RELEASING: Releasing 1 package(s)

Releases: create-llama@0.0.28 [skip ci]
fix: add missing check env logic (#636 )
2026-07-15 06:52:45 -04:00 · 2024-03-12 13:41:36 +07:00 · 2024-03-12 12:29:00 +07:00 · 2024-03-12 11:10:31 +07:00 · 2024-03-12 09:22:21 +07:00 · 2024-03-11 14:16:15 +07:00
120 changed files with 1742 additions and 402 deletions
@@ -0,0 +1,5 @@
+---
+"llamaindex": patch
+---
+
+feat: experimental package + json query engine
@@ -1,5 +0,0 @@
---
-"create-llama": patch
---
-
-Add LlamaParse option when selecting a pdf file or a folder
@@ -0,0 +1,12 @@
+---
+"llamaindex": patch
+"@llamaindex/core-test": patch
+---
+
+- Add missing exports:
+  - `IndexStructType`,
+  - `IndexDict`,
+  - `jsonToIndexStruct`,
+  - `IndexList`,
+  - `IndexStruct`
+- Fix `IndexDict.toJson()` method
@@ -0,0 +1,5 @@
+---
+"llamaindex": patch
+---
+
+Add streaming to agents
@@ -0,0 +1,5 @@
+---
+"llamaindex": minor
+---
+
+Use parameter object for retrieve function of Retriever (to align usage with query function of QueryEngine)
@@ -11,5 +11,13 @@ module.exports = {
    "max-params": ["error", 4],
    "prefer-const": "error",
  },
+  overrides: [
+    {
+      files: ["examples/**/*.ts"],
+      rules: {
+        "turbo/no-undeclared-env-vars": "off",
+      },
+    },
+  ],
  ignorePatterns: ["dist/", "lib/"],
 };
@@ -0,0 +1,28 @@
+name: Publish
+on:
+  push:
+    branches:
+      - main
+
+jobs:
+  publish:
+    runs-on: ubuntu-latest
+
+    permissions:
+      contents: read
+      id-token: write
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Publish @llamaindex/env
+        run: npx jsr publish
+        working-directory: packages/env
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Publish @llamaindex/core
+        run: npx jsr publish --allow-slow-types
+        working-directory: packages/core
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
@@ -45,6 +45,7 @@ playwright-report/
 blob-report/
 playwright/.cache/
 .tsbuildinfo
+packages/create-llama/e2e/cache

 # intellij
 **/.idea
@@ -4,3 +4,4 @@ pnpm-lock.yaml
 lib/
 dist/
 .docusaurus/
+packages/create-llama/e2e/cache/
@@ -1,6 +1,6 @@
 # Transformations

-A transformation is something that takes a list of nodes as an input, and returns a list of nodes. Each component that implements the Transformatio class has both a `transform` definition responsible for transforming the nodes
+A transformation is something that takes a list of nodes as an input, and returns a list of nodes. Each component that implements the Transformation class has both a `transform` definition responsible for transforming the nodes.

 Currently, the following components are Transformation objects:

@@ -100,7 +100,7 @@ const response = await queryEngine.query("<user_query>");
 ```ts
 import { SimilarityPostprocessor } from "llamaindex";

-nodes = await index.asRetriever().retrieve("test query str");
+nodes = await index.asRetriever().retrieve({ query: "test query str" });

 const processor = new SimilarityPostprocessor({
  similarityCutoff: 0.7,
@@ -11,7 +11,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Fetch nodes!
-const nodesWithScore = await retriever.retrieve("query string");
+const nodesWithScore = await retriever.retrieve({ query: "query string" });
 ```

 ## API Reference
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // جلب العقد!
-const nodesWithScore = await retriever.retrieve("سلسلة الاستعلام");
+const nodesWithScore = await retriever.retrieve({ query: "سلسلة الاستعلام" });
 ```

 ## مرجع الواجهة البرمجية (API Reference)
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Извличане на върхове!
-const nodesWithScore = await retriever.retrieve("query string");
+const nodesWithScore = await retriever.retrieve({ query: "query string" });
 ```

