fix: resolve Electron startup failures on Windows and Linux

- Try legacy port (61337) first to preserve existing users' localStorage,
  fall back to 13370 which is below the Windows Hyper-V ephemeral range (#705)
- Bind server and all URL references to 127.0.0.1 instead of localhost
  to fix IPv4/IPv6 mismatch on Linux (#684)
- Add OS-assigned port fallback (port 0) so startup never throws
- Log error codes in port checks for easier debugging
- Update localhost guards in index.ts and window-manager.ts to also
  match 127.0.0.1

Related: #705, #684

app/api/parse-url/route.ts

@@ -5,6 +5,7 @@ import { allowPrivateUrls, isPrivateUrl } from "@/lib/ssrf-protection"
 
 const MAX_CONTENT_LENGTH = 150000 // Match PDF limit
 const EXTRACT_TIMEOUT_MS = 15000
+const USER_AGENT = "Mozilla/5.0 (compatible; NextAIDrawio/1.0)"
 
 export async function POST(req: Request) {
     try {
@@ -34,6 +35,31 @@ export async function POST(req: Request) {
                 { status: 400 },
             )
         }
+        const headController = new AbortController()
+        const headTimeout = setTimeout(() => headController.abort(), 3000)
+        try {
+            const headResponse = await fetch(url, {
+                method: "HEAD",
+                headers: { "User-Agent": USER_AGENT },
+                signal: headController.signal,
+            })
+            const contentType = headResponse.headers.get("content-type")
+            if (contentType?.includes("application/pdf")) {
+                return NextResponse.json(
+                    {
+                        error: "PDF URLs are not supported. Please download and upload the PDF file directly",
+                    },
+                    { status: 422 },
+                )
+            }
+        } catch (err) {
+            console.warn(
+                "HEAD pre-check failed, proceeding with extraction:",
+                err,
+            )
+        } finally {
+            clearTimeout(headTimeout)
+        }
 
         // Extract article content with timeout to avoid tying up server resources
         const controller = new AbortController()
@@ -44,9 +70,7 @@ export async function POST(req: Request) {
         let article
         try {
             article = await extract(url, undefined, {
-                headers: {
-                    "User-Agent": "Mozilla/5.0 (compatible; NextAIDrawio/1.0)",
-                },
+                headers: { "User-Agent": USER_AGENT },
                 signal: controller.signal,
             })
         } catch (err: any) {

app/api/validate-model/route.ts

@@ -174,10 +174,21 @@ export async function POST(req: Request) {
             }
 
             case "ollama": {
-                const ollama = createOllama({
-                    baseURL: baseUrl || "http://localhost:11434",
+                // SECURITY: Mirror ai-providers.ts guard — only use server
+                // OLLAMA_API_KEY when the URL is also from server config.
+                const ollamaApiKey = baseUrl
+                    ? apiKey || undefined
+                    : apiKey || process.env.OLLAMA_API_KEY || undefined
+                const ollamaProvider = createOllama({
+                    baseURL:
+                        baseUrl ||
+                        process.env.OLLAMA_BASE_URL ||
+                        "https://ollama.com/api",
+                    ...(ollamaApiKey && {
+                        headers: { Authorization: `Bearer ${ollamaApiKey}` },
+                    }),
                 })
-                model = ollama(modelId)
+                model = ollamaProvider(modelId)
                 break
             }
 

components/model-config-dialog.tsx

@@ -282,6 +282,7 @@ export function ModelConfigDialog({
         // Check credentials based on provider type
         const isBedrock = selectedProvider.provider === "bedrock"
         const isEdgeOne = selectedProvider.provider === "edgeone"
+        const isOllama = selectedProvider.provider === "ollama"
         const isVertexAI = selectedProvider.provider === "vertexai"
         if (isBedrock) {
             if (
@@ -296,7 +297,7 @@ export function ModelConfigDialog({
             if (!selectedProvider.vertexApiKey) {
                 return
             }
-        } else if (!isEdgeOne && !selectedProvider.apiKey) {
+        } else if (!isEdgeOne && !isOllama && !selectedProvider.apiKey) {
             return
         }
 
