@juspay/neurolink 12.47.1 → 12.47.3

package/dist/providers/anthropic/client.d.ts

@@ -11,6 +11,15 @@ export declare class AnthropicProvider extends BaseProvider {
     private readonly authMethod;
     private readonly subscriptionTier;
     private readonly enableBetaFeatures;
+    /**
+     * Where requests go when not to `api.anthropic.com`: the credentials'
+     * `baseURL`, else the config's, else `ANTHROPIC_BASE_URL` — normalized once
+     * (no trailing slash, no `/vN` suffix: the SDK appends `/v1` itself).
+     * Undefined means the vendor's own endpoint; every "proxy in use" decision
+     * reads this, never the environment directly, so a per-instance gateway is
+     * honoured the same way the environment one always was.
+     */
+    private readonly baseURL;
     private oauthToken;
     private lastResponseMetadata;
     private usageInfo;
@@ -25,6 +34,7 @@ export declare class AnthropicProvider extends BaseProvider {
     constructor(modelName?: string, sdk?: unknown, config?: AnthropicProviderConfig, credentials?: {
         apiKey?: string;
         oauthToken?: string;
+        baseURL?: string;
     });
     /**
      * Get authentication headers based on current auth method and configuration.

package/dist/providers/anthropic/client.js

@@ -58,6 +58,31 @@ const getAnthropicApiKey = () => {
 const getDefaultAnthropicModel = () => {
     return getProviderModel("ANTHROPIC_MODEL", AnthropicModels.CLAUDE_SONNET_4_6);
 };
+/**
+ * The official Anthropic SDK builds `${baseURL}/v1/messages` itself, so a
+ * version-suffixed base URL — the form the previous @ai-sdk/anthropic
+ * implementation REQUIRED (`https://api.anthropic.com/v1`) — would double up
+ * as `/v1/v1/messages`. Normalize the inverse way: strip trailing slashes and
+ * a trailing `/vN` segment, so both historical forms keep working whether the
+ * URL comes from the credentials, the config or `ANTHROPIC_BASE_URL`. Blank
+ * means "the vendor's endpoint".
+ */
+function normalizeAnthropicBaseURL(raw) {
+    const value = raw?.trim();
+    if (!value) {
+        return undefined;
+    }
+    const trimmed = value.replace(/\/+$/, "");
+    const stripped = trimmed.replace(/\/v\d+$/, "");
+    if (stripped !== trimmed) {
+        logger.debug("[AnthropicProvider] Stripping the version suffix from the base URL — " +
+            "the official Anthropic SDK appends /v1 to the base URL itself.", {
+            baseURL: redactUrlCredentials(value),
+            rewrittenTo: redactUrlCredentials(stripped),
+        });
+    }
+    return stripped;
+}
 const streamTracer = trace.getTracer("neurolink.provider.anthropic");
 /**
  * Get OAuth token from stored credentials file or environment.
@@ -476,6 +501,15 @@ export class AnthropicProvider extends BaseProvider {
     authMethod;
     subscriptionTier;
     enableBetaFeatures;
+    /**
+     * Where requests go when not to `api.anthropic.com`: the credentials'
+     * `baseURL`, else the config's, else `ANTHROPIC_BASE_URL` — normalized once
+     * (no trailing slash, no `/vN` suffix: the SDK appends `/v1` itself).
+     * Undefined means the vendor's own endpoint; every "proxy in use" decision
+     * reads this, never the environment directly, so a per-instance gateway is
+     * honoured the same way the environment one always was.
+     */
+    baseURL;
     oauthToken;
     lastResponseMetadata = null;
     usageInfo = null;
@@ -511,12 +545,16 @@ export class AnthropicProvider extends BaseProvider {
         const subscriptionTier = config?.subscriptionTier ??
             (authMethod === "oauth" ? detectSubscriptionTier(oauthToken) : "api");
         const targetModel = modelName || getDefaultAnthropicModel();
+        // Resolved before super(): the tier check below needs it, and the
+        // credentials win over the config, which wins over the environment —
+        // the same precedence `apiKey` has.
+        const baseURL = normalizeAnthropicBaseURL(credentials?.baseURL ?? config?.baseURL ?? process.env.ANTHROPIC_BASE_URL);
         // Determine effective model based on tier access.
-        // Skip tier validation when a proxy is in use (ANTHROPIC_BASE_URL is set)
-        // — the proxy handles model access and auth, so the SDK should pass
-        // the requested model through without downgrading.
+        // Skip tier validation when a proxy is in use (a base URL is set) — the
+        // proxy handles model access and auth, so the SDK should pass the
+        // requested model through without downgrading.
         let effectiveModel = targetModel;
-        const usingProxy = !!process.env.ANTHROPIC_BASE_URL;
+        const usingProxy = baseURL !== undefined;
         if (!usingProxy &&
             subscriptionTier !== "api" &&
             !isModelAvailableForTier(targetModel, subscriptionTier)) {
@@ -533,6 +571,7 @@ export class AnthropicProvider extends BaseProvider {
         // Store computed values
         this.oauthToken = oauthToken;
         this.subscriptionTier = subscriptionTier;
+        this.baseURL = baseURL;
         // Use the auth method already resolved above (before tier computation)
         this.authMethod = authMethod;
         // Build headers based on auth method and subscription tier
@@ -570,6 +609,9 @@ export class AnthropicProvider extends BaseProvider {
             // The claude-code-20250219 beta header triggers "credential only for Claude Code" error
             client = new Anthropic({
                 apiKey: "oauth-authenticated", // Placeholder, actual auth is in fetch wrapper
+                // The same gateway the API-key branch honours: without it, OAuth
+                // credentials that also name a base URL would still reach the vendor.
+                ...(this.baseURL && { baseURL: this.baseURL }),
                 // Note: No headers passed - fetch wrapper sets oauth-2025-04-20 beta header
                 // Limit capture wraps the OAuth fetch so subscription quota headers
                 // (anthropic-ratelimit-unified-*) are recorded on every request —
@@ -599,33 +641,10 @@ export class AnthropicProvider extends BaseProvider {
         else {
             // Traditional API key authentication
             const apiKeyToUse = credentials?.apiKey ?? config?.apiKey ?? getAnthropicApiKey();
-            // The official Anthropic SDK builds `${baseURL}/v1/messages` itself, so
-            // a version-suffixed base URL — the form the previous @ai-sdk/anthropic
-            // implementation REQUIRED (`https://api.anthropic.com/v1`) — would
-            // double up as `/v1/v1/messages`. Normalize the inverse way now: strip
-            // a trailing `/vN` segment when present so both historical forms of
-            // ANTHROPIC_BASE_URL keep working.
-            const normalizedBaseURL = (() => {
-                const raw = process.env.ANTHROPIC_BASE_URL;
-                if (!raw) {
-                    return undefined;
-                }
-                const trimmed = raw.replace(/\/+$/, "");
-                const stripped = trimmed.replace(/\/v\d+$/, "");
-                if (stripped !== trimmed) {
-                    logger.debug("[AnthropicProvider] Stripping the version suffix from " +
-                        "ANTHROPIC_BASE_URL — the official Anthropic SDK appends /v1 " +
-                        "to the base URL itself.", {
-                        baseURL: redactUrlCredentials(raw),
-                        rewrittenTo: redactUrlCredentials(stripped),
-                    });
-                }
-                return stripped;
-            })();
             client = new Anthropic({
                 apiKey: apiKeyToUse,
                 defaultHeaders: headers,
-                ...(normalizedBaseURL && { baseURL: normalizedBaseURL }),
+                ...(this.baseURL && { baseURL: this.baseURL }),
                 // Same capture as the OAuth branch: works for direct API-key traffic
                 // (legacy requests/tokens counters) and for the NeuroLink Claude proxy
                 // (verbatim unified quota plus x-neurolink-* account/pool state).
@@ -676,10 +695,11 @@ export class AnthropicProvider extends BaseProvider {
      */
     getAuthHeaders() {
         const headers = {};
-        // When routing through proxy (ANTHROPIC_BASE_URL set), use the full
-        // OAuth beta set so the proxy forwards them upstream. Without these,
-        // Anthropic treats the request with tighter non-subscription rate limits.
-        const usingProxy = !!process.env.ANTHROPIC_BASE_URL;
+        // When routing through a proxy (a base URL is set — per instance or
+        // ANTHROPIC_BASE_URL), use the full OAuth beta set so the proxy forwards
+        // them upstream. Without these, Anthropic treats the request with tighter
+        // non-subscription rate limits.
+        const usingProxy = this.baseURL !== undefined;
         if (this.enableBetaFeatures) {
             if (usingProxy) {
                 // The 1M-context beta requires a plan upgrade on most accounts;
@@ -705,9 +725,9 @@ export class AnthropicProvider extends BaseProvider {
             }
         }
         if (usingProxy) {
-            // WAFs in front of ANTHROPIC_BASE_URL proxies commonly block the bare
-            // SDK UA ("Anthropic/JS x.y.z"); send the claude-cli UA the OAuth path
-            // already uses. Direct-to-Anthropic traffic keeps the honest SDK UA.
+            // WAFs in front of Anthropic proxies commonly block the bare SDK UA
+            // ("Anthropic/JS x.y.z"); send the claude-cli UA the OAuth path already
+            // uses. Direct-to-Anthropic traffic keeps the honest SDK UA.
             headers["User-Agent"] = CLAUDE_CLI_USER_AGENT;
         }
         // Add subscription-specific headers if applicable
@@ -736,8 +756,8 @@ export class AnthropicProvider extends BaseProvider {
         // Proxy mode: bypass tier validation entirely — the proxy handles model
         // access. Log at debug level so users can tell why an unknown model name
         // "validated" when their proxy may not actually expose it.
-        if (process.env.ANTHROPIC_BASE_URL) {
-            logger.debug("[validateModelAccess] Bypassing tier check (ANTHROPIC_BASE_URL set — proxy enforces access)", { model });
+        if (this.baseURL !== undefined) {
+            logger.debug("[validateModelAccess] Bypassing tier check (a base URL is set — the proxy enforces access)", { model });
             return true;
         }
         // API tier has access to all models