 ## API Reference (API справка)
@@ -13,7 +13,7 @@ const recuperador = vector_index.asRetriever();
 recuperador.similarityTopK = 3;

 // Obteniu els nodes!
-const nodesAmbPuntuació = await recuperador.retrieve("cadena de consulta");
+const nodesAmbPuntuació = await recuperador.retrieve({ query: "cadena de consulta" });
 ```

 ## Referència de l'API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Získání uzlů!
-const nodesWithScore = await retriever.retrieve("dotazovací řetězec");
+const nodesWithScore = await retriever.retrieve({ query: "dotazovací řetězec" });
 ```

 ## API Reference (Odkazy na rozhraní)
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Hent noder!
-const nodesWithScore = await retriever.retrieve("forespørgselsstreng");
+const nodesWithScore = await retriever.retrieve({ query: "forespørgselsstreng" });
 ```

 ## API Reference
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Knoten abrufen!
-const nodesWithScore = await retriever.retrieve("Abfragezeichenfolge");
+const nodesWithScore = await retriever.retrieve({ query: "Abfragezeichenfolge" });
 ```

 ## API-Referenz
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Ανάκτηση κόμβων!
-const nodesWithScore = await retriever.retrieve("συμβολοσειρά ερωτήματος");
+const nodesWithScore = await retriever.retrieve({ query: "συμβολοσειρά ερωτήματος" });
 ```

 ## Αναφορά API
@@ -13,7 +13,7 @@ const recuperador = vector_index.asRetriever();
 recuperador.similarityTopK = 3;

 // ¡Obtener nodos!
-const nodosConPuntuación = await recuperador.retrieve("cadena de consulta");
+const nodosConPuntuación = await recuperador.retrieve({ query: "cadena de consulta" });
 ```

 ## Referencia de la API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Too sõlmed!
-const nodesWithScore = await retriever.retrieve("päringu string");
+const nodesWithScore = await retriever.retrieve({ query: "päringu string" });
 ```

 ## API viide
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // بازیابی گره ها!
-const nodesWithScore = await retriever.retrieve("رشته پرس و جو");
+const nodesWithScore = await retriever.retrieve({ query: "رشته پرس و جو" });
 ```

 ## مرجع API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Hae solmut!
-const nodesWithScore = await retriever.retrieve("kyselymerkkijono");
+const nodesWithScore = await retriever.retrieve({ query: "kyselymerkkijono" });
 ```

 ## API-viite
@@ -11,7 +11,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Récupérer les nœuds !
-const nodesWithScore = await retriever.retrieve("chaîne de requête");
+const nodesWithScore = await retriever.retrieve({ query: "chaîne de requête" });
 ```

 ## Référence de l'API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // אחזור צמתים!
-const nodesWithScore = await retriever.retrieve("מחרוזת שאילתה");
+const nodesWithScore = await retriever.retrieve({ query: "מחרוזת שאילתה" });
 ```

 ## מדריך לממשק API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // नोड्स प्राप्त करें!
-const nodesWithScore = await retriever.retrieve("क्वेरी स्ट्रिंग");
+const nodesWithScore = await retriever.retrieve({ query: "क्वेरी स्ट्रिंग" });
 ```

 ## एपीआई संदर्भ (API Reference)
@@ -13,7 +13,7 @@ const dohvatnik = vector_index.asRetriever();
 dohvatnik.similarityTopK = 3;

 // Dohvati čvorove!
-const čvoroviSaRezultatom = await dohvatnik.retrieve("upitni niz");
+const čvoroviSaRezultatom = await dohvatnik.retrieve({ query: "upitni niz" });
 ```

 ## API Referenca
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Node-ok lekérése!
-const nodesWithScore = await retriever.retrieve("lekérdezési karakterlánc");
+const nodesWithScore = await retriever.retrieve({ query: "lekérdezési karakterlánc" });
 ```