@@ -1030,9 +1031,7 @@ export function ModelConfigDialog({
                                                     </div>
                                                 </>
                                             ) : selectedProvider.provider ===
-                                                  "ollama" ||
-                                              selectedProvider.provider ===
-                                                  "edgeone" ? (
+                                              "edgeone" ? (
                                                 <div className="space-y-3">
                                                     <div className="flex items-center gap-2">
                                                         <Button
@@ -1100,6 +1099,9 @@ export function ModelConfigDialog({
                                                                 dict.modelConfig
                                                                     .apiKey
                                                             }
+                                                            {selectedProvider.provider ===
+                                                                "ollama" &&
+                                                                ` ${dict.modelConfig.optional}`}
                                                         </Label>
                                                         <div className="flex gap-2">
                                                             <div className="relative flex-1">
@@ -1163,7 +1165,9 @@ export function ModelConfigDialog({
                                                                     handleValidate
                                                                 }
                                                                 disabled={
-                                                                    !selectedProvider.apiKey ||
+                                                                    (selectedProvider.provider !==
+                                                                        "ollama" &&
+                                                                        !selectedProvider.apiKey) ||
                                                                     validationStatus ===
                                                                         "validating"
                                                                 }

electron/main/config-manager.ts

@@ -359,9 +359,9 @@ const PROVIDER_ENV_MAP: Record<string, { apiKey: string; baseUrl: string }> = {
         baseUrl: "MODELSCOPE_BASE_URL",
     },
     gateway: { apiKey: "AI_GATEWAY_API_KEY", baseUrl: "AI_GATEWAY_BASE_URL" },
-    // bedrock and ollama don't use API keys in the same way
+    // bedrock doesn't use API keys in the same way
     bedrock: { apiKey: "", baseUrl: "" },
-    ollama: { apiKey: "", baseUrl: "OLLAMA_BASE_URL" },
+    ollama: { apiKey: "OLLAMA_API_KEY", baseUrl: "OLLAMA_BASE_URL" },
 }
 
 /**

electron/main/index.ts

@@ -94,7 +94,8 @@ if (!gotTheLock) {
             if (
                 url.includes("diagrams.net") ||
                 url.includes("draw.io") ||
-                url.startsWith("http://localhost")
+                url.startsWith("http://localhost") ||
+                url.startsWith("http://127.0.0.1")
             ) {
                 return { action: "allow" }
             }

electron/main/next-server.ts

@@ -68,7 +68,7 @@ export async function startNextServer(): Promise<string> {
     const env: Record<string, string> = {
         NODE_ENV: "production",
         PORT: String(port),
-        HOSTNAME: "localhost",
+        HOSTNAME: "127.0.0.1",
         // Enable Node.js built-in proxy support for fetch (Node.js 24+)
         NODE_USE_ENV_PROXY: "1",
     }

electron/main/port-manager.ts

@@ -9,9 +9,11 @@ import { app } from "electron"
 const PORT_CONFIG = {
     // Development mode uses fixed port for hot reload compatibility
     development: 6002,
-    // Production mode uses fixed port (61337) to preserve localStorage
-    // Falls back to sequential ports if unavailable
-    production: 61337,
+    // Legacy production port — tried first to preserve localStorage for existing users
+    legacyProduction: 61337,
+    // New production port below the ephemeral range (49152-65535)
+    // to avoid conflicts with Windows Hyper-V / ephemeral port reservations
+    production: 13370,
     // Maximum attempts to find an available port (fallback)
     maxAttempts: 100,
 }
@@ -27,7 +29,10 @@ let allocatedPort: number | null = null
 export function isPortAvailable(port: number): Promise<boolean> {
     return new Promise((resolve) => {
         const server = net.createServer()
-        server.once("error", () => resolve(false))
+        server.once("error", (err: NodeJS.ErrnoException) => {
+            console.warn(`Port ${port} unavailable: ${err.code}`)
+            resolve(false)
+        })
         server.once("listening", () => {
             server.close()
             resolve(true)
@@ -39,12 +44,12 @@ export function isPortAvailable(port: number): Promise<boolean> {
 /**
  * Find an available port
  * - In development: uses fixed port (6002)
- * - In production: uses fixed port (61337) to preserve localStorage
+ * - In production: uses fixed port (13370) to preserve localStorage
  * - Falls back to sequential ports if preferred port is unavailable
+ * - Last resort: lets the OS assign a port (port 0)
  *
  * @param reuseExisting If true, try to reuse the previously allocated port
  * @returns Promise<number> The available port
- * @throws Error if no available port found after max attempts
  */
 export async function findAvailablePort(reuseExisting = true): Promise<number> {
     const isDev = !app.isPackaged
@@ -64,7 +69,16 @@ export async function findAvailablePort(reuseExisting = true): Promise<number> {
         allocatedPort = null
     }
 
