@rouvanpm/rouva 0.1.0 → 0.1.3

package/README.md

@@ -5,13 +5,13 @@ Official Node.js SDK for [Rouva](https://rouva.io) — managed AI gateway with i
 ## Installation
 
 ```bash
-npm install rouva
+npm install @rouvanpm/rouva
 ```
 
 ## Quick Start
 
 ```typescript
-import { Rouva } from 'rouva'
+import { Rouva } from '@rouvanpm/rouva'
 
 const rouva = new Rouva({ apiKey: 'rva_...' })
 
@@ -22,7 +22,31 @@ const response = await rouva.chat.completions.create({
 console.log(response.choices[0].message.content)
 ```
 
-## Drop-in replacement for OpenAI
+## Provider agnostic
+
+Rouva works with all connected providers — Anthropic, OpenAI, Gemini, DeepSeek, and Mistral. You can request a specific model, force a specific provider, or omit both and let Rouva route to the cheapest capable model automatically.
+
+```typescript
+// Request a specific model
+const res = await rouva.chat.completions.create({
+  model: 'gpt-4o',
+  messages,
+})
+
+// Force a specific provider + model
+const res = await rouva.chat.completions.create({
+  provider: 'gemini',
+  model: 'gemini-2.5-pro',
+  messages,
+})
+
+// Let Rouva decide — routes to cheapest model for the task
+const res = await rouva.chat.completions.create({
+  messages,
+})
+```
+
+## OpenAI-style request shape
 
 ```typescript
 // Before
@@ -31,7 +55,7 @@ const openai = new OpenAI({ apiKey: '...' })
 const res = await openai.chat.completions.create({ messages, model: 'gpt-4o' })
 
 // After — Rouva routes to the cheapest capable model automatically
-import { Rouva } from 'rouva'
+import { Rouva } from '@rouvanpm/rouva'
 const rouva = new Rouva({ apiKey: 'rva_...' })
 const res = await rouva.chat.completions.create({ messages })
 ```
@@ -54,6 +78,8 @@ while (true) {
 }
 ```
 
+Streaming responses are normalized to OpenAI-style SSE chunks, even when Rouva routes the request to Anthropic.
+
 ## Options
 
 ```typescript
@@ -65,7 +91,7 @@ const rouva = new Rouva({
 
 ## Response metadata
 
-Every response includes a `_rouva` field with routing and cost details:
+Parsed non-stream responses may include a `_rouva` field with gateway header metadata when available:
 
 ```typescript
 const res = await rouva.chat.completions.create({ messages })
@@ -73,9 +99,7 @@ const res = await rouva.chat.completions.create({ messages })
 console.log(res._rouva)
 // {
 //   model_used: 'gpt-4o-mini',
-//   cost: 0.000012,
-//   savings: 0.000088,
-//   intelligently_routed: true,
+//   provider_used: 'openai',
 //   task_type: 'summarize'
 // }
 ```
@@ -88,4 +112,4 @@ console.log(res._rouva)
 
 ## License
 
-MIT
+MIT

package/dist/index.d.mts

@@ -8,10 +8,24 @@ interface Message {
     role: 'system' | 'user' | 'assistant';
     content: string;
 }
+type RouvaProvider = 'anthropic' | 'openai' | 'gemini' | 'deepseek' | 'mistral' | (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' | (string & {});
 interface ChatCompletionParams {
     messages: Message[];
-    /** Target model — omit to let Rouva route intelligently */
-    model?: string;
+    /**
+     * Target model — omit to let Rouva route intelligently to the cheapest capable model.
+     * Supports models from any connected provider (Anthropic, OpenAI).
+     */
+    model?: RouvaModel;
+    /**
+     * Force an exact provider when paired with `model`.
+     * Omit to let Rouva auto-route based on your connected keys.
+     */
+    provider?: RouvaProvider;
     /** Maximum tokens to generate */
     max_tokens?: number;
     /** Sampling temperature 0–1 */
@@ -40,16 +54,14 @@ interface ChatCompletion {
     _rouva?: RouvaResponseMeta;
 }
 interface RouvaResponseMeta {
-    /** 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;
+    /** 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;
 }
 
 declare class Rouva {
@@ -64,4 +76,4 @@ declare class Rouva {
     private _createChatCompletion;
 }
 