 ## API Referencia
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Mengambil node!
-const nodesWithScore = await retriever.retrieve("string query");
+const nodesWithScore = await retriever.retrieve({ query: "string query" });
 ```

 ## Referensi API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Recupera i nodi!
-const nodesWithScore = await retriever.retrieve("stringa di query");
+const nodesWithScore = await retriever.retrieve({ query: "stringa di query" });
 ```

 ## Riferimento API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // ノードを取得します！
-const nodesWithScore = await retriever.retrieve("クエリ文字列");
+const nodesWithScore = await retriever.retrieve({ query: "クエリ文字列" });
 ```

 ## API リファレンス
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // 노드를 가져옵니다!
-const nodesWithScore = await retriever.retrieve("쿼리 문자열");
+const nodesWithScore = await retriever.retrieve({ query: "쿼리 문자열" });
 ```

 ## API 참조
@@ -13,7 +13,7 @@ const gavėjas = vector_index.asRetriever();
 gavėjas.similarityTopK = 3;

 // Išgaunami mazgai!
-const mazgaiSuRezultatu = await gavėjas.retrieve("užklausos eilutė");
+const mazgaiSuRezultatu = await gavėjas.retrieve({ query: "užklausos eilutė" });
 ```

 ## API nuorodos (API Reference)
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Haal knooppunten op!
-const nodesWithScore = await retriever.retrieve("zoekopdracht");
+const nodesWithScore = await retriever.retrieve({ query: "zoekopdracht" });
 ```

 ## API Referentie
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Hent noder!
-const nodesWithScore = await retriever.retrieve("spørringsstreng");
+const nodesWithScore = await retriever.retrieve({ query: "spørringsstreng" });
 ```

 ## API-referanse
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Pobierz węzły!
-const nodesWithScore = await retriever.retrieve("ciąg zapytania");
+const nodesWithScore = await retriever.retrieve({ query: "ciąg zapytania" });
 ```

 ## Dokumentacja interfejsu API
@@ -13,7 +13,7 @@ const recuperador = vector_index.asRetriever();
 recuperador.similarityTopK = 3;

 // Buscar nós!
-const nósComPontuação = await recuperador.retrieve("string de consulta");
+const nósComPontuação = await recuperador.retrieve({ query: "string de consulta" });
 ```

 ## Referência da API
@@ -13,7 +13,7 @@ const recuperator = vector_index.asRetriever();
 recuperator.similarityTopK = 3;

 // Preia nodurile!
-const noduriCuScor = await recuperator.retrieve("șir de interogare");
+const noduriCuScor = await recuperator.retrieve({ query: "șir de interogare" });
 ```

 ## Referință API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Получение узлов!
-const nodesWithScore = await retriever.retrieve("строка запроса");
+const nodesWithScore = await retriever.retrieve({ query: "строка запроса" });
 ```

 ## Справочник по API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Dohvati čvorove!
-const nodesWithScore = await retriever.retrieve("upitni niz");
+const nodesWithScore = await retriever.retrieve({ query: "upitni niz" });
 ```

 ## API Referenca
@@ -13,7 +13,7 @@ const pridobitelj = vector_index.asRetriever();
 pridobitelj.similarityTopK = 3;

 // Pridobivanje vozlišč!
-const vozliščaZRezultatom = await pridobitelj.retrieve("poizvedbeni niz");
+const vozliščaZRezultatom = await pridobitelj.retrieve({ query: "poizvedbeni niz" });
 ```

 ## API Sklic
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Získajte uzly!
-const nodesWithScore = await retriever.retrieve("reťazec dotazu");
+const nodesWithScore = await retriever.retrieve({ query: "reťazec dotazu" });
 ```