-    // Try preferred port first
+    // In production, try legacy port first to preserve existing users' localStorage
+    if (!isDev) {
+        const legacyPort = PORT_CONFIG.legacyProduction
+        if (await isPortAvailable(legacyPort)) {
+            allocatedPort = legacyPort
+            return legacyPort
+        }
+    }
+
+    // Try preferred port
     if (await isPortAvailable(preferredPort)) {
         allocatedPort = preferredPort
         return preferredPort
@@ -84,9 +98,23 @@ export async function findAvailablePort(reuseExisting = true): Promise<number> {
         }
     }
 
-    throw new Error(
-        `Failed to find available port after ${PORT_CONFIG.maxAttempts} attempts`,
+    // Last resort: let the OS pick an available port
+    console.warn(
+        "All sequential ports failed. Requesting OS-assigned port (localStorage may not persist across restarts).",
     )
+    const osPort = await new Promise<number>((resolve, reject) => {
+        const server = net.createServer()
+        server.once("error", reject)
+        server.once("listening", () => {
+            const addr = server.address()
+            const port = (addr as net.AddressInfo).port
+            server.close(() => resolve(port))
+        })
+        server.listen(0, "127.0.0.1")
+    })
+    allocatedPort = osPort
+    console.log(`OS assigned port: ${osPort}`)
+    return osPort
 }
 
 /**
@@ -113,5 +141,5 @@ export function getServerUrl(): string {
             "No port allocated yet. Call findAvailablePort() first.",
         )
     }
-    return `http://localhost:${allocatedPort}`
+    return `http://127.0.0.1:${allocatedPort}`
 }

electron/main/window-manager.ts

@@ -66,7 +66,11 @@ export function createWindow(serverUrl: string): BrowserWindow {
 
     // Handle page title updates
     mainWindow.webContents.on("page-title-updated", (event, title) => {
-        if (title && !title.includes("localhost")) {
+        if (
+            title &&
+            !title.includes("localhost") &&
+            !title.includes("127.0.0.1")
+        ) {
             mainWindow?.setTitle(title)
         } else {
             event.preventDefault()

env.example

@@ -59,8 +59,9 @@ AI_MODEL=global.anthropic.claude-sonnet-4-5-20250929-v1:0
 # AZURE_REASONING_EFFORT=low                 # Optional: Azure reasoning effort (low, medium, high)
 # AZURE_REASONING_SUMMARY=detailed
 
-# Ollama (Local) Configuration
-# OLLAMA_BASE_URL=http://localhost:11434/api  # Optional, defaults to localhost
+# Ollama Configuration (Local or Cloud)
+# OLLAMA_BASE_URL=https://ollama.com/api  # Optional, defaults to Ollama Cloud
+# OLLAMA_API_KEY=your-ollama-cloud-api-key    # Optional: For Ollama Cloud or authenticated remote instances
 # OLLAMA_ENABLE_THINKING=true                 # Optional: Enable thinking for models that support it (e.g., qwen3)
 