package/dist/providers/googleAiStudio/client.js

@@ -21,7 +21,7 @@ import { resolveToolExecutionRecords } from "../../core/toolExecutionRecorder.js
 import { buildDedupedEngineTools, buildGeminiResponseSchema, buildLoopExitMessage, buildNativeConfig, buildWrapupNudgeText, computeMaxSteps, createContextGuard, createTurnClock, buildUserPartsWithMultimodal, extractThoughtSignature, geminiContentsToV3Prompt, handleMaxStepsTermination, mapGeminiFinishReason, prependConversationMessages, resolveTurnStopReason, v3PromptToGeminiContents, } from "../googleNativeGemini3/index.js";
 import { createStreamChannel } from "../../core/streamChannel.js";
 import { toNativeToolDeclarations } from "../../core/nativeToolFormat.js";
-import { warnGoogleSdkIgnoresProxy } from "../../proxy/proxyFetch.js";
+import { googleSdkProxyHttpOptions } from "../../proxy/proxyFetch.js";
 // ── Model middleware stream/generate bridges ──
 //
 // Caller model middleware (transformParams / wrapGenerate / wrapStream) is
@@ -145,20 +145,19 @@ async function createGoogleGenAIClient(apiKey, baseURL) {
         });
     }
     const Ctor = ctor;
-    // httpOptions carries the endpoint override and nothing else. It used to
-    // also pass a proxy fetch, which the SDK silently ignored — see
-    // warnGoogleSdkIgnoresProxy for why that is not fixable here.
+    // httpOptions carries the endpoint override and, when a proxy is configured,
+    // the proxy-aware fetch the SDK sends its requests through.
     //
     // baseUrl is only included when resolved — verified against
     // @google/genai's ApiClient (dist/node/index.cjs) that it falls back to
     // its own default whenever httpOptions.baseUrl is undefined, so omitting
     // the key and passing `baseUrl: undefined` behave identically; the key is
     // still omitted outright for a cleaner outbound config object.
-    warnGoogleSdkIgnoresProxy("GoogleAIStudio");
     return new Ctor({
         apiKey,
         httpOptions: {
             ...(baseURL ? { baseUrl: baseURL } : {}),
+            ...googleSdkProxyHttpOptions(),
         },
     });
 }
@@ -2202,7 +2201,17 @@ export class GoogleAIStudioProvider extends BaseProvider {
             }
             queue.push(item);
         };
-        const session = await client.live.connect({
+        // @google/genai 2.x waits inside connect() for the server's setupComplete
+        // message and does not reject when the socket errors or closes first, so a
+        // refused key or model would leave this await pending forever. Fail on the
+        // first error or close that arrives before the session is up. Once it is
+        // up, the callbacks below end the stream as they always did.
+        let sessionOpen = false;
+        let failConnect = () => undefined;
+        const connectFailed = new Promise((_resolve, reject) => {
+            failConnect = reject;
+        });
+        const connecting = client.live.connect({
             model,
             callbacks: {
                 onopen: () => {
@@ -2230,9 +2239,15 @@ export class GoogleAIStudioProvider extends BaseProvider {
                     }
                 },
                 onerror: (e) => {
+                    if (!sessionOpen) {
+                        failConnect(new Error("Gemini Live connection failed before setup completed"));
+                    }
                     push({ type: "error", error: e });
                 },
                 onclose: (_e) => {
+                    if (!sessionOpen) {
+                        failConnect(new Error("Gemini Live connection closed before setup completed"));
+                    }
                     push({ type: "end" });
                 },
             },
@@ -2243,6 +2258,8 @@ export class GoogleAIStudioProvider extends BaseProvider {
                 },
             },
         });
+        const session = await Promise.race([connecting, connectFailed]);
+        sessionOpen = true;
         // Feed upstream audio frames concurrently
         (async () => {
             try {
@@ -2411,11 +2428,35 @@ export class GoogleAIStudioProvider extends BaseProvider {
         try {
             const apiKey = this.getApiKey();
             const client = await createGoogleGenAIClient(apiKey, this.getBaseURL());
-            const result = await client.models.embedContent({
-                model: embeddingModelName,
-                contents: texts,
-            });
-            const embeddings = (result.embeddings || []).map((e) => e.values || []);
+            const embeddings = [];
+            if (embeddingModelName.includes("gemini-embedding-2")) {
+                // @google/genai 2.x reads an array of strings for these models as ONE
+                // content with several parts, so a batch would come back as a single
+                // vector. Embed each text on its own, a few at a time, in order.
+                const concurrency = 8;
+                for (let start = 0; start < texts.length; start += concurrency) {
+                    const group = texts.slice(start, start + concurrency);
+                    const results = await Promise.all(group.map((text) => client.models.embedContent({
+                        model: embeddingModelName,
+                        contents: [text],
+                    })));
+                    for (const result of results) {
+                        embeddings.push(result.embeddings?.[0]?.values || []);
+                    }
+                }
+            }
+            else {
+                const result = await client.models.embedContent({
+                    model: embeddingModelName,
+                    contents: texts,
+                });
+                embeddings.push(...(result.embeddings || []).map((e) => e.values || []));
+            }
+            // One vector per text, or fail: a silent mismatch would pair texts with
+            // the wrong vectors downstream.
+            if (embeddings.length !== texts.length) {
+                throw new Error(`Embedding response held ${embeddings.length} vectors for ${texts.length} texts`);
+            }
             logger.debug("Batch embeddings generated successfully", {
                 provider: this.providerName,
                 model: embeddingModelName,