 ## API Referencia
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Hämta noder!
-const nodesWithScore = await retriever.retrieve("frågesträng");
+const nodesWithScore = await retriever.retrieve({ query: "frågesträng" });
 ```

 ## API-referens
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // เรียกคืนโหนด!
-const nodesWithScore = await retriever.retrieve("query string");
+const nodesWithScore = await retriever.retrieve({ query: "query string" });
 ```

 ## API Reference (การอ้างอิง API)
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Düğümleri getir!
-const nodesWithScore = await retriever.retrieve("sorgu dizesi");
+const nodesWithScore = await retriever.retrieve({ query: "sorgu dizesi" });
 ```

 ## API Referansı
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Отримати вузли!
-const nodesWithScore = await retriever.retrieve("рядок запиту");
+const nodesWithScore = await retriever.retrieve({ query: "рядок запиту" });
 ```

 ## Довідник API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Lấy các node!
-const nodesWithScore = await retriever.retrieve("chuỗi truy vấn");
+const nodesWithScore = await retriever.retrieve({ query: "chuỗi truy vấn" });
 ```

 ## Tài liệu tham khảo API
@@ -11,7 +11,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // 获取节点！
-const nodesWithScore = await retriever.retrieve("查询字符串");
+const nodesWithScore = await retriever.retrieve({ query: "查询字符串" });
 ```