 # OpenRouter Configuration

lib/ai-providers.ts

@@ -596,7 +596,7 @@ function validateProviderCredentials(
  * - GOOGLE_GENERATIVE_AI_API_KEY: Google API key
  * - AZURE_RESOURCE_NAME, AZURE_API_KEY: Azure OpenAI credentials
  * - AWS_REGION, AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY: AWS Bedrock credentials
- * - OLLAMA_BASE_URL: Ollama server URL (optional, defaults to http://localhost:11434)
+ * - OLLAMA_BASE_URL: Ollama server URL (optional, defaults to https://ollama.com/api)
  * - OPENROUTER_API_KEY: OpenRouter API key
  * - DEEPSEEK_API_KEY: DeepSeek API key
  * - DEEPSEEK_BASE_URL: DeepSeek endpoint (optional)
@@ -611,13 +611,15 @@ export function getAIModel(overrides?: ClientOverrides): ModelConfig {
     // SECURITY: Prevent SSRF attacks (GHSA-9qf7-mprq-9qgm)
     // If a custom baseUrl is provided, an API key MUST also be provided.
     // This prevents attackers from redirecting server API keys to malicious endpoints.
-    // Exception: EdgeOne and Ollama providers don't require API keys
+    // Exception: EdgeOne doesn't require API keys.
+    // Ollama is exempt only when no server OLLAMA_API_KEY is configured;
+    // when it IS configured, the outer guard also enforces client apiKey for custom baseUrls.
     if (
         overrides?.baseUrl &&
         !overrides?.apiKey &&
         !(overrides?.provider === "vertexai" && overrides?.vertexApiKey) &&
         overrides?.provider !== "edgeone" &&
-        overrides?.provider !== "ollama"
+        !(overrides?.provider === "ollama" && !process.env.OLLAMA_API_KEY)
     ) {
         throw new Error(
             `API key is required when using a custom base URL. ` +
@@ -878,8 +880,19 @@ export function getAIModel(overrides?: ClientOverrides): ModelConfig {
 
         case "ollama": {
             const baseURL = overrides?.baseUrl || process.env.OLLAMA_BASE_URL
-            if (baseURL) {
-                const customOllama = createOllama({ baseURL })
+            // SECURITY: When client provides a custom base URL, only use
+            // client-provided API key. Never fall back to server OLLAMA_API_KEY
+            // to prevent leaking server credentials to user-controlled endpoints.
+            const apiKey = overrides?.baseUrl
+                ? overrides?.apiKey || undefined
+                : resolveApiKey(overrides, "OLLAMA_API_KEY")
+            if (baseURL || apiKey) {
+                const customOllama = createOllama({
+                    ...(baseURL && { baseURL }),
+                    ...(apiKey && {
+                        headers: { Authorization: `Bearer ${apiKey}` },
+                    }),
+                })
                 model = customOllama(modelId)
             } else {
                 model = ollama(modelId)
@@ -1217,7 +1230,12 @@ export function supportsImageInput(modelId: string): boolean {
     }
 
     // Qwen text models (not vision variants like qwen-vl)
-    if (lowerModelId.includes("qwen") && !hasVisionIndicator) {
+    // qwen3.5-plus is a vision model
+    if (
+        lowerModelId.includes("qwen") &&
+        !hasVisionIndicator &&
+        !lowerModelId.includes("qwen3.5-plus")
+    ) {
         return false
     }
 

lib/types/model-config.ts

@@ -104,7 +104,7 @@ export const PROVIDER_INFO: Record<
     bedrock: { label: "Amazon Bedrock" },
     ollama: {
         label: "Ollama",
-        defaultBaseUrl: "http://localhost:11434",
+        defaultBaseUrl: "https://ollama.com/api",
     },
     openrouter: {
         label: "OpenRouter",
@@ -264,6 +264,7 @@ export const SUGGESTED_MODELS: Partial<Record<ProviderName, string[]>> = {
         "Qwen/Qwen2.5-Coder-32B-Instruct",
         "Qwen/Qwen2.5-7B-Instruct",
         "Qwen/Qwen2-VL-72B-Instruct",
+        "qwen3.5-plus",
     ],
     sglang: [
         // SGLang is OpenAI-compatible, models depend on deployment
@@ -293,6 +294,7 @@ export const SUGGESTED_MODELS: Partial<Record<ProviderName, string[]>> = {
         "Qwen/Qwen3-235B-A22B-Instruct-2507",
         "Qwen/Qwen3-VL-235B-A22B-Instruct",
         "Qwen/Qwen3-32B",
+        "qwen3.5-plus",
         // DeepSeek
         "deepseek-ai/DeepSeek-R1-0528",
         "deepseek-ai/DeepSeek-V3.2",