package/dist/providers/googleVertex/client.js

@@ -13,7 +13,7 @@ import { resolveRequestKind } from "../../core/resolveRequestKind.js";
 import { ModelConfigurationManager } from "../../core/modelConfiguration.js";
 import { isSchemaComplexityError } from "../../core/modules/structuredOutputPolicy.js";
 import { redactUrlForError, stringifyContentSafe, } from "../../utils/logSanitize.js";
-import { warnGoogleSdkIgnoresProxy } from "../../proxy/proxyFetch.js";
+import { googleSdkProxyHttpOptions } from "../../proxy/proxyFetch.js";
 import { AuthenticationError, InvalidModelError, NetworkError, ProviderError, RateLimitError, } from "../../types/index.js";
 import { classifyProviderError } from "../../utils/errorClassifier.js";
 import { ERROR_CODES, NeuroLinkError } from "../../utils/errorHandling.js";
@@ -1122,7 +1122,6 @@ export class GoogleVertexProvider extends BaseProvider {
      * Create @google/genai client configured for Vertex AI
      */
     async createVertexGenAIClient(regionOverride) {
-        warnGoogleSdkIgnoresProxy("GoogleVertex");
         const expressApiKey = this.resolveExpressApiKey();
         // Resolved only on the ADC path, and from the per-instance projectId the
         // constructor already settled (credentials.projectId if supplied, else
@@ -1147,14 +1146,14 @@ export class GoogleVertexProvider extends BaseProvider {
         const Ctor = ctor;
         const baseUrl = this.resolveBaseURL();
         const httpOptions = {
-            // The endpoint override and nothing else. This object used to also pass
-            // a proxy fetch, which the SDK silently ignored — see
-            // warnGoogleSdkIgnoresProxy for why that is not fixable here.
+            // The endpoint override and, when a proxy is configured, the
+            // proxy-aware fetch the SDK sends its requests through.
             //
-            // Only set when resolved: the SDK falls back to its own default
-            // whenever httpOptions.baseUrl is undefined, so omitting the key and
-            // passing undefined behave identically.
+            // baseUrl is only set when resolved: the SDK falls back to its own
+            // default whenever httpOptions.baseUrl is undefined, so omitting the key
+            // and passing undefined behave identically.
             ...(baseUrl ? { baseUrl } : {}),
+            ...googleSdkProxyHttpOptions(),
         };
         if (expressApiKey) {
             // Express Mode: an API key replaces project/location entirely. Passing

package/dist/proxy/proxyFetch.d.ts

@@ -3,6 +3,7 @@
  * Supports HTTP/HTTPS, SOCKS4/5, authentication, and NO_PROXY bypass
  * Lightweight implementation extracted from research of major proxy packages
  */
+import type { GoogleGenAIHttpOptions } from "../types/index.js";
 /**
  * Classify a fetch failure as a transient network error worth retrying.
  *
@@ -47,4 +48,18 @@ export declare function getProxyStatus(): {
     method: string;
     capabilities: string[];
 };
-export declare function warnGoogleSdkIgnoresProxy(providerLabel: string): void;
+/**
+ * `httpOptions` fragment that routes the @google/genai SDK through the
+ * configured proxy, or nothing when none is configured.
+ *
+ * The SDK calls global `fetch` unless `HttpOptions.fetch` is set (declared and
+ * used from 2.23.0), and global `fetch` does not read HTTP_PROXY / HTTPS_PROXY
+ * unless Node itself was started with env-proxy support. The proxy-aware fetch
+ * is passed only when a proxy is configured: with none, the SDK keeps its own
+ * request path instead of gaining this module's trace headers and retries.
+ *
+ * Covers the SDK's `generateContent`, `generateContentStream` and
+ * `embedContent` requests. It does not reach Gemini Live websockets or the
+ * Application Default Credentials token requests.
+ */
+export declare function googleSdkProxyHttpOptions(): GoogleGenAIHttpOptions;

package/dist/proxy/proxyFetch.js

@@ -698,31 +698,19 @@ export function getProxyStatus() {
     };
 }
 /**
- * One-time warning that the @google/genai SDK cannot honour a configured
- * proxy.
+ * `httpOptions` fragment that routes the @google/genai SDK through the
+ * configured proxy, or nothing when none is configured.
  *
- * Both Google providers passed `httpOptions: { fetch: createProxyFetch() }`,
- * which does nothing. `HttpOptions` in @google/genai 1.46.0 declares only
- * baseUrl, baseUrlResourceScope, apiVersion, headers, timeout, extraBody and
- * retryOptions — there is no `fetch` on it. That property belongs to a
- * different interface (`ClientOptions`), which `GoogleGenAIOptions` does not
- * accept, and the SDK's request path calls global `fetch`. The option was
- * silently dropped and requests went direct.
+ * The SDK calls global `fetch` unless `HttpOptions.fetch` is set (declared and
+ * used from 2.23.0), and global `fetch` does not read HTTP_PROXY / HTTPS_PROXY
+ * unless Node itself was started with env-proxy support. The proxy-aware fetch
+ * is passed only when a proxy is configured: with none, the SDK keeps its own
+ * request path instead of gaining this module's trace headers and retries.
  *
- * It type-checked only because those constructors are reached through a
- * loosely-typed local alias, which turns off excess-property checking.
- *
- * This SDK version offers no supported injection point, so rather than keep a
- * line that reads like working proxy support, the situation is reported once
- * per process — and only to someone who actually configured a proxy. Silent
- * bypass is the worst outcome available: a corporate user believes their
- * traffic is proxied when it is not.
+ * Covers the SDK's `generateContent`, `generateContentStream` and
+ * `embedContent` requests. It does not reach Gemini Live websockets or the
+ * Application Default Credentials token requests.
  */
-let proxyUnsupportedWarned = false;
-export function warnGoogleSdkIgnoresProxy(providerLabel) {
-    if (proxyUnsupportedWarned || !getProxyStatus().enabled) {
-        return;
-    }
-    proxyUnsupportedWarned = true;
-    logger.warn(`[${providerLabel}] A proxy is configured, but the @google/genai SDK provides no way to route its requests through it (HttpOptions has no 'fetch', and GoogleGenAIOptions accepts none). Requests from this provider go direct.`);
+export function googleSdkProxyHttpOptions() {
+    return getProxyStatus().enabled ? { fetch: createProxyFetch() } : {};
 }

package/dist/types/providers.d.ts

@@ -105,9 +105,16 @@ export type NeurolinkCredentials = {
         apiKey?: string;
         baseURL?: string;
     };
+    /**
+     * Anthropic. `baseURL` points the official SDK client at a gateway or proxy
+     * instead of `api.anthropic.com` (with or without a trailing `/v1`); it
+     * takes precedence over `ANTHROPIC_BASE_URL`, exactly as `apiKey` does over
+     * `ANTHROPIC_API_KEY`.
+     */
     anthropic?: {
         apiKey?: string;
         oauthToken?: string;
+        baseURL?: string;
     };
     googleAiStudio?: {
         apiKey?: string;