-export { type ChatCompletion, type ChatCompletionChoice, type ChatCompletionParams, type ChatCompletionUsage, type Message, Rouva, type RouvaOptions, type RouvaResponseMeta };
+export { type ChatCompletion, type ChatCompletionChoice, type ChatCompletionParams, type ChatCompletionUsage, type Message, Rouva, type RouvaOptions, type RouvaProvider, type RouvaResponseMeta };

package/dist/index.d.ts

@@ -8,10 +8,24 @@ interface Message {
     role: 'system' | 'user' | 'assistant';
     content: string;
 }
+type RouvaProvider = 'anthropic' | 'openai' | 'gemini' | 'deepseek' | 'mistral' | (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' | (string & {});
 interface ChatCompletionParams {
     messages: Message[];
-    /** Target model — omit to let Rouva route intelligently */
-    model?: string;
+    /**
+     * Target model — omit to let Rouva route intelligently to the cheapest capable model.
+     * Supports models from any connected provider (Anthropic, OpenAI).
+     */
+    model?: RouvaModel;
+    /**
+     * Force an exact provider when paired with `model`.
+     * Omit to let Rouva auto-route based on your connected keys.
+     */
+    provider?: RouvaProvider;
     /** Maximum tokens to generate */
     max_tokens?: number;
     /** Sampling temperature 0–1 */
@@ -40,16 +54,14 @@ interface ChatCompletion {
     _rouva?: RouvaResponseMeta;
 }
 interface RouvaResponseMeta {
-    /** 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;
+    /** 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;
 }
 
 declare class Rouva {
@@ -64,4 +76,4 @@ declare class Rouva {
     private _createChatCompletion;
 }
 
-export { type ChatCompletion, type ChatCompletionChoice, type ChatCompletionParams, type ChatCompletionUsage, type Message, Rouva, type RouvaOptions, type RouvaResponseMeta };
+export { type ChatCompletion, type ChatCompletionChoice, type ChatCompletionParams, type ChatCompletionUsage, type Message, Rouva, type RouvaOptions, type RouvaProvider, type RouvaResponseMeta };

package/dist/index.js

@@ -24,6 +24,273 @@ __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 {
@@ -46,7 +313,7 @@ var Rouva = class {
       method: "POST",
       headers: {
         "Content-Type": "application/json",
-        "x-api-key": this.apiKey
+        "Authorization": `Bearer ${this.apiKey}`
       },
       body: JSON.stringify(params)
     });
@@ -54,11 +321,10 @@ var Rouva = class {
       const body = await res.text().catch(() => res.statusText);
       throw new Error(`[Rouva] Gateway error ${res.status}: ${body}`);
     }
-    if (params.stream) {
-      if (!res.body) throw new Error("[Rouva] No response body for streaming request");
-      return res.body;
-    }
-    return res.json();
+    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));
   }
 };
 // Annotate the CommonJS export names for ESM import in node:

package/dist/index.mjs

@@ -1,3 +1,270 @@
+// 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 {
@@ -20,7 +287,7 @@ var Rouva = class {
       method: "POST",
       headers: {
         "Content-Type": "application/json",
-        "x-api-key": this.apiKey
+        "Authorization": `Bearer ${this.apiKey}`
       },
       body: JSON.stringify(params)
     });
@@ -28,11 +295,10 @@ var Rouva = class {
       const body = await res.text().catch(() => res.statusText);
       throw new Error(`[Rouva] Gateway error ${res.status}: ${body}`);
     }
-    if (params.stream) {
-      if (!res.body) throw new Error("[Rouva] No response body for streaming request");
-      return res.body;
-    }
-    return res.json();
+    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));
   }
 };
 export {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@rouvanpm/rouva",
-  "version": "0.1.0",
+  "version": "0.1.3",
   "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"
   }
-}
+}