package.json

@@ -43,7 +43,7 @@
         "@aws-sdk/client-dynamodb": "^3.957.0",
         "@aws-sdk/credential-providers": "^3.943.0",
         "@extractus/article-extractor": "^8.0.18",
-        "@formatjs/intl-localematcher": "^0.7.2",
+        "@formatjs/intl-localematcher": "^0.8.0",
         "@langfuse/client": "^4.4.9",
         "@langfuse/otel": "^4.4.4",
         "@langfuse/tracing": "^4.4.9",
@@ -51,7 +51,7 @@
         "@opennextjs/cloudflare": "1.16.1",
         "@openrouter/ai-sdk-provider": "^1.5.4",
         "@opentelemetry/api": "^1.9.0",
-        "@opentelemetry/exporter-trace-otlp-http": "^0.209.0",
+        "@opentelemetry/exporter-trace-otlp-http": "^0.211.0",
         "@opentelemetry/sdk-trace-node": "^2.2.0",
         "@radix-ui/react-alert-dialog": "^1.1.15",
         "@radix-ui/react-collapsible": "^1.1.12",
@@ -72,7 +72,7 @@
         "cmdk": "^1.1.1",
         "idb": "^8.0.3",
         "jsonrepair": "^3.13.1",
-        "lucide-react": "^0.562.0",
+        "lucide-react": "^0.563.0",
         "motion": "^12.23.25",
         "nanoid": "^5.0.0",
         "negotiator": "^1.0.0",

