@rouvanpm/rouva 0.1.4 → 0.2.0

package/README.md

@@ -24,7 +24,7 @@ console.log(response.choices[0].message.content)
 
 ## Provider agnostic
 
-Rouva works with all connected providers — Anthropic, OpenAI, Gemini, DeepSeek, Mistral, and xAI Grok. You can request a specific model, force a specific provider, or omit both and let Rouva route to the cheapest capable model automatically.
+Rouva works with all connected providers — Anthropic and OpenAI today, more coming. You can request a specific model or omit it entirely and let Rouva route to the cheapest capable model automatically.
 
 ```typescript
 // Request a specific model
@@ -33,10 +33,9 @@ const res = await rouva.chat.completions.create({
   messages,
 })
 
-// Force a specific provider + model
+// Request a specific Anthropic model
 const res = await rouva.chat.completions.create({
-  provider: 'xai',
-  model: 'grok-4.3',
+  model: 'claude-sonnet-4-6',
   messages,
 })
 
@@ -46,7 +45,7 @@ const res = await rouva.chat.completions.create({
 })
 ```
 
-## OpenAI-style request shape
+## Drop-in replacement for OpenAI
 
 ```typescript
 // Before
@@ -78,8 +77,6 @@ while (true) {
 }
 ```
 
-Streaming responses are normalized to OpenAI-style SSE chunks, even when Rouva routes the request to Anthropic.
-
 ## Options
 
 ```typescript
@@ -89,21 +86,6 @@ const rouva = new Rouva({
 })
 ```
 
-## Response metadata
-
-Parsed non-stream responses may include a `_rouva` field with gateway header metadata when available:
-
-```typescript
-const res = await rouva.chat.completions.create({ messages })
-
-console.log(res._rouva)
-// {
-//   model_used: 'gpt-4o-mini',
-//   provider_used: 'openai',
-//   task_type: 'summarize'
-// }
-```
-
 ## Getting your API key
 
 1. Sign in to [app.rouva.io](https://app.rouva.io)
@@ -112,4 +94,4 @@ console.log(res._rouva)
 
 ## License
 
-MIT
+MIT

package/dist/index.d.mts

@@ -8,24 +8,30 @@ interface Message {
     role: 'system' | 'user' | 'assistant';
     content: string;
 }
-type RouvaProvider = 'anthropic' | 'openai' | 'gemini' | 'deepseek' | 'mistral' | 'xai' | (string & {});
 /**
  * Supported models across all providers.
  * Omit `model` entirely to let Rouva route to the cheapest capable model automatically.
  */
-type RouvaModel = 'claude-opus-4-6' | 'claude-opus-4-7' | 'claude-opus-4-8' | 'claude-sonnet-4-6' | 'claude-haiku-4-5-20251001' | 'claude-fable-5' | 'gpt-5-nano' | 'gpt-5-mini' | 'gpt-5' | 'gpt-5.5' | 'gpt-5.5-pro' | 'gpt-4.1-nano' | 'gpt-4.1-mini' | 'gpt-4.1' | 'gpt-4o' | 'gpt-4o-mini' | 'gemini-2.5-flash' | 'gemini-2.5-pro' | 'deepseek-chat' | 'deepseek-reasoner' | 'mistral-small-latest' | 'mistral-large-latest' | 'grok-4.3' | 'grok-4.20-0309-reasoning' | (string & {});
+type RouvaModel = 'claude-opus-4-6' | 'claude-sonnet-4-6' | 'claude-haiku-4-5-20251001' | 'gpt-4o' | 'gpt-4o-mini' | 'gpt-4-turbo' | 'gpt-3.5-turbo' | (string & {});
+/** Anthropic-style text block — the array form of the `system` param */
+interface SystemTextBlock {
+    type: 'text';
+    text: string;
+}
 interface ChatCompletionParams {
     messages: Message[];
     /**
-     * Target model — omit to let Rouva route intelligently to the cheapest capable model.
-     * Supports models from any connected provider.
+     * System prompt — a plain string or Anthropic-style text blocks.
+     * Equivalent to a `role: "system"` message at the head of `messages`;
+     * the gateway accepts both forms and re-emits whichever dialect the
+     * routed provider expects, so it survives cross-provider routing.
      */
-    model?: RouvaModel;
+    system?: string | SystemTextBlock[];
     /**
-     * Force an exact provider when paired with `model`.
-     * Omit to let Rouva auto-route based on your connected keys.
+     * Target model — omit to let Rouva route intelligently to the cheapest capable model.
+     * Supports models from any connected provider (Anthropic, OpenAI).
      */
-    provider?: RouvaProvider;
+    model?: RouvaModel;
     /** Maximum tokens to generate */
     max_tokens?: number;
     /** Sampling temperature 0–1 */
@@ -54,14 +60,16 @@ interface ChatCompletion {
     _rouva?: RouvaResponseMeta;
 }
 interface RouvaResponseMeta {
-    /** Actual model used when exposed by the gateway */
-    model_used?: string;
-    /** Actual provider used when exposed by the gateway */
-    provider_used?: string;
-    /** Task type classified by Rouva when exposed by the gateway */
-    task_type?: string;
-    /** Semantic cache status when exposed by the gateway */
-    cache?: string;
+    /** Actual model used (may differ from requested when intelligently routed) */
+    model_used: string;
+    /** USD cost for this request */
+    cost: number;
+    /** USD saved vs your baseline model */
+    savings: number;
+    /** Whether intelligent routing selected the model */
+    intelligently_routed: boolean;
+    /** Task type classified by Rouva */
+    task_type: string;
 }
 