package/docs-site/static/search-index.json

@@ -4131,15 +4131,15 @@
 {"objectID":"1d8625a67cba7bc640e454fe15f76d86e2be4fbe79e9496fc085e5a6df496086","title":"Future Enhancements","url":"/docs/features/pdf-support#future-enhancements","content":"Planned features for PDF support:\nOCR Integration: Extract text from scanned PDFs\nPage Selection: Analyze specific pages only\nPDF Generation: Create PDFs from AI responses\nForm Filling: Extract and populate PDF forms","hierarchy":{"lvl0":"Features","lvl1":"PDF File Support","lvl2":"Future Enhancements","lvl3":""}},
 {"objectID":"20bb5119cb74199a04bebd17c621d8c6a0438c0833f12103b9f3914c5f026ceb","title":"Feedback and Support","url":"/docs/features/pdf-support#feedback-and-support","content":"Found a bug or have a feature request? Please:\nCheck existing issues on GitHub\nCreate a new issue with:\nProvider used\nPDF file details (size, pages)\nError message or unexpected behavior\nSample code (if possible)","hierarchy":{"lvl0":"Features","lvl1":"PDF File Support","lvl2":"Feedback and Support","lvl3":""}},
 {"objectID":"7c47c75b9023d438078ad3ed696c167cb9b5beaa369fad844635f7f32cce74a5","title":"Version 9.2.0 (Current)","url":"/docs/features/pdf-support#version-920-current","content":"✅ Initial PDF support for Vertex AI, Anthropic, Bedrock, AI Studio\n✅ Auto-detection via --file flag\n✅ Multiple PDF processing\n✅ Size and page limit validation\n✅ Comprehensive error messages\n✅ CLI and SDK integration\n✅ Streaming support\n✅ Mixed multimodal inputs (PDF + CSV + images)\n\nNext: Multimodal Chat Guide | CSV Support","hierarchy":{"lvl0":"Features","lvl1":"PDF File Support","lvl2":"Version 9.2.0 (Current)","lvl3":""}},
-{"objectID":"2a46dad649fcc5c7b389d9db519211899183703302b4127f0a4a2285018e1362","title":"Per-Request Credentials","url":"/docs/features/per-request-credentials","content":"Per-Request Credentials\n\nStatus: Stable | Availability: SDK only\n\nOverview\n\nNeuroLink allows provider credentials to be supplied at two levels below the environment-variable default: on the NeuroLink constructor (instance level) and on individual generate() / stream() calls (per-call level). Credentials are resolved in the following order of precedence:\n\nThis enables multi-tenant architectures where different users or tenants supply their own provider API keys (bring-your-own-key, BYOK), without requiring separate NeuroLink instances per user or touching the process environment.\n\nTypical use cases:\nMulti-tenant SaaS — each API request carries the calling user's provider key; no shared key leakage between tenants\nBYOK products — end users paste their OpenAI / Anthropic keys in settings; you forward them to NeuroLink per call\nTesting and CI — inject credentials programmatically without setting environment variables\nProvider switching — override only the active provider's credentials while leaving others to fall through to env vars\n\nQuick Start\n\nInstance-Level vs Per-Call\n\nInstance-Level Credentials\n\nSet credentials in the NeuroLink constructor. These apply as the default for every generate() and stream() call made on that instance. Useful when serving a single tenant or when you have a known key for the duration of the instance lifecycle.\n\nPer-Call Credentials\n\nSet credentials directly on generate() or stream(). These override the instance-level credentials for that single call only. Only the providers you explicitly set are overridden — others continue falling through to instance credentials and then environment variables.\n\nPrecedence Rules\n\n| Level       | Scope                 | Set on                                                                              |\n| ----------- | --------------------- | ----------------------------------------------------------------------------------- |\n| Per-call    | Single request only   | generate({ credentials }), stream({ credentials }) or decide({ credentials }) |\n| Instance    | All calls on instance | new NeuroLink({ credentials })                                                    |\n| Environment | Process-wide fallback | OPENAI_API_KEY, ANTHROPIC_API_KEY, …                                            |\n\nUnset providers at any level fall through to the next. You never need to repeat a credential at the per-call level if the instance default is correct.\n\nProvider Credential Reference\n\nAll fields are optional — omit any field you want to fall through to a lower-precedence level.\n\n| Provider             | Key                 | Fields                                                                                               |\n| -------------------- | ------------------- | ---------------------------------------------------------------------------------------------------- |\n| OpenAI               | openai            | apiKey, baseURL                                                                                  |\n| Anthropic            | anthropic         | apiKey, oauthToken                                                                               |\n| Google AI Studio     | googleAiStudio    | apiKey, baseURL                                                                                  |\n| Google Vertex AI     | vertex            | projectId, location, apiKey (Express Mode), serviceAccountKey, clientEmail, privateKey   |\n| Amazon Bedrock       | bedrock           | accessKeyId, secretAccessKey, sessionToken, region                                           |\n| Amazon SageMaker     | sagemaker         | accessKeyId, secretAccessKey, sessionToken, region, endpoint                               |\n| Azure OpenAI         | azure             | apiKey, resourceName, deploymentName, apiVersion                                             |\n| Mistral              | mistral           | apiKey                                                                                             |\n| Hugging Face         | huggingFace       | apiKey, baseURL                                                                                  |\n| OpenRouter           | openrouter        | apiKey, baseURL                                                                                  |\n| LiteLLM              | litellm           | apiKey, baseURL                                                                                  |\n| OpenAI-Compatible    | openaiCompatible  | apiKey, baseURL                                                                                  |\n| Cerebras             | cerebras          | apiKey, baseURL                                                                                  |\n| SambaNova            | sambanova         | apiKey, baseURL                                                                                  |\n| Ollama               | ollama            | baseURL                                                                                     ","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"","lvl3":""}},
+{"objectID":"2a46dad649fcc5c7b389d9db519211899183703302b4127f0a4a2285018e1362","title":"Per-Request Credentials","url":"/docs/features/per-request-credentials","content":"Per-Request Credentials\n\nStatus: Stable | Availability: SDK only\n\nOverview\n\nNeuroLink allows provider credentials to be supplied at two levels below the environment-variable default: on the NeuroLink constructor (instance level) and on individual generate() / stream() calls (per-call level). Credentials are resolved in the following order of precedence:\n\nThis enables multi-tenant architectures where different users or tenants supply their own provider API keys (bring-your-own-key, BYOK), without requiring separate NeuroLink instances per user or touching the process environment.\n\nTypical use cases:\nMulti-tenant SaaS — each API request carries the calling user's provider key; no shared key leakage between tenants\nBYOK products — end users paste their OpenAI / Anthropic keys in settings; you forward them to NeuroLink per call\nTesting and CI — inject credentials programmatically without setting environment variables\nProvider switching — override only the active provider's credentials while leaving others to fall through to env vars\n\nQuick Start\n\nInstance-Level vs Per-Call\n\nInstance-Level Credentials\n\nSet credentials in the NeuroLink constructor. These apply as the default for every generate() and stream() call made on that instance. Useful when serving a single tenant or when you have a known key for the duration of the instance lifecycle.\n\nPer-Call Credentials\n\nSet credentials directly on generate() or stream(). These override the instance-level credentials for that single call only. Only the providers you explicitly set are overridden — others continue falling through to instance credentials and then environment variables.\n\nPrecedence Rules\n\n| Level       | Scope                 | Set on                                                                              |\n| ----------- | --------------------- | ----------------------------------------------------------------------------------- |\n| Per-call    | Single request only   | generate({ credentials }), stream({ credentials }) or decide({ credentials }) |\n| Instance    | All calls on instance | new NeuroLink({ credentials })                                                    |\n| Environment | Process-wide fallback | OPENAI_API_KEY, ANTHROPIC_API_KEY, …                                            |\n\nUnset providers at any level fall through to the next. You never need to repeat a credential at the per-call level if the instance default is correct.\n\nProvider Credential Reference\n\nAll fields are optional — omit any field you want to fall through to a lower-precedence level.