 ## API 参考
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // 提取節點！
-const nodesWithScore = await retriever.retrieve("查詢字串");
+const nodesWithScore = await retriever.retrieve({ query: "查詢字串" });
 ```

 ## API 參考
@@ -0,0 +1,77 @@
+import { FunctionTool, OpenAIAgent } from "llamaindex";
+
+// Define a function to sum two numbers
+function sumNumbers({ a, b }: { a: number; b: number }): number {
+  return a + b;
+}
+
+// Define a function to divide two numbers
+function divideNumbers({ a, b }: { a: number; b: number }): number {
+  return a / b;
+}
+
+// Define the parameters of the sum function as a JSON schema
+const sumJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The first number",
+    },
+    b: {
+      type: "number",
+      description: "The second number",
+    },
+  },
+  required: ["a", "b"],
+};
+
+const divideJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The dividend",
+    },
+    b: {
+      type: "number",
+      description: "The divisor",
+    },
+  },
+  required: ["a", "b"],
+};
+
+async function main() {
+  // Create a function tool from the sum function
+  const functionTool = new FunctionTool(sumNumbers, {
+    name: "sumNumbers",
+    description: "Use this function to sum two numbers",
+    parameters: sumJSON,
+  });
+
+  // Create a function tool from the divide function
+  const functionTool2 = new FunctionTool(divideNumbers, {
+    name: "divideNumbers",
+    description: "Use this function to divide two numbers",
+    parameters: divideJSON,
+  });
+
+  // Create an OpenAIAgent with the function tools
+  const agent = new OpenAIAgent({
+    tools: [functionTool, functionTool2],
+    verbose: false,
+  });
+
+  const stream = await agent.chat({
+    message: "Divide 16 by 2 then add 20",
+    stream: true,
+  });
+
+  for await (const chunk of stream.response) {
+    process.stdout.write(chunk.response);
+  }
+}
+
+main().then(() => {
+  console.log("\nDone");
+});
@@ -27,9 +27,9 @@ async function main() {
  // retrieve documents using the index
  const index = await createIndex();
  const retriever = index.asRetriever({ similarityTopK: 3 });
-  const results = await retriever.retrieve(
-    "what are Vincent van Gogh's famous paintings",
-  );
+  const results = await retriever.retrieve({
+    query: "what are Vincent van Gogh's famous paintings",
+  });
  for (const result of results) {
    const node = result.node;
    if (!node) {
@@ -13,8 +13,8 @@
    "type-check": "tsc -b --diagnostics",
    "release": "pnpm run build:release && changeset publish",
    "new-llamaindex": "pnpm run build:release && changeset version --ignore create-llama",
-    "new-create-llama": "pnpm run build:release && changeset version --ignore llamaindex",
-    "new-snapshots": "pnpm run build:release && changeset version --snapshot"
+    "new-create-llama": "pnpm run build:release && changeset version --ignore llamaindex --ignore @llamaindex/core-test",
+    "new-experimental": "pnpm run build:release && changeset version --ignore create-llama"
  },
  "devDependencies": {
    "@changesets/cli": "^2.27.1",
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/core",
-  "version": "0.1.20",
+  "version": "0.1.21",
  "exports": "./src/index.ts",
  "imports": {
    "@llamaindex/env": "jsr:@llamaindex/env@0.0.5"
@@ -92,9 +92,9 @@
  "scripts": {
    "lint": "eslint .",
    "build": "rm -rf ./dist && pnpm run build:esm && pnpm run build:cjs && pnpm run build:type",
-    "build:esm": "swc src -d dist --strip-leading-paths --config-file .swcrc",
-    "build:cjs": "swc src -d dist/cjs --strip-leading-paths --config-file .cjs.swcrc",
-    "build:type": "pnpm run -w type-check",
+    "build:esm": "swc src -d dist --strip-leading-paths --config-file ../../.swcrc",
+    "build:cjs": "swc src -d dist/cjs --strip-leading-paths --config-file ../../.cjs.swcrc",
+    "build:type": "tsc -p tsconfig.json",
    "copy": "cp -r ../../README.md ../../LICENSE .",
    "postbuild": "pnpm run copy && node -e \"require('fs').writeFileSync('./dist/cjs/package.json', JSON.stringify({ type: 'commonjs' }))\"",
    "circular-check": "madge -c ./src/index.ts",
@@ -2,14 +2,16 @@ import type { Event } from "./callbacks/CallbackManager.js";
 import type { NodeWithScore } from "./Node.js";
 import type { ServiceContext } from "./ServiceContext.js";