 declare class Rouva {
@@ -76,4 +84,4 @@ declare class Rouva {
     private _createChatCompletion;
 }
 
-export { type ChatCompletion, type ChatCompletionChoice, type ChatCompletionParams, type ChatCompletionUsage, type Message, Rouva, type RouvaOptions, type RouvaProvider, type RouvaResponseMeta };
+export { type ChatCompletion, type ChatCompletionChoice, type ChatCompletionParams, type ChatCompletionUsage, type Message, Rouva, type RouvaOptions, type RouvaResponseMeta, type SystemTextBlock };

package/dist/index.d.ts

@@ -8,24 +8,30 @@ interface Message {
     role: 'system' | 'user' | 'assistant';
     content: string;
 }
-type RouvaProvider = 'anthropic' | 'openai' | 'gemini' | 'deepseek' | 'mistral' | 'xai' | (string & {});
 /**
  * Supported models across all providers.
  * Omit `model` entirely to let Rouva route to the cheapest capable model automatically.
  */
-type RouvaModel = 'claude-opus-4-6' | 'claude-opus-4-7' | 'claude-opus-4-8' | 'claude-sonnet-4-6' | 'claude-haiku-4-5-20251001' | 'claude-fable-5' | 'gpt-5-nano' | 'gpt-5-mini' | 'gpt-5' | 'gpt-5.5' | 'gpt-5.5-pro' | 'gpt-4.1-nano' | 'gpt-4.1-mini' | 'gpt-4.1' | 'gpt-4o' | 'gpt-4o-mini' | 'gemini-2.5-flash' | 'gemini-2.5-pro' | 'deepseek-chat' | 'deepseek-reasoner' | 'mistral-small-latest' | 'mistral-large-latest' | 'grok-4.3' | 'grok-4.20-0309-reasoning' | (string & {});
+type RouvaModel = 'claude-opus-4-6' | 'claude-sonnet-4-6' | 'claude-haiku-4-5-20251001' | 'gpt-4o' | 'gpt-4o-mini' | 'gpt-4-turbo' | 'gpt-3.5-turbo' | (string & {});
+/** Anthropic-style text block — the array form of the `system` param */
+interface SystemTextBlock {
+    type: 'text';
+    text: string;
+}
 interface ChatCompletionParams {
     messages: Message[];
     /**
-     * Target model — omit to let Rouva route intelligently to the cheapest capable model.
-     * Supports models from any connected provider.
+     * System prompt — a plain string or Anthropic-style text blocks.
+     * Equivalent to a `role: "system"` message at the head of `messages`;
+     * the gateway accepts both forms and re-emits whichever dialect the
+     * routed provider expects, so it survives cross-provider routing.
      */
-    model?: RouvaModel;
+    system?: string | SystemTextBlock[];
     /**
-     * Force an exact provider when paired with `model`.
-     * Omit to let Rouva auto-route based on your connected keys.
+     * Target model — omit to let Rouva route intelligently to the cheapest capable model.
+     * Supports models from any connected provider (Anthropic, OpenAI).
      */
-    provider?: RouvaProvider;
+    model?: RouvaModel;
     /** Maximum tokens to generate */
     max_tokens?: number;
     /** Sampling temperature 0–1 */
@@ -54,14 +60,16 @@ interface ChatCompletion {
     _rouva?: RouvaResponseMeta;
 }
 interface RouvaResponseMeta {
-    /** Actual model used when exposed by the gateway */
-    model_used?: string;
-    /** Actual provider used when exposed by the gateway */
-    provider_used?: string;
-    /** Task type classified by Rouva when exposed by the gateway */
-    task_type?: string;
-    /** Semantic cache status when exposed by the gateway */
-    cache?: string;
+    /** Actual model used (may differ from requested when intelligently routed) */
+    model_used: string;
+    /** USD cost for this request */
+    cost: number;
+    /** USD saved vs your baseline model */
+    savings: number;
+    /** Whether intelligent routing selected the model */
+    intelligently_routed: boolean;
+    /** Task type classified by Rouva */
+    task_type: string;
 }
 