\n\n| Provider             | Key                 | Fields                                                                                               |\n| -------------------- | ------------------- | ---------------------------------------------------------------------------------------------------- |\n| OpenAI               | openai            | apiKey, baseURL                                                                                  |\n| Anthropic            | anthropic         | apiKey, oauthToken, baseURL                                                                    |\n| Google AI Studio     | googleAiStudio    | apiKey, baseURL                                                                                  |\n| Google Vertex AI     | vertex            | projectId, location, apiKey (Express Mode), serviceAccountKey, clientEmail, privateKey   |\n| Amazon Bedrock       | bedrock           | accessKeyId, secretAccessKey, sessionToken, region                                           |\n| Amazon SageMaker     | sagemaker         | accessKeyId, secretAccessKey, sessionToken, region, endpoint                               |\n| Azure OpenAI         | azure             | apiKey, resourceName, deploymentName, apiVersion                                             |\n| Mistral              | mistral           | apiKey                                                                                             |\n| Hugging Face         | huggingFace       | apiKey, baseURL                                                                                  |\n| OpenRouter           | openrouter        | apiKey, baseURL                                                                                  |\n| LiteLLM              | litellm           | apiKey, baseURL                                                                                  |\n| OpenAI-Compatible    | openaiCompatible  | apiKey, baseURL                                                                                  |\n| Cerebras             | cerebras          | apiKey, baseURL                                                                                  |\n| SambaNova            | sambanova         | apiKey, baseURL                                                                                  |\n| Ollama               | ollama            | baseURL                                                                                       ","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"","lvl3":""}},
 {"objectID":"12291afcfaaf4af65b01b07e93a9eac08280ff7c61394f6051a86c1b152a7b4a","title":"Per-Request Credentials","url":"/docs/features/per-request-credentials#per-request-credentials","content":"Status: Stable | Availability: SDK only","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"Per-Request Credentials","lvl3":""}},
 {"objectID":"6c6756b0806344240b3c86d6b0063587db50f2c4ccf936a32e87d4377b77014c","title":"Overview","url":"/docs/features/per-request-credentials#overview","content":"NeuroLink allows provider credentials to be supplied at two levels below the environment-variable default: on the NeuroLink constructor (instance level) and on individual generate() / stream() calls (per-call level). Credentials are resolved in the following order of precedence:\n\nThis enables multi-tenant architectures where different users or tenants supply their own provider API keys (bring-your-own-key, BYOK), without requiring separate NeuroLink instances per user or touching the process environment.\n\nTypical use cases:\nMulti-tenant SaaS — each API request carries the calling user's provider key; no shared key leakage between tenants\nBYOK products — end users paste their OpenAI / Anthropic keys in settings; you forward them to NeuroLink per call\nTesting and CI — inject credentials programmatically without setting environment variables\nProvider switching — override only the active provider's credentials while leaving others to fall through to env vars","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"Overview","lvl3":""}},
 {"objectID":"d4e86918319aa78b4e4461dbf29a60ebe90935e74fc70617f59bf524ba3e1eae","title":"Instance-Level Credentials","url":"/docs/features/per-request-credentials#instance-level-credentials","content":"Set credentials in the NeuroLink constructor. These apply as the default for every generate() and stream() call made on that instance. Useful when serving a single tenant or when you have a known key for the duration of the instance lifecycle.","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"Instance-Level Credentials","lvl3":""}},
 {"objectID":"250b1dd11e018f5f252b27ce408e3dade21eff1cf26eed4aaf81126fad6ab691","title":"Per-Call Credentials","url":"/docs/features/per-request-credentials#per-call-credentials","content":"Set credentials directly on generate() or stream(). These override the instance-level credentials for that single call only. Only the providers you explicitly set are overridden — others continue falling through to instance credentials and then environment variables.","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"Per-Call Credentials","lvl3":""}},
 {"objectID":"4166cbaf6e6263ac98e24d7ede9950b44aade18e787b57e71138d2fc2deaaa07","title":"Precedence Rules","url":"/docs/features/per-request-credentials#precedence-rules","content":"| Level       | Scope                 | Set on                                                                              |\n| ----------- | --------------------- | ----------------------------------------------------------------------------------- |\n| Per-call    | Single request only   | generate({ credentials }), stream({ credentials }) or decide({ credentials }) |\n| Instance    | All calls on instance | new NeuroLink({ credentials })                                                    |\n| Environment | Process-wide fallback | OPENAI_API_KEY, ANTHROPIC_API_KEY, …                                            |\n\nUnset providers at any level fall through to the next. You never need to repeat a credential at the per-call level if the instance default is correct.","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"Precedence Rules","lvl3":""}},
-{"objectID":"b739fcf7819dd6f055ebe3bcdb161ff839067dc172235f9d0072f9cf79d7bcd3","title":"Provider Credential Reference","url":"/docs/features/per-request-credentials#provider-credential-reference","content":"All fields are optional — omit any field you want to fall through to a lower-precedence level.\n\n| Provider             | Key                 | Fields                                                                                               |\n| -------------------- | ------------------- | ---------------------------------------------------------------------------------------------------- |\n| OpenAI               | openai            | apiKey, baseURL                                                                                  |\n| Anthropic            | anthropic         | apiKey, oauthToken                                                                               |\n| Google AI Studio     | googleAiStudio    | apiKey, baseURL                                                                                  |\n| Google Vertex AI     | vertex            | projectId, location, apiKey (Express Mode), serviceAccountKey, clientEmail, privateKey   |\n| Amazon Bedrock       | bedrock           | accessKeyId, secretAccessKey, sessionToken, region                                           |\n| Amazon SageMaker     | sagemaker         | accessKeyId, secretAccessKey, sessionToken, region, endpoint                               |\n| Azure OpenAI         | azure             | apiKey, resourceName, deploymentName, apiVersion                                             |\n| Mistral              | mistral           | apiKey                                                                                             |\n| Hugging Face         | huggingFace       | apiKey, baseURL                                                                                  |\n| OpenRouter           | openrouter        | apiKey, baseURL                                                                                  |\n| LiteLLM              | litellm           | apiKey, baseURL                                                                                  |\n| OpenAI-Compatible    |","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"Provider Credential Reference","lvl3":""}},
+{"objectID":"b739fcf7819dd6f055ebe3bcdb161ff839067dc172235f9d0072f9cf79d7bcd3","title":"Provider Credential Reference","url":"/docs/features/per-request-credentials#provider-credential-reference","content":"All fields are optional — omit any field you want to fall through to a lower-precedence level.\n\n| Provider             | Key                 | Fields                                                                                               |\n| -------------------- | ------------------- | ---------------------------------------------------------------------------------------------------- |\n| OpenAI               | openai            | apiKey, baseURL                                                                                  |\n| Anthropic            | anthropic         | apiKey, oauthToken, baseURL                                                                    |\n| Google AI Studio     | googleAiStudio    | apiKey, baseURL                                                                                  |\n| Google Vertex AI     | vertex            | projectId, location, apiKey (Express Mode), serviceAccountKey, clientEmail, privateKey   |\n| Amazon Bedrock       | bedrock           | accessKeyId, secretAccessKey, sessionToken, region                                           |\n| Amazon SageMaker     | sagemaker         | accessKeyId, secretAccessKey, sessionToken, region, endpoint                               |\n| Azure OpenAI         | azure             | apiKey, resourceName, deploymentName, apiVersion                                             |\n| Mistral              | mistral           | apiKey                                                                                             |\n| Hugging Face         | huggingFace       | apiKey, baseURL                                                                                  |\n| OpenRouter           | openrouter        | apiKey, baseURL                                                                                  |\n| LiteLLM              | litellm           | apiKey, baseURL                                                                                  |\n| OpenAI-Compatible    | openaiCompatible","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"Provider Credential Reference","lvl3":""}},