+export type RetrieveParams = {
+  query: string;
+  parentEvent?: Event;
+  preFilters?: unknown;
+};
+
 /**
 * Retrievers retrieve the nodes that most closely match our query in similarity.
 */
 export interface BaseRetriever {
-  retrieve(
-    query: string,
-    parentEvent?: Event,
-    preFilters?: unknown,
-  ): Promise<NodeWithScore[]>;
+  retrieve(params: RetrieveParams): Promise<NodeWithScore[]>;
  getServiceContext(): ServiceContext;
 }
@@ -1,10 +1,10 @@
-// Assuming that the necessary interfaces and classes (like BaseTool, OpenAI, ChatMessage, CallbackManager, etc.) are defined elsewhere
-
 import { randomUUID } from "@llamaindex/env";
+import { Response } from "../../Response.js";
 import type { CallbackManager } from "../../callbacks/CallbackManager.js";
 import {
  AgentChatResponse,
  ChatResponseMode,
+  StreamingAgentChatResponse,
 } from "../../engines/chat/types.js";
 import type {
  ChatMessage,
@@ -12,6 +12,7 @@ import type {
  ChatResponseChunk,
 } from "../../llm/index.js";
 import { OpenAI } from "../../llm/index.js";
+import { streamConverter, streamReducer } from "../../llm/utils.js";
 import { ChatMemoryBuffer } from "../../memory/ChatMemoryBuffer.js";
 import type { ObjectRetriever } from "../../objects/base.js";
 import type { ToolOutput } from "../../tools/types.js";
@@ -192,13 +193,40 @@ export class OpenAIAgentWorker implements AgentWorker {
  private _processMessage(
    task: Task,
    chatResponse: ChatResponse,
-  ): AgentChatResponse | AsyncIterable<ChatResponseChunk> {
+  ): AgentChatResponse {
    const aiMessage = chatResponse.message;
    task.extraState.newMemory.put(aiMessage);

    return new AgentChatResponse(aiMessage.content, task.extraState.sources);
  }

+  private async _getStreamAiResponse(
+    task: Task,
+    llmChatKwargs: any,
+  ): Promise<StreamingAgentChatResponse> {
+    const stream = await this.llm.chat({
+      stream: true,
+      ...llmChatKwargs,
+    });
+
+    const iterator = streamConverter(
+      streamReducer({
+        stream,
+        initialValue: "",
+        reducer: (accumulator, part) => (accumulator += part.delta),
+        finished: (accumulator) => {
+          task.extraState.newMemory.put({
+            content: accumulator,
+            role: "assistant",
+          });
+        },
+      }),
+      (r: ChatResponseChunk) => new Response(r.delta),
+    );
+
+    return new StreamingAgentChatResponse(iterator, task.extraState.sources);
+  }
+
  /**
   * Get agent response.
   * @param task: task
@@ -210,7 +238,7 @@ export class OpenAIAgentWorker implements AgentWorker {
    task: Task,
    mode: ChatResponseMode,
    llmChatKwargs: any,
-  ): Promise<AgentChatResponse> {
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse> {
    if (mode === ChatResponseMode.WAIT) {
      const chatResponse = (await this.llm.chat({
        stream: false,
@@ -218,9 +246,11 @@ export class OpenAIAgentWorker implements AgentWorker {
      })) as unknown as ChatResponse;

      return this._processMessage(task, chatResponse) as AgentChatResponse;
-    } else {
-      throw new Error("Not implemented");
+    } else if (mode === ChatResponseMode.STREAM) {
+      return this._getStreamAiResponse(task, llmChatKwargs);
    }
+
+    throw new Error("Invalid mode");
  }

  /**
@@ -4,6 +4,7 @@ import type { ChatEngineAgentParams } from "../../engines/chat/index.js";
 import {
  AgentChatResponse,
  ChatResponseMode,
+  StreamingAgentChatResponse,
 } from "../../engines/chat/index.js";
 import type { ChatMessage, LLM } from "../../llm/index.js";
 import { ChatMemoryBuffer } from "../../memory/ChatMemoryBuffer.js";
@@ -231,23 +232,26 @@ export class AgentRunner extends BaseAgentRunner {
    taskId: string,
    stepOutput: TaskStepOutput,
    kwargs?: any,
-  ): Promise<AgentChatResponse> {
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse> {
    if (!stepOutput) {
      stepOutput =
        this.getCompletedSteps(taskId)[
          this.getCompletedSteps(taskId).length - 1
        ];
    }
+
    if (!stepOutput.isLast) {
      throw new Error(
        "finalizeResponse can only be called on the last step output",
      );
    }

-    if (!(stepOutput.output instanceof AgentChatResponse)) {
-      throw new Error(