 declare class Rouva {
@@ -76,4 +84,4 @@ declare class Rouva {
     private _createChatCompletion;
 }
 
-export { type ChatCompletion, type ChatCompletionChoice, type ChatCompletionParams, type ChatCompletionUsage, type Message, Rouva, type RouvaOptions, type RouvaProvider, type RouvaResponseMeta };
+export { type ChatCompletion, type ChatCompletionChoice, type ChatCompletionParams, type ChatCompletionUsage, type Message, Rouva, type RouvaOptions, type RouvaResponseMeta, type SystemTextBlock };

package/dist/index.js

@@ -24,273 +24,6 @@ __export(index_exports, {
 });
 module.exports = __toCommonJS(index_exports);
 
-// src/sse.ts
-function parseSseEvent(line) {
-  const raw = line.trim();
-  if (!raw.startsWith("data:")) return null;
-  const data = raw.slice(5).trim();
-  if (!data) return null;
-  try {
-    return { raw, data, json: data === "[DONE]" ? null : JSON.parse(data) };
-  } catch {
-    return { raw, data, json: null };
-  }
-}
-function stringifySse(payload) {
-  return `data: ${JSON.stringify(payload)}
-
-`;
-}
-function detectProvider(payload) {
-  if (!payload || typeof payload !== "object") return null;
-  if ("choices" in payload || "usage" in payload) return "openai";
-  if ("type" in payload) return "anthropic";
-  return null;
-}
-function normalizeChatId(id) {
-  return id.startsWith("chatcmpl_") ? id : `chatcmpl_${id}`;
-}
-function toFinishReason(stopReason) {
-  if (!stopReason) return null;
-  if (stopReason === "end_turn" || stopReason === "stop_sequence") return "stop";
-  if (stopReason === "max_tokens") return "length";
-  return stopReason;
-}
-function initialState() {
-  return {
-    provider: null,
-    id: `chatcmpl_${Date.now()}`,
-    model: "unknown",
-    created: Math.floor(Date.now() / 1e3),
-    promptTokens: 0,
-    completionTokens: 0,
-    roleEmitted: false,
-    finalChunkEmitted: false
-  };
-}
-function normalizeAnthropicPayload(payload, state) {
-  const eventType = typeof payload.type === "string" ? payload.type : null;
-  const chunks = [];
-  if (eventType === "message_start") {
-    const message = payload.message ?? {};
-    const usage = message.usage ?? {};
-    state.id = normalizeChatId(typeof message.id === "string" ? message.id : state.id);
-    state.model = typeof message.model === "string" ? message.model : state.model;
-    state.promptTokens = typeof usage.input_tokens === "number" ? usage.input_tokens : 0;
-    if (!state.roleEmitted) {
-      chunks.push(stringifySse({
-        id: state.id,
-        object: "chat.completion.chunk",
-        created: state.created,
-        model: state.model,
-        choices: [{ index: 0, delta: { role: "assistant" }, finish_reason: null }]
-      }));
-      state.roleEmitted = true;
-    }
-  }
-  if (eventType === "content_block_delta") {
-    const delta = payload.delta ?? {};
-    if (delta.type === "text_delta" && typeof delta.text === "string") {
-      if (!state.roleEmitted) {
-        chunks.push(stringifySse({
-          id: state.id,
-          object: "chat.completion.chunk",
-          created: state.created,
-          model: state.model,
-          choices: [{ index: 0, delta: { role: "assistant" }, finish_reason: null }]
-        }));
-        state.roleEmitted = true;
-      }
-      chunks.push(stringifySse({
-        id: state.id,
-        object: "chat.completion.chunk",
-        created: state.created,
-        model: state.model,
-        choices: [{ index: 0, delta: { content: delta.text }, finish_reason: null }]
-      }));
-    }
-  }
-  if (eventType === "message_delta") {
-    const delta = payload.delta ?? {};
-    const usage = payload.usage ?? {};