 {"objectID":"575e2e50c5fd00e53fce66deee556264f3df8962e2188ac446e6e7d4ad29b2ed","title":"OpenAI","url":"/docs/features/per-request-credentials#openai","content":"Custom base URL for OpenAI-compatible proxies:","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"OpenAI","lvl3":""}},
-{"objectID":"499e7c1d012d337ad54e06db35770d5b3471afb0f9e3f471564642e8ba7e5c6b","title":"Anthropic","url":"/docs/features/per-request-credentials#anthropic","content":"OAuth token (Anthropic Claude subscription):","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"Anthropic","lvl3":""}},
+{"objectID":"499e7c1d012d337ad54e06db35770d5b3471afb0f9e3f471564642e8ba7e5c6b","title":"Anthropic","url":"/docs/features/per-request-credentials#anthropic","content":"OAuth token (Anthropic Claude subscription):\n\nA gateway or proxy in front of Anthropic (baseURL wins over ANTHROPIC_BASE_URL, with or without a trailing /v1; a gateway that does its own auth accepts any apiKey):","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"Anthropic","lvl3":""}},
 {"objectID":"3f5ea871d2750f36fca227c33f109de53f5d0bc4c184ea956d0b8c3736babe29","title":"Vertex AI — Express Mode","url":"/docs/features/per-request-credentials#vertex-ai-express-mode","content":"Express Mode uses a simple API key instead of service-account credentials, making it suitable for per-request BYOK flows:\n\nFull service-account credentials (server-side only — keep private keys out of client code):","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"Vertex AI — Express Mode","lvl3":""}},
 {"objectID":"af5efdaa7ba1811d537c1c7aca8a0143cd8496be1e9db86526d34003cf427dbb","title":"Streaming with Credentials","url":"/docs/features/per-request-credentials#streaming-with-credentials","content":"credentials works identically on stream(). Note that neurolink.stream()\nreturns a StreamResult wrapper — iterate over its .stream property:","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"Streaming with Credentials","lvl3":""}},
 {"objectID":"ea70e911923bd079e70af1c36bda14c16bdd9bfcec49f236a9d0550d6602502a","title":"Multi-Tenant Request Handler","url":"/docs/features/per-request-credentials#multi-tenant-request-handler","content":"A typical pattern for a multi-tenant API endpoint. Note the provider-name →\ncredential-key mapping: the registered provider names (google-ai,\nopenai-compatible, huggingface) differ from their NeurolinkCredentials\nslot keys (googleAiStudio, openaiCompatible, huggingFace), so map\nexplicitly to avoid runtime surprises.\n\nNote: This example assumes API-key authentication. Providers like Bedrock\n(which use accessKeyId/secretAccessKey) and Ollama (which use baseURL)\nrequire different credential shapes — see the Provider Credential Reference above.","hierarchy":{"lvl0":"Features","lvl1":"Per-Request Credentials","lvl2":"Multi-Tenant Request Handler","lvl3":""}},
@@ -9687,8 +9687,8 @@
 {"objectID":"a5276d835ceba15da6ec36261ab78001e17f8b78ba9c8cb1c5e1282aa2f377e3","title":"Error-message fidelity","url":"/docs/provider-integration/openai-compat-catalog#error-message-fidelity","content":"Each entry's errorRules is a direct, order-preserving translation of its\noriginal subclass's formatProviderError if/else ladder into rule data\n(status code and/or case-insensitive pattern), classified via\nclassifyProviderError()\n(src/lib/utils/errorClassifier.ts). Every bespoke message string is\npreserved verbatim — including xAI's \"top up your account\" quota URL and\nGroq's decommissioned-vs-not-found distinction — via each rule's own\nmessage field (string | ((ctx) => string)), with model-name\ninterpolation carried through ctx.modelName. There is no message-wording\nregression here. Timeout classification is likewise unchanged: every catalog\nprovider except Groq maps TimeoutError to NetworkError (the classifier's\ndefault), and Groq alone maps it to ProviderError. Groq's subclass override is\npreserved verbatim via the JSON's quirks.timeoutErrorClass, so no\nprovider's timeout class changed during migration.","hierarchy":{"lvl0":"Provider Integration","lvl1":"OpenAI-Compatible Provider Catalog","lvl2":"Error-message fidelity","lvl3":""}},
 {"objectID":"2a22fddc1a2c383a5e75a53e2d512d764f01ad66ab03bfe6edf12acb5d204814","title":"Known pre-existing quirk this migration preserved (not fixed)","url":"/docs/provider-integration/openai-compat-catalog#known-pre-existing-quirk-this-migration-preserved-not-fixed","content":"Mistral's provider registration passes a defaultModel value to\nProviderFactory.registerProvider() that does not check MISTRAL_MODEL\n(MistralModels.MISTRAL_LARGE_LATEST, a bare literal), while\nConfiguredOpenAICompatProvider.getDefaultModel() for Mistral does\ncheck MISTRAL_MODEL (falling back to MistralModels.MISTRAL_SMALL_2506).\nEvery other catalog provider's registry default and class default agree.\nThis is expressed via the JSON's quirks.registryDefaultIgnoresModelEnvVar\n(true only for Mistral). The JSON migration did reconcile one half of it:\ngetDefaultModel(MISTRAL) now returns the real generation default\n(mistral-small-2506) rather than the registry literal — a disclosed\nbug-fix-grade delta, since the two disagreed before.","hierarchy":{"lvl0":"Provider Integration","lvl1":"OpenAI-Compatible Provider Catalog","lvl2":"Known pre-existing quirk this migration preserved (not fixed)","lvl3":""}},
 {"objectID":"1d94ab5116c71ed28e2428889216922df96e1d6d4d5fdc955ee65a4171c483e3","title":"What comes next","url":"/docs/provider-integration/openai-compat-catalog#what-comes-next","content":"The credential-free growth queue in docs/provider-integration/growth-queue.json lists every\nverified candidate vendor by wave (W1 = hosted OpenAI-compatible, the catalog path). Entries\nonboarded from it carry evidence.liveMatrix: null until a live run with a real key.","hierarchy":{"lvl0":"Provider Integration","lvl1":"OpenAI-Compatible Provider Catalog","lvl2":"What comes next","lvl3":""}},
-{"objectID":"bf3e1f8f1233a9b3d12aedbdbec4bf8cd196d18bf56b7bb104b34e31e90b61e0","title":"Provider Onboarding Tiers","url":"/docs/provider-integration/tiers/README","content":"Provider Onboarding Tiers\n\nFour tiers, ordered by effort. Always pick the lowest tier that's\nactually true for the provider you're adding — a provider that's\nOpenAI-wire-compatible but gets built as a bespoke Tier 3 subclass \"to be\nsafe\" is exactly the copy-pasted-boilerplate problem this redesign\nexists to eliminate (see ../adr/0002-catalog-over-subclass-default.md).\n\n| Tier                       | Example                                                                            | Code required                                                     | Time                              |\n| -------------------------- | ---------------------------------------------------------------------------------- | ----------------------------------------------------------------- | --------------------------------- |\n| 1 — Aggregator passthrough | A new model id on an existing LiteLLM/OpenRouter route                             | None                                                              | Minutes                           |\n| 2 — Catalog entry          | A new zero-quirk OpenAI-compatible vendor (the Groq/xAI/Together shape)            | One data row + one mocked-test section                            | ~1 hour                           |\n| 3 — Adapter-based native   | A vendor with its own SDK/wire format but a normal request/response HTTP lifecycle | One provider class                                                | Days                              |\n| 4 — Full custom            | SageMaker-class: non-HTTP protocol, SDK-signed auth, bespoke lifecycle             | Custom executeStream/doGenerate, possibly own CLI subcommands | Days, needs written justification |\n\nTier 1 adds zero provider coverage — it is a usage change against an\nalready-counted aggregator, never reflected in\ndocs/reference/provider-comparison.md or the README's provider count.\n\nEvery tier that adds a new AIProviderName member (Tier 2 and above) ends\nthe same way: a manifest at\ndocs/provider-integration/manifests/<provider>.json\n(see ../manifests/README.md) and a green run\nof pnpm run verify:provider-onboarding (see\n../../../tools/verify-provider-onboarding.ts).\nTier 1 needs no manifest and no gate — see\ntier-1-aggregator-passthrough.md.\n\nUse ../../../tools/scaffold-provider.ts\n(pnpm run scaffold:provider) to generate the starting-point snippets for\nTiers 2–4 instead of copy-pasting from an existing provider by hand. Both\ntools ship in the tree; there is no manual-fallback era anymore — a PR\nthat skips the gate locally just fails it in CI.","hierarchy":{"lvl0":"Provider Integration","lvl1":"Provider Onboarding Tiers","lvl2":"","lvl3":""}},
-{"objectID":"de01fcaf93bee302e2fca5e18f85cba5def099685383ec01b61d18489dbca5aa","title":"Provider Onboarding Tiers","url":"/docs/provider-integration/tiers/README#provider-onboarding-tiers","content":"Four tiers, ordered by effort. Always pick the lowest tier that's\nactually true for the provider you're adding — a provider that's\nOpenAI-wire-compatible but gets built as a bespoke Tier 3 subclass \"to be\nsafe\" is exactly the copy-pasted-boilerplate problem this redesign\nexists to eliminate (see ../adr/0002-catalog-over-subclass-default.md).\n\n| Tier                       | Example                                                                            | Code required                                                     | Time                              |\n| -------------------------- | ---------------------------------------------------------------------------------- | ----------------------------------------------------------------- | --------------------------------- |\n| 1 — Aggregator passthrough | A new model id on an existing LiteLLM/OpenRouter route                             | None                                                              | Minutes                           |\n| 2 — Catalog entry          | A new zero-quirk OpenAI-compatible vendor (the Groq/xAI/Together shape)            | One data row + one mocked-test section                            | ~1 hour                           |\n| 3 — Adapter-based native   | A vendor with its own SDK/wire format but a normal request/response HTTP lifecycle | One provider class                                                | Days                              |\n| 4 — Full custom            | SageMaker-class: non-HTTP protocol, SDK-signed auth, bespoke lifecycle             | Custom executeStream/doGenerate, possibly own CLI subcommands | Days, needs written justification |\n\nTier 1 adds zero provider coverage — it is a usage change against an\nalready-counted aggregator, never reflected in\ndocs/reference/provider-comparison.md or the README's provider count.","hierarchy":{"lvl0":"Provider Integration","lvl1":"Provider Onboarding Tiers","lvl2":"Provider Onboarding Tiers","lvl3":""}},