packages/mcp-server/package-lock.json

@@ -1,12 +1,12 @@
 {
     "name": "@next-ai-drawio/mcp-server",
-    "version": "0.1.12",
+    "version": "0.1.16",
     "lockfileVersion": 3,
     "requires": true,
     "packages": {
         "": {
             "name": "@next-ai-drawio/mcp-server",
-            "version": "0.1.12",
+            "version": "0.1.16",
             "license": "Apache-2.0",
             "dependencies": {
                 "@modelcontextprotocol/sdk": "^1.0.4",
@@ -469,9 +469,9 @@
             }
         },
         "node_modules/@hono/node-server": {
-            "version": "1.19.7",
-            "resolved": "https://registry.npmjs.org/@hono/node-server/-/node-server-1.19.7.tgz",
-            "integrity": "sha512-vUcD0uauS7EU2caukW8z5lJKtoGMokxNbJtBiwHgpqxEXokaHCBkQUmCHhjFB1VUTWdqj25QoMkMKzgjq+uhrw==",
+            "version": "1.19.9",
+            "resolved": "https://registry.npmjs.org/@hono/node-server/-/node-server-1.19.9.tgz",
+            "integrity": "sha512-vHL6w3ecZsky+8P5MD+eFfaGTyCeOHUIFYMGpQGbrBTSmNNoxv0if69rEZ5giu36weC5saFuznL411gRX7bJDw==",
             "license": "MIT",
             "engines": {
                 "node": ">=18.14.1"
@@ -481,12 +481,12 @@
             }
         },
         "node_modules/@modelcontextprotocol/sdk": {
-            "version": "1.25.2",
-            "resolved": "https://registry.npmjs.org/@modelcontextprotocol/sdk/-/sdk-1.25.2.tgz",
-            "integrity": "sha512-LZFeo4F9M5qOhC/Uc1aQSrBHxMrvxett+9KLHt7OhcExtoiRN9DKgbZffMP/nxjutWDQpfMDfP3nkHI4X9ijww==",
+            "version": "1.26.0",
+            "resolved": "https://registry.npmjs.org/@modelcontextprotocol/sdk/-/sdk-1.26.0.tgz",
+            "integrity": "sha512-Y5RmPncpiDtTXDbLKswIJzTqu2hyBKxTNsgKqKclDbhIgg1wgtf1fRuvxgTnRfcnxtvvgbIEcqUOzZrJ6iSReg==",
             "license": "MIT",
             "dependencies": {
-                "@hono/node-server": "^1.19.7",
+                "@hono/node-server": "^1.19.9",
                 "ajv": "^8.17.1",
                 "ajv-formats": "^3.0.1",
                 "content-type": "^1.0.5",
@@ -494,14 +494,15 @@
                 "cross-spawn": "^7.0.5",
                 "eventsource": "^3.0.2",
                 "eventsource-parser": "^3.0.0",
-                "express": "^5.0.1",
-                "express-rate-limit": "^7.5.0",
-                "jose": "^6.1.1",
+                "express": "^5.2.1",
+                "express-rate-limit": "^8.2.1",
+                "hono": "^4.11.4",
+                "jose": "^6.1.3",
                 "json-schema-typed": "^8.0.2",
                 "pkce-challenge": "^5.0.0",
                 "raw-body": "^3.0.0",
                 "zod": "^3.25 || ^4.0",
-                "zod-to-json-schema": "^3.25.0"
+                "zod-to-json-schema": "^3.25.1"
             },
             "engines": {
                 "node": ">=18"
@@ -520,9 +521,9 @@
             }
         },
         "node_modules/@types/node": {
-            "version": "24.10.6",
-            "resolved": "https://registry.npmjs.org/@types/node/-/node-24.10.6.tgz",
-            "integrity": "sha512-B8h60xgJMR/xmgyX9fncRzEW9gCxoJjdenUhke2v1JGOd/V66KopmWrLPXi5oUI4VuiGK+d+HlXJjDRZMj21EQ==",
+            "version": "24.10.12",
+            "resolved": "https://registry.npmjs.org/@types/node/-/node-24.10.12.tgz",
+            "integrity": "sha512-68e+T28EbdmLSTkPgs3+UacC6rzmqrcWFPQs1C8mwJhI/r5Uxr0yEuQotczNRROd1gq30NGxee+fo0rSIxpyAw==",
             "dev": true,
             "license": "MIT",
             "dependencies": {
@@ -1074,10 +1075,13 @@