-    state.completionTokens = typeof usage.output_tokens === "number" ? usage.output_tokens : state.completionTokens;
-    chunks.push(stringifySse({
-      id: state.id,
-      object: "chat.completion.chunk",
-      created: state.created,
-      model: state.model,
-      choices: [{ index: 0, delta: {}, finish_reason: toFinishReason(typeof delta.stop_reason === "string" ? delta.stop_reason : null) }],
-      usage: {
-        prompt_tokens: state.promptTokens,
-        completion_tokens: state.completionTokens,
-        total_tokens: state.promptTokens + state.completionTokens
-      }
-    }));
-    state.finalChunkEmitted = true;
-  }
-  if (eventType === "message_stop") {
-    chunks.push("data: [DONE]\n\n");
-  }
-  if (eventType === "error") {
-    chunks.push(stringifySse(payload));
-  }
-  return chunks;
-}
-function normalizeGatewayStream(stream) {
-  const encoder = new TextEncoder();
-  const decoder = new TextDecoder();
-  let buffer = "";
-  const state = initialState();
-  return new ReadableStream({
-    async start(controller) {
-      const reader = stream.getReader();
-      try {
-        while (true) {
-          const { done, value } = await reader.read();
-          if (done) break;
-          if (!value) continue;
-          buffer += decoder.decode(value, { stream: true });
-          const lines = buffer.split("\n");
-          buffer = lines.pop() ?? "";
-          for (const line of lines) {
-            const event = parseSseEvent(line);
-            if (!event) continue;
-            if (event.data === "[DONE]") {
-              controller.enqueue(encoder.encode("data: [DONE]\n\n"));
-              continue;
-            }
-            const provider = detectProvider(event.json);
-            if (provider && !state.provider) state.provider = provider;
-            if (state.provider === "anthropic" && event.json && typeof event.json === "object") {
-              for (const chunk of normalizeAnthropicPayload(event.json, state)) {
-                controller.enqueue(encoder.encode(chunk));
-              }
-              continue;
-            }
-            controller.enqueue(encoder.encode(`${event.raw}
-
-`));
-          }
-        }
-        if (buffer.trim()) {
-          const event = parseSseEvent(buffer);
-          if (event) {
-            if (event.data === "[DONE]") {
-              controller.enqueue(encoder.encode("data: [DONE]\n\n"));
-            } else if (state.provider === "anthropic" && event.json && typeof event.json === "object") {
-              for (const chunk of normalizeAnthropicPayload(event.json, state)) {
-                controller.enqueue(encoder.encode(chunk));
-              }
-            } else {
-              controller.enqueue(encoder.encode(`${event.raw}
-
-`));
-            }
-          }
-        }
-      } finally {
-        controller.close();
-      }
-    }
-  });
-}
-async function readChatCompletionFromSse(stream, metadata) {
-  const reader = stream.getReader();
-  const decoder = new TextDecoder();
-  let buffer = "";
-  let id = `chatcmpl_${Date.now()}`;
-  let model = metadata?.model_used ?? "unknown";
-  let created = Math.floor(Date.now() / 1e3);
-  let role = "assistant";
-  let content = "";
-  let finishReason = null;
-  let usage = {
-    prompt_tokens: 0,
-    completion_tokens: 0,
-    total_tokens: 0
-  };
-  try {
-    while (true) {
-      const { done, value } = await reader.read();
-      if (done) break;
-      if (!value) continue;
-      buffer += decoder.decode(value, { stream: true });
-      const lines = buffer.split("\n");
-      buffer = lines.pop() ?? "";
-      for (const line of lines) {
-        const event = parseSseEvent(line);