+{"objectID":"bf3e1f8f1233a9b3d12aedbdbec4bf8cd196d18bf56b7bb104b34e31e90b61e0","title":"Provider Onboarding Tiers","url":"/docs/provider-integration/tiers/README","content":"Provider Onboarding Tiers\n\nFour tiers, ordered by effort. Always pick the lowest tier that's\nactually true for the provider you're adding — a provider that's\nOpenAI-wire-compatible but gets built as a bespoke Tier 3 subclass \"to be\nsafe\" is exactly the copy-pasted-boilerplate problem this redesign\nexists to eliminate (see ../adr/0002-catalog-over-subclass-default.md).\n\nProvider Integration index\n\n| Tier                       | Example                                                                            | Code required                                                     | Time                              |\n| -------------------------- | ---------------------------------------------------------------------------------- | ----------------------------------------------------------------- | --------------------------------- |\n| 1 — Aggregator passthrough | A new model id on an existing LiteLLM/OpenRouter route                             | None                                                              | Minutes                           |\n| 2 — Catalog entry          | A new zero-quirk OpenAI-compatible vendor (the Groq/xAI/Together shape)            | One data row + one mocked-test section                            | ~1 hour                           |\n| 3 — Adapter-based native   | A vendor with its own SDK/wire format but a normal request/response HTTP lifecycle | One provider class                                                | Days                              |\n| 4 — Full custom            | SageMaker-class: non-HTTP protocol, SDK-signed auth, bespoke lifecycle             | Custom executeStream/doGenerate, possibly own CLI subcommands | Days, needs written justification |\n\nTier 1 adds zero provider coverage — it is a usage change against an\nalready-counted aggregator, never reflected in\ndocs/reference/provider-comparison.md or the README's provider count.\n\nEvery tier that adds a new AIProviderName member (Tier 2 and above) ends\nthe same way: a manifest at\ndocs/provider-integration/manifests/<provider>.json\n(see ../manifests/README.md) and a green run\nof pnpm run verify:provider-onboarding (see\n../../../tools/verify-provider-onboarding.ts).\nTier 1 needs no manifest and no gate — see\ntier-1-aggregator-passthrough.md.\n\nUse ../../../tools/scaffold-provider.ts\n(pnpm run scaffold:provider) to generate the starting-point snippets for\nTiers 2–4 instead of copy-pasting from an existing provider by hand. Both\ntools ship in the tree; there is no manual-fallback era anymore — a PR\nthat skips the gate locally just fails it in CI.","hierarchy":{"lvl0":"Provider Integration","lvl1":"Provider Onboarding Tiers","lvl2":"","lvl3":""}},
+{"objectID":"de01fcaf93bee302e2fca5e18f85cba5def099685383ec01b61d18489dbca5aa","title":"Provider Onboarding Tiers","url":"/docs/provider-integration/tiers/README#provider-onboarding-tiers","content":"Four tiers, ordered by effort. Always pick the lowest tier that's\nactually true for the provider you're adding — a provider that's\nOpenAI-wire-compatible but gets built as a bespoke Tier 3 subclass \"to be\nsafe\" is exactly the copy-pasted-boilerplate problem this redesign\nexists to eliminate (see ../adr/0002-catalog-over-subclass-default.md).\n\nProvider Integration index\n\n| Tier                       | Example                                                                            | Code required                                                     | Time                              |\n| -------------------------- | ---------------------------------------------------------------------------------- | ----------------------------------------------------------------- | --------------------------------- |\n| 1 — Aggregator passthrough | A new model id on an existing LiteLLM/OpenRouter route                             | None                                                              | Minutes                           |\n| 2 — Catalog entry          | A new zero-quirk OpenAI-compatible vendor (the Groq/xAI/Together shape)            | One data row + one mocked-test section                            | ~1 hour                           |\n| 3 — Adapter-based native   | A vendor with its own SDK/wire format but a normal request/response HTTP lifecycle | One provider class                                                | Days                              |\n| 4 — Full custom            | SageMaker-class: non-HTTP protocol, SDK-signed auth, bespoke lifecycle             | Custom executeStream/doGenerate, possibly own CLI subcommands | Days, needs written justification |\n\nTier 1 adds zero provider coverage — it is a usage change against an\nalready-counted aggregator, never reflected in\ndocs/reference/provider-comparison.md or the README's provider count.","hierarchy":{"lvl0":"Provider Integration","lvl1":"Provider Onboarding Tiers","lvl2":"Provider Onboarding Tiers","lvl3":""}},
 {"objectID":"a9753e41ed762a57c15471c49e70dc9917e8388eca055e3dc1c4ce90db6804bb","title":"Tier 1 — Aggregator Passthrough","url":"/docs/provider-integration/tiers/tier-1-aggregator-passthrough","content":"Tier 1 — Aggregator Passthrough\n\nWhen this applies: the model you want is already reachable through a\nprovider NeuroLink already registers as a pass-through aggregator —\ntoday that's litellm (any backend the user's LiteLLM proxy exposes) or\nopenrouter (any model in OpenRouter's catalog). No new\nAIProviderName member, no new provider class, no new catalog row.\n\nWhat you're actually doing: picking a model id string and confirming\nit works — this is a usage change, not an integration change.\n\nChecklist\n[ ] Confirm the aggregator actually serves the model. For LiteLLM,\n      check the proxy's /v1/models (or its config.yaml) for the\n      model's model_name. For OpenRouter, check\n      https://openrouter.ai/models for the exact vendor/model-id\n      slug.\n[ ] No AIProviderName enum change. No PROVIDER_DESCRIPTORS change.\n      No OPENAI_COMPAT_CATALOG change. If you find yourself editing any\n      of those three for a \"Tier 1\" provider, it isn't Tier 1 — restart\n      from ../tiers/README.md's decision tree.\n[ ] Optional: if the model needs a friendlier default alias, add it to\n      MODEL_ALIASES / DEFAULT_MODEL_ALIASES in\n      src/lib/models/modelRegistry.ts. Not required for the model to\n      work.\n[ ] Optional: if the model needs a documented env var (e.g., a\n      dedicated LiteLLM route), document it in\n      docs/getting-started/environment-variables.md.\n[ ] Manually smoke-test the model end-to-end.\n[ ] No manifest file is required — tools/verify-provider-onboarding.ts\n      (see ../../../tools/verify-provider-onboarding.ts) only gates\n      new AIProviderName members, and Tier 1 never adds one.\n[ ] Confirmed: no edit to docs/reference/provider-comparison.md or\n      README provider count for this change.\n\nVerification commands\n\nBoth should return a normal GenerateResult with non-empty content. If\neither 400s with an \"unknown model\" style error, the aggregator doesn't\nactually serve that model yet — fix the aggregator-side config, not\nNeuroLink.","hierarchy":{"lvl0":"Provider Integration","lvl1":"Tier 1 — Aggregator Passthrough","lvl2":"","lvl3":""}},
 {"objectID":"f04f6a19785cf4ff03d1ed0808f35ddbedb0508a51ca0590d1a57ff6703b40eb","title":"Tier 1 — Aggregator Passthrough","url":"/docs/provider-integration/tiers/tier-1-aggregator-passthrough#tier-1-aggregator-passthrough","content":"When this applies: the model you want is already reachable through a\nprovider NeuroLink already registers as a pass-through aggregator —\ntoday that's litellm (any backend the user's LiteLLM proxy exposes) or\nopenrouter (any model in OpenRouter's catalog). No new\nAIProviderName member, no new provider class, no new catalog row.\n\nWhat you're actually doing: picking a model id string and confirming\nit works — this is a usage change, not an integration change.","hierarchy":{"lvl0":"Provider Integration","lvl1":"Tier 1 — Aggregator Passthrough","lvl2":"Tier 1 — Aggregator Passthrough","lvl3":""}},
 {"objectID":"209d4e31f0601ea928a348869f63ad32c049e6624269f06d4fe795f15d30d311","title":"Checklist","url":"/docs/provider-integration/tiers/tier-1-aggregator-passthrough#checklist","content":"[ ] Confirm the aggregator actually serves the model. For LiteLLM,\n      check the proxy's /v1/models (or its config.yaml) for the\n      model's model_name. For OpenRouter, check\n      https://openrouter.ai/models for the exact vendor/model-id\n      slug.\n[ ] No AIProviderName enum change. No PROVIDER_DESCRIPTORS change.\n      No OPENAI_COMPAT_CATALOG change. If you find yourself editing any\n      of those three for a \"Tier 1\" provider, it isn't Tier 1 — restart\n      from ../tiers/README.md's decision tree.\n[ ] Optional: if the model needs a friendlier default alias, add it to\n      MODEL_ALIASES / DEFAULT_MODEL_ALIASES in\n      src/lib/models/modelRegistry.ts. Not required for the model to\n      work.\n[ ] Optional: if the model needs a documented env var (e.g., a\n      dedicated LiteLLM route), document it in\n      docs/getting-started/environment-variables.md.\n[ ] Manually smoke-test the model end-to-end.\n[ ] No manifest file is required — tools/verify-provider-onboarding.ts\n      (see ../../../tools/verify-provider-onboarding.ts) only gates\n      new AIProviderName members, and Tier 1 never adds one.\n[ ] Confirmed: no edit to docs/reference/provider-comparison.md or\n      README provider count for this change.","hierarchy":{"lvl0":"Provider Integration","lvl1":"Tier 1 — Aggregator Passthrough","lvl2":"Checklist","lvl3":""}},

