7 @@ async function loadManagedConfig() {
 	const apiEndpoint = process.env.ALFE_API_ENDPOINT ?? getEndpointFromToken(apiKey);
 	logger$1.debug({ apiEndpoint }, "Environment loaded, resolving identity...");
 	const identity = await resolveAgentIdentity(apiKey, apiEndpoint);
-	const gatewayWsUrl = deriveGatewayWsUrl(apiEndpoint);
+	const gatewayWsUrl = _resolveManagedGatewayWsUrl(apiEndpoint);
 	logger$1.debug({
 		agentId: identity.agentId,
 		orgId: identity.orgId,
@@ -6950,6 +6969,15 @@ const SYSTEMD_SERVICE = "alfe-gateway";
 const MANAGED_ENV_VARS = [
 	"ALFE_MANAGED",
 	"ALFE_API_KEY",
+	"ALFE_API_ENDPOINT",
+	"ALFE_API_URL",
+	"ALFE_AI_PROXY_URL",
+	"ALFE_CHAT_WS_URL",
+	"ALFE_GATEWAY_WS_URL",
+	"ALFE_SYNC_RELAY_WS_URL",
+	"ALFE_PRIVATE_DATA_PLANE",
+	"ALFE_PRIVATE_DATA_PLANE_REGION",
+	"ALFE_PRIVATE_DATA_PLANE_RELEASE",
 	"LOG_LEVEL",
 	"ALFE_CLI_VERSION"
 ];
@@ -7992,12 +8020,18 @@ async function startAiProxy(apiKey) {
 	try {
 		const { createProxyServer, DEFAULT_AI_PROXY_PORT } = await import("@alfe.ai/ai-proxy-local");
 		const { getAiServiceUrlFromToken } = await import("@alfe.ai/config");
-		const proxyUrl = getAiServiceUrlFromToken(apiKey);
+		const privateDataPlane = process.env.ALFE_PRIVATE_DATA_PLANE === "true";
+		const proxyUrlOverride = process.env.ALFE_AI_PROXY_URL?.trim();
+		const configuredProxyUrl = proxyUrlOverride && proxyUrlOverride.length > 0 ? proxyUrlOverride : void 0;
+		if (privateDataPlane && configuredProxyUrl === void 0) throw new Error("AWS-private AI proxy target is not configured");
+		const proxyUrl = configuredProxyUrl ?? getAiServiceUrlFromToken(apiKey);
 		const port = DEFAULT_AI_PROXY_PORT ?? 18193;
 		handle.server = createProxyServer({
 			port,
 			apiKey,
-			proxyUrl
+			proxyUrl,
+			allowPrivateAwsTarget: privateDataPlane,
+			requirePrivateAwsTarget: privateDataPlane
 		});
 		const server = handle.server;
 		await new Promise((resolve, reject) => {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@alfe.ai/gateway",
-  "version": "0.9.13",
+  "version": "0.9.14",
   "description": "Alfe local gateway daemon — persistent control plane for agent integrations",
   "type": "module",
   "bin": {
@@ -23,8 +23,8 @@
     "pino-roll": "^1.2.0",
     "smol-toml": ">=1.6.1",
     "ws": "^8.18.0",
-    "@alfe.ai/agent-api-client": "^0.17.0",
-    "@alfe.ai/ai-proxy-local": "^0.0.16",
+    "@alfe.ai/agent-api-client": "^0.17.1",
+    "@alfe.ai/ai-proxy-local": "^0.0.17",
     "@alfe.ai/config": "^0.4.1",
     "@alfe.ai/integration-manifest": "^0.4.0",
     "@alfe.ai/integrations": "^0.6.4",