-        if (!event || event.data === "[DONE]") continue;
-        if (!event.json || typeof event.json !== "object") continue;
-        const payload = event.json;
-        if (payload.type === "error" && payload.error) {
-          const error = payload.error;
-          throw new Error(
-            `[Rouva] Upstream error in stream: ${typeof error.message === "string" ? error.message : "Unknown error"}`
-          );
-        }
-        if (typeof payload.id === "string") id = payload.id;
-        if (typeof payload.model === "string") model = payload.model;
-        if (typeof payload.created === "number") created = payload.created;
-        const choices = Array.isArray(payload.choices) ? payload.choices : [];
-        const firstChoice = choices[0];
-        const delta = firstChoice?.delta;
-        if (delta?.role === "assistant") {
-          role = "assistant";
-        }
-        if (typeof delta?.content === "string") {
-          content += delta.content;
-        }
-        if (typeof firstChoice?.finish_reason === "string" || firstChoice?.finish_reason === null) {
-          finishReason = firstChoice.finish_reason;
-        }
-        const payloadUsage = payload.usage;
-        if (payloadUsage) {
-          const promptTokens = typeof payloadUsage.prompt_tokens === "number" ? payloadUsage.prompt_tokens : 0;
-          const completionTokens = typeof payloadUsage.completion_tokens === "number" ? payloadUsage.completion_tokens : 0;
-          usage = {
-            prompt_tokens: promptTokens,
-            completion_tokens: completionTokens,
-            total_tokens: typeof payloadUsage.total_tokens === "number" ? payloadUsage.total_tokens : promptTokens + completionTokens
-          };
-        }
-      }
-    }
-  } finally {
-    reader.releaseLock();
-  }
-  const response = {
-    id,
-    object: "chat.completion",
-    created,
-    model,
-    choices: [
-      {
-        index: 0,
-        message: { role, content },
-        finish_reason: finishReason
-      }
-    ],
-    usage
-  };
-  if (metadata && Object.keys(metadata).length > 0) {
-    response._rouva = metadata;
-  }
-  return response;
-}
-function getRouvaMetadata(headers) {
-  const modelUsed = headers.get("x-rouva-model") ?? void 0;
-  const providerUsed = headers.get("x-rouva-provider") ?? void 0;
-  const taskType = headers.get("x-rouva-task") ?? void 0;
-  const cache = headers.get("x-rouva-cache") ?? void 0;
-  if (!modelUsed && !providerUsed && !taskType && !cache) return void 0;
-  return {
-    model_used: modelUsed,
-    provider_used: providerUsed,
-    task_type: taskType,
-    cache
-  };
-}
-
 // src/client.ts
 var DEFAULT_BASE_URL = "https://app.rouva.io";
 var Rouva = class {
@@ -313,7 +46,7 @@ var Rouva = class {
       method: "POST",
       headers: {
         "Content-Type": "application/json",
-        "Authorization": `Bearer ${this.apiKey}`
+        "x-api-key": this.apiKey
       },
       body: JSON.stringify(params)
     });
@@ -321,10 +54,11 @@ var Rouva = class {
       const body = await res.text().catch(() => res.statusText);
       throw new Error(`[Rouva] Gateway error ${res.status}: ${body}`);
     }
-    if (!res.body) throw new Error("[Rouva] No response body from gateway");
-    const normalizedStream = normalizeGatewayStream(res.body);
-    if (params.stream) return normalizedStream;
-    return readChatCompletionFromSse(normalizedStream, getRouvaMetadata(res.headers));
+    if (params.stream) {
+      if (!res.body) throw new Error("[Rouva] No response body for streaming request");
+      return res.body;
+    }
+    return res.json();
   }
 };
 // Annotate the CommonJS export names for ESM import in node:

package/dist/index.mjs

@@ -1,270 +1,3 @@
-// src/sse.ts
-function parseSseEvent(line) {
-  const raw = line.trim();
-  if (!raw.startsWith("data:")) return null;
-  const data = raw.slice(5).trim();
-  if (!data) return null;
-  try {
-    return { raw, data, json: data === "[DONE]" ? null : JSON.parse(data) };
-  } catch {
-    return { raw, data, json: null };
-  }
-}
-function stringifySse(payload) {
-  return `data: ${JSON.stringify(payload)}
-
-`;
-}
-function detectProvider(payload) {
-  if (!payload || typeof payload !== "object") return null;
-  if ("choices" in payload || "usage" in payload) return "openai";
-  if ("type" in payload) return "anthropic";
-  return null;
-}
-function normalizeChatId(id) {
-  return id.startsWith("chatcmpl_") ? id : `chatcmpl_${id}`;
-}
-function toFinishReason(stopReason) {
-  if (!stopReason) return null;
-  if (stopReason === "end_turn" || stopReason === "stop_sequence") return "stop";
-  if (stopReason === "max_tokens") return "length";
-  return stopReason;
-}
-function initialState() {
-  return {
-    provider: null,
-    id: `chatcmpl_${Date.now()}`,
-    model: "unknown",
-    created: Math.floor(Date.now() / 1e3),
-    promptTokens: 0,
-    completionTokens: 0,
-    roleEmitted: false,
-    finalChunkEmitted: false
-  };
-}
-function normalizeAnthropicPayload(payload, state) {
-  const eventType = typeof payload.type === "string" ? payload.type : null;
-  const chunks = [];
-  if (eventType === "message_start") {
-    const message = payload.message ?? {};
-    const usage = message.usage ?? {};
-    state.id = normalizeChatId(typeof message.id === "string" ? message.id : state.id);
-    state.model = typeof message.model === "string" ? message.model : state.model;
-    state.promptTokens = typeof usage.input_tokens === "number" ? usage.input_tokens : 0;
-    if (!state.roleEmitted) {
-      chunks.push(stringifySse({
-        id: state.id,
-        object: "chat.completion.chunk",
-        created: state.created,
-        model: state.model,
-        choices: [{ index: 0, delta: { role: "assistant" }, finish_reason: null }]
-      }));
-      state.roleEmitted = true;
-    }
-  }
-  if (eventType === "content_block_delta") {
-    const delta = payload.delta ?? {};
-    if (delta.type === "text_delta" && typeof delta.text === "string") {
-      if (!state.roleEmitted) {
-        chunks.push(stringifySse({
-          id: state.id,
-          object: "chat.completion.chunk",
-          created: state.created,
-          model: state.model,
-          choices: [{ index: 0, delta: { role: "assistant" }, finish_reason: null }]
-        }));
-        state.roleEmitted = true;
-      }
-      chunks.push(stringifySse({
-        id: state.id,
-        object: "chat.completion.chunk",
-        created: state.created,
-        model: state.model,
-        choices: [{ index: 0, delta: { content: delta.text }, finish_reason: null }]
-      }));
-    }
-  }
-  if (eventType === "message_delta") {
-    const delta = payload.delta ?? {};
-    const usage = payload.usage ?? {};
-    state.completionTokens = typeof usage.output_tokens === "number" ? usage.output_tokens : state.completionTokens;
-    chunks.push(stringifySse({
-      id: state.id,
-      object: "chat.completion.chunk",
-      created: state.created,
-      model: state.model,
-      choices: [{ index: 0, delta: {}, finish_reason: toFinishReason(typeof delta.stop_reason === "string" ? delta.stop_reason : null) }],
-      usage: {
-        prompt_tokens: state.promptTokens,
-        completion_tokens: state.completionTokens,