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@juspay/neurolink",
-  "version": "12.47.1",
+  "version": "12.47.3",
   "packageManager": "pnpm@10.15.1",
   "description": "The pipe layer of an AI nervous system: one interface connecting provider neurons to your application, across three inference types — generate, stream and decide. `decide` returns typed, calibrated judgements from a non-generative model (~400ms, ~$0.00002/call) for routing, tool selection and context budgeting. MCP-native (4 transports), voice TTS/STT/realtime, RAG, agents, memory, compaction, 9 observability exporters. OpenAI · Anthropic · Gemini · Bedrock · Azure · Ollama · TypeSafe Jev and more.",
   "author": {
@@ -280,6 +280,8 @@
     "test:anthropic-loop-characterization": "tsx test/continuous-test-suite-anthropic-loop-characterization.ts",
     "test:anthropic-execution-control": "tsx test/continuous-test-suite-anthropic-execution-control.ts",
     "test:aistudio-loop-characterization": "tsx test/continuous-test-suite-aistudio-loop-characterization.ts",
+    "test:google-genai-proxy": "tsx test/continuous-test-suite-google-genai-proxy.ts",
+    "test:google-genai-sdk": "tsx test/continuous-test-suite-google-genai-sdk.ts",
     "test:docs-mcp": "pnpm exec tsx test/continuous-test-suite-docs-mcp.ts",
     "test:docs-search-index": "pnpm exec tsx test/continuous-test-suite-docs-search-index.ts",
     "test:docs-snippets": "pnpm exec tsx test/continuous-test-suite-docs-snippets.ts",
@@ -430,7 +432,7 @@
     "@anthropic-ai/vertex-sdk": "^0.16.0",
     "@aws-sdk/types": "^3.862.0",
     "@cfworker/json-schema": "^4.1.1",
-    "@google/genai": "^1.43.0",
+    "@google/genai": "^2.23.0",
     "@modelcontextprotocol/sdk": "^1.27.1",
     "@opentelemetry/api-logs": "^0.214.0",
     "@opentelemetry/context-async-hooks": "^2.6.1",