             }
         },
         "node_modules/express-rate-limit": {
-            "version": "7.5.1",
-            "resolved": "https://registry.npmjs.org/express-rate-limit/-/express-rate-limit-7.5.1.tgz",
-            "integrity": "sha512-7iN8iPMDzOMHPUYllBEsQdWVB6fPDMPqwjBaFrgr4Jgr/+okjvzAy+UHlYYL/Vs0OsOrMkwS6PJDkFlJwoxUnw==",
+            "version": "8.2.1",
+            "resolved": "https://registry.npmjs.org/express-rate-limit/-/express-rate-limit-8.2.1.tgz",
+            "integrity": "sha512-PCZEIEIxqwhzw4KF0n7QF4QqruVTcF73O5kFKUnGOyjbCCgizBBiFaYpd/fnBLUMPw/BWw9OsiN7GgrNYr7j6g==",
             "license": "MIT",
+            "dependencies": {
+                "ip-address": "10.0.1"
+            },
             "engines": {
                 "node": ">= 16"
             },
@@ -1260,9 +1264,9 @@
             }
         },
         "node_modules/hono": {
-            "version": "4.11.1",
-            "resolved": "https://registry.npmjs.org/hono/-/hono-4.11.1.tgz",
-            "integrity": "sha512-KsFcH0xxHes0J4zaQgWbYwmz3UPOOskdqZmItstUG93+Wk1ePBLkLGwbP9zlmh1BFUiL8Qp+Xfu9P7feJWpGNg==",
+            "version": "4.11.9",
+            "resolved": "https://registry.npmjs.org/hono/-/hono-4.11.9.tgz",
+            "integrity": "sha512-Eaw2YTGM6WOxA6CXbckaEvslr2Ne4NFsKrvc0v97JD5awbmeBLO5w9Ho9L9kmKonrwF9RJlW6BxT1PVv/agBHQ==",
             "license": "MIT",
             "peer": true,
             "engines": {
@@ -1348,6 +1352,15 @@
             "integrity": "sha512-k/vGaX4/Yla3WzyMCvTQOXYeIHvqOKtnqBduzTHpzpQZzAskKMhZ2K+EnBiSM9zGSoIFeMpXKxa4dYeZIQqewQ==",
             "license": "ISC"
         },
+        "node_modules/ip-address": {
+            "version": "10.0.1",
+            "resolved": "https://registry.npmjs.org/ip-address/-/ip-address-10.0.1.tgz",
+            "integrity": "sha512-NWv9YLW4PoW2B7xtzaS3NCot75m6nK7Icdv0o3lfMceJVRfSoQwqD4wEH5rLwoKJwUiZ/rfpiVBhnaF0FK4HoA==",
+            "license": "MIT",
+            "engines": {
+                "node": ">= 12"
+            }
+        },
         "node_modules/ipaddr.js": {
             "version": "1.9.1",
             "resolved": "https://registry.npmjs.org/ipaddr.js/-/ipaddr.js-1.9.1.tgz",
@@ -2051,9 +2064,9 @@
             }
         },
         "node_modules/zod": {
-            "version": "4.3.5",
-            "resolved": "https://registry.npmjs.org/zod/-/zod-4.3.5.tgz",
-            "integrity": "sha512-k7Nwx6vuWx1IJ9Bjuf4Zt1PEllcwe7cls3VNzm4CQ1/hgtFUK2bRNG3rvnpPUhFjmqJKAKtjV576KnUkHocg/g==",
+            "version": "4.3.6",
+            "resolved": "https://registry.npmjs.org/zod/-/zod-4.3.6.tgz",
+            "integrity": "sha512-rftlrkhHZOcjDwkGlnUtZZkvaPHCsDATp4pGpuOOMDaTdDDXF91wuVDJoWoPsKX/3YPQ5fHuF3STjcYyKr+Qhg==",
             "license": "MIT",
             "peer": true,
             "funding": {
@@ -2061,9 +2074,9 @@
             }
         },
         "node_modules/zod-to-json-schema": {
-            "version": "3.25.0",
-            "resolved": "https://registry.npmjs.org/zod-to-json-schema/-/zod-to-json-schema-3.25.0.tgz",
-            "integrity": "sha512-HvWtU2UG41LALjajJrML6uQejQhNJx+JBO9IflpSja4R03iNWfKXrj6W2h7ljuLyc1nKS+9yDyL/9tD1U/yBnQ==",
+            "version": "3.25.1",
+            "resolved": "https://registry.npmjs.org/zod-to-json-schema/-/zod-to-json-schema-3.25.1.tgz",
+            "integrity": "sha512-pM/SU9d3YAggzi6MtR4h7ruuQlqKtad8e9S0fmxcMi+ueAK5Korys/aWcV9LIIHTVbj01NdzxcnXSN+O74ZIVA==",
             "license": "ISC",
             "peerDependencies": {
                 "zod": "^3.25 || ^4"