-        total_tokens: state.promptTokens + state.completionTokens
-      }
-    }));
-    state.finalChunkEmitted = true;
-  }
-  if (eventType === "message_stop") {
-    chunks.push("data: [DONE]\n\n");
-  }
-  if (eventType === "error") {
-    chunks.push(stringifySse(payload));
-  }
-  return chunks;
-}
-function normalizeGatewayStream(stream) {
-  const encoder = new TextEncoder();
-  const decoder = new TextDecoder();
-  let buffer = "";
-  const state = initialState();
-  return new ReadableStream({
-    async start(controller) {
-      const reader = stream.getReader();
-      try {
-        while (true) {
-          const { done, value } = await reader.read();
-          if (done) break;
-          if (!value) continue;
-          buffer += decoder.decode(value, { stream: true });
-          const lines = buffer.split("\n");
-          buffer = lines.pop() ?? "";
-          for (const line of lines) {
-            const event = parseSseEvent(line);
-            if (!event) continue;
-            if (event.data === "[DONE]") {
-              controller.enqueue(encoder.encode("data: [DONE]\n\n"));
-              continue;
-            }
-            const provider = detectProvider(event.json);
-            if (provider && !state.provider) state.provider = provider;
-            if (state.provider === "anthropic" && event.json && typeof event.json === "object") {
-              for (const chunk of normalizeAnthropicPayload(event.json, state)) {
-                controller.enqueue(encoder.encode(chunk));
-              }
-              continue;
-            }
-            controller.enqueue(encoder.encode(`${event.raw}
-
-`));
-          }
-        }
-        if (buffer.trim()) {
-          const event = parseSseEvent(buffer);
-          if (event) {
-            if (event.data === "[DONE]") {
-              controller.enqueue(encoder.encode("data: [DONE]\n\n"));
-            } else if (state.provider === "anthropic" && event.json && typeof event.json === "object") {
-              for (const chunk of normalizeAnthropicPayload(event.json, state)) {
-                controller.enqueue(encoder.encode(chunk));
-              }
-            } else {
-              controller.enqueue(encoder.encode(`${event.raw}
-
-`));
-            }
-          }
-        }
-      } finally {
-        controller.close();
-      }
-    }
-  });
-}
-async function readChatCompletionFromSse(stream, metadata) {
-  const reader = stream.getReader();
-  const decoder = new TextDecoder();
-  let buffer = "";
-  let id = `chatcmpl_${Date.now()}`;
-  let model = metadata?.model_used ?? "unknown";
-  let created = Math.floor(Date.now() / 1e3);
-  let role = "assistant";
-  let content = "";
-  let finishReason = null;
-  let usage = {
-    prompt_tokens: 0,
-    completion_tokens: 0,
-    total_tokens: 0
-  };
-  try {
-    while (true) {
-      const { done, value } = await reader.read();
-      if (done) break;
-      if (!value) continue;
-      buffer += decoder.decode(value, { stream: true });
-      const lines = buffer.split("\n");
-      buffer = lines.pop() ?? "";
-      for (const line of lines) {
-        const event = parseSseEvent(line);
-        if (!event || event.data === "[DONE]") continue;
-        if (!event.json || typeof event.json !== "object") continue;
-        const payload = event.json;
-        if (payload.type === "error" && payload.error) {
-          const error = payload.error;
-          throw new Error(
-            `[Rouva] Upstream error in stream: ${typeof error.message === "string" ? error.message : "Unknown error"}`
-          );
-        }
-        if (typeof payload.id === "string") id = payload.id;
-        if (typeof payload.model === "string") model = payload.model;
-        if (typeof payload.created === "number") created = payload.created;
-        const choices = Array.isArray(payload.choices) ? payload.choices : [];
-        const firstChoice = choices[0];
-        const delta = firstChoice?.delta;
-        if (delta?.role === "assistant") {
-          role = "assistant";
-        }
-        if (typeof delta?.content === "string") {
-          content += delta.content;
-        }
-        if (typeof firstChoice?.finish_reason === "string" || firstChoice?.finish_reason === null) {
-          finishReason = firstChoice.finish_reason;
-        }
-        const payloadUsage = payload.usage;
-        if (payloadUsage) {
-          const promptTokens = typeof payloadUsage.prompt_tokens === "number" ? payloadUsage.prompt_tokens : 0;
-          const completionTokens = typeof payloadUsage.completion_tokens === "number" ? payloadUsage.completion_tokens : 0;
-          usage = {
-            prompt_tokens: promptTokens,
-            completion_tokens: completionTokens,
-            total_tokens: typeof payloadUsage.total_tokens === "number" ? payloadUsage.total_tokens : promptTokens + completionTokens
-          };
-        }
-      }
-    }
-  } finally {
-    reader.releaseLock();
-  }
-  const response = {
-    id,
-    object: "chat.completion",
-    created,
-    model,
-    choices: [
-      {
-        index: 0,
-        message: { role, content },
-        finish_reason: finishReason
-      }
-    ],
-    usage
-  };
-  if (metadata && Object.keys(metadata).length > 0) {
-    response._rouva = metadata;
-  }
-  return response;
-}
-function getRouvaMetadata(headers) {
-  const modelUsed = headers.get("x-rouva-model") ?? void 0;
-  const providerUsed = headers.get("x-rouva-provider") ?? void 0;
-  const taskType = headers.get("x-rouva-task") ?? void 0;
-  const cache = headers.get("x-rouva-cache") ?? void 0;
-  if (!modelUsed && !providerUsed && !taskType && !cache) return void 0;
-  return {
-    model_used: modelUsed,
-    provider_used: providerUsed,
-    task_type: taskType,
-    cache
-  };
-}
-
 // src/client.ts
 var DEFAULT_BASE_URL = "https://app.rouva.io";
 var Rouva = class {
@@ -287,7 +20,7 @@ var Rouva = class {
       method: "POST",
       headers: {
         "Content-Type": "application/json",
-        "Authorization": `Bearer ${this.apiKey}`
+        "x-api-key": this.apiKey
       },
       body: JSON.stringify(params)
     });
@@ -295,10 +28,11 @@ var Rouva = class {
       const body = await res.text().catch(() => res.statusText);
       throw new Error(`[Rouva] Gateway error ${res.status}: ${body}`);
     }
-    if (!res.body) throw new Error("[Rouva] No response body from gateway");
-    const normalizedStream = normalizeGatewayStream(res.body);
-    if (params.stream) return normalizedStream;
-    return readChatCompletionFromSse(normalizedStream, getRouvaMetadata(res.headers));
+    if (params.stream) {
+      if (!res.body) throw new Error("[Rouva] No response body for streaming request");
+      return res.body;
+    }
+    return res.json();
   }
 };
 export {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@rouvanpm/rouva",
-  "version": "0.1.4",
+  "version": "0.2.0",
   "description": "Official Node.js SDK for Rouva — managed AI gateway with intelligent routing and spend tracking",
   "main": "dist/index.js",
   "module": "dist/index.mjs",
@@ -38,4 +38,4 @@
     "typescript": "^5.0.0",
     "vitest": "^1.0.0"
   }
-}
+}