tests/unit/ai-providers.test.ts

@@ -1,5 +1,6 @@
-import { describe, expect, it } from "vitest"
+import { afterEach, beforeEach, describe, expect, it, vi } from "vitest"
 import {
+    getAIModel,
     resolveBaseURL,
     supportsImageInput,
     supportsPromptCaching,
@@ -189,3 +190,120 @@ describe("supportsImageInput", () => {
         expect(supportsImageInput("gemini-pro")).toBe(true)
     })
 })
+
+vi.mock("ollama-ai-provider-v2", () => {
+    const mockModel = { modelId: "test-model" }
+    const mockProviderFn = vi.fn(() => mockModel)
+    const mockCreateOllama = vi.fn(() => mockProviderFn)
+    const mockOllama = vi.fn(() => mockModel)
+    return { createOllama: mockCreateOllama, ollama: mockOllama }
+})
+
+describe("Ollama API key security", () => {
+    let createOllamaMock: ReturnType<typeof vi.fn>
+    const savedEnv: Record<string, string | undefined> = {}
+
+    beforeEach(async () => {
+        savedEnv.OLLAMA_API_KEY = process.env.OLLAMA_API_KEY
+        savedEnv.OLLAMA_BASE_URL = process.env.OLLAMA_BASE_URL
+        delete process.env.OLLAMA_BASE_URL
+
+        const mod = await import("ollama-ai-provider-v2")
+        createOllamaMock = mod.createOllama as ReturnType<typeof vi.fn>
+        createOllamaMock.mockClear()
+    })
+
+    afterEach(() => {
+        process.env.OLLAMA_API_KEY = savedEnv.OLLAMA_API_KEY
+        process.env.OLLAMA_BASE_URL = savedEnv.OLLAMA_BASE_URL
+    })
+
+    it("applies server OLLAMA_API_KEY when no client baseUrl is provided", () => {
+        process.env.OLLAMA_API_KEY = "server-secret-key"
+
+        getAIModel({ provider: "ollama", modelId: "llama2" })
+
+        expect(createOllamaMock).toHaveBeenCalledWith(
+            expect.objectContaining({
+                headers: { Authorization: "Bearer server-secret-key" },
+            }),
+        )
+    })
+
+    it("does NOT leak server OLLAMA_API_KEY when client provides a custom baseUrl", () => {
+        process.env.OLLAMA_API_KEY = "server-secret-key"
+
+        // When server has OLLAMA_API_KEY, the SSRF guard rejects
+        // client-provided baseUrl without an apiKey outright
+        expect(() =>
+            getAIModel({
+                provider: "ollama",
+                baseUrl: "https://evil-server.com",
+                modelId: "llama2",
+            }),
+        ).toThrow("API key is required")
+    })
+
+    it("uses client API key when client provides both baseUrl and apiKey", () => {
+        process.env.OLLAMA_API_KEY = "server-secret-key"
+
+        getAIModel({
+            provider: "ollama",
+            baseUrl: "https://my-ollama.com",
+            apiKey: "client-key",
+            modelId: "llama2",
+        })
+
+        expect(createOllamaMock).toHaveBeenCalledWith(
+            expect.objectContaining({
+                baseURL: "https://my-ollama.com",
+                headers: { Authorization: "Bearer client-key" },
+            }),
+        )
+    })
+
+    it("applies both server OLLAMA_BASE_URL and OLLAMA_API_KEY when no client overrides", () => {
+        process.env.OLLAMA_BASE_URL = "https://cloud.ollama.com"
+        process.env.OLLAMA_API_KEY = "server-key"
+
+        getAIModel({ provider: "ollama", modelId: "llama2" })
+
+        expect(createOllamaMock).toHaveBeenCalledWith(
+            expect.objectContaining({
+                baseURL: "https://cloud.ollama.com",
+                headers: { Authorization: "Bearer server-key" },
+            }),
+        )
+    })
+
+    it("works when OLLAMA_API_KEY is set but OLLAMA_BASE_URL is not", () => {
+        process.env.OLLAMA_API_KEY = "server-key"
+        delete process.env.OLLAMA_BASE_URL
+
+        getAIModel({ provider: "ollama", modelId: "llama2" })
+
+        expect(createOllamaMock).toHaveBeenCalledTimes(1)
+        const callArgs = createOllamaMock.mock.calls[0][0]
+        expect(callArgs).not.toHaveProperty("baseURL")
+        expect(callArgs).toEqual(
+            expect.objectContaining({
+                headers: { Authorization: "Bearer server-key" },
+            }),
+        )
+    })
+
+    it("allows client custom baseUrl without apiKey when no server OLLAMA_API_KEY", () => {
+        delete process.env.OLLAMA_API_KEY
+
+        getAIModel({
+            provider: "ollama",
+            baseUrl: "https://my-ollama.com",
+            modelId: "llama2",
+        })
+
+        expect(createOllamaMock).toHaveBeenCalledTimes(1)
+        const callArgs = createOllamaMock.mock.calls[0][0]
+        expect(callArgs.baseURL).toBe("https://my-ollama.com")
+        expect(callArgs).not.toHaveProperty("headers")
+    })
+})