-        `When \`isLast\` is True, cur_step_output.output must be AGENT_CHAT_RESPONSE_TYPE: ${stepOutput.output}`,
-      );
+    if (!(stepOutput.output instanceof StreamingAgentChatResponse)) {
+      if (!(stepOutput.output instanceof AgentChatResponse)) {
+        throw new Error(
+          `When \`isLast\` is True, cur_step_output.output must be AGENT_CHAT_RESPONSE_TYPE: ${stepOutput.output}`,
+        );
+      }
    }

    this.agentWorker.finalizeTask(this.getTask(taskId), kwargs);
@@ -262,20 +266,32 @@ export class AgentRunner extends BaseAgentRunner {
  protected async _chat({
    message,
    toolChoice,
-  }: ChatEngineAgentParams & { mode: ChatResponseMode }) {
+    stream,
+  }: ChatEngineAgentParams): Promise<AgentChatResponse>;
+  protected async _chat({
+    message,
+    toolChoice,
+    stream,
+  }: ChatEngineAgentParams & {
+    stream: true;
+  }): Promise<StreamingAgentChatResponse>;
+  protected async _chat({
+    message,
+    toolChoice,
+    stream,
+  }: ChatEngineAgentParams): Promise<
+    AgentChatResponse | StreamingAgentChatResponse
+  > {
    const task = this.createTask(message as string);

    let resultOutput;

+    const mode = stream ? ChatResponseMode.STREAM : ChatResponseMode.WAIT;
+
    while (true) {
-      const curStepOutput = await this._runStep(
-        task.taskId,
-        undefined,
-        ChatResponseMode.WAIT,
-        {
-          toolChoice,
-        },
-      );
+      const curStepOutput = await this._runStep(task.taskId, undefined, mode, {
+        toolChoice,
+      });

      if (curStepOutput.isLast) {
        resultOutput = curStepOutput;
@@ -299,7 +315,26 @@ export class AgentRunner extends BaseAgentRunner {
    message,
    chatHistory,
    toolChoice,
-  }: ChatEngineAgentParams): Promise<AgentChatResponse> {
+    stream,
+  }: ChatEngineAgentParams & {
+    stream?: false;
+  }): Promise<AgentChatResponse>;
+  public async chat({
+    message,
+    chatHistory,
+    toolChoice,
+    stream,
+  }: ChatEngineAgentParams & {
+    stream: true;
+  }): Promise<StreamingAgentChatResponse>;
+  public async chat({
+    message,
+    chatHistory,
+    toolChoice,
+    stream,
+  }: ChatEngineAgentParams): Promise<
+    AgentChatResponse | StreamingAgentChatResponse
+  > {
    if (!toolChoice) {
      toolChoice = this.defaultToolChoice;
    }
@@ -308,7 +343,7 @@ export class AgentRunner extends BaseAgentRunner {
      message,
      chatHistory,
      toolChoice,
-      mode: ChatResponseMode.WAIT,
+      stream,
    });

    return chatResponse;
@@ -1,4 +1,7 @@
-import type { AgentChatResponse } from "../../engines/chat/index.js";
+import type {
+  AgentChatResponse,
+  StreamingAgentChatResponse,
+} from "../../engines/chat/index.js";
 import type { Task, TaskStep, TaskStepOutput } from "../types.js";
 import { BaseAgent } from "../types.js";

@@ -57,7 +60,7 @@ export abstract class BaseAgentRunner extends BaseAgent {
    taskId: string,
    stepOutput: TaskStepOutput,
    kwargs?: any,
-  ): Promise<AgentChatResponse>;
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse>;

  abstract undoStep(taskId: string): void;
 }
@@ -1,7 +1,9 @@
 import type {
  AgentChatResponse,
  ChatEngineAgentParams,
+  StreamingAgentChatResponse,
 } from "../engines/chat/index.js";
+
 import type { QueryEngineParamsNonStreaming } from "../types.js";

 export interface AgentWorker {
@@ -12,11 +14,15 @@ export interface AgentWorker {
 }

 interface BaseChatEngine {
-  chat(params: ChatEngineAgentParams): Promise<AgentChatResponse>;
+  chat(
+    params: ChatEngineAgentParams,
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse>;
 }

 interface BaseQueryEngine {
-  query(params: QueryEngineParamsNonStreaming): Promise<AgentChatResponse>;
+  query(
+    params: QueryEngineParamsNonStreaming,
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse>;
 }

 /**
@@ -31,7 +37,10 @@ export abstract class BaseAgent implements BaseChatEngine, BaseQueryEngine {
    return [];
  }

-  abstract chat(params: ChatEngineAgentParams): Promise<AgentChatResponse>;
+  abstract chat(
+    params: ChatEngineAgentParams,
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse>;
+
  abstract reset(): void;

  /**
@@ -41,7 +50,7 @@ export abstract class BaseAgent implements BaseChatEngine, BaseQueryEngine {
   */
  async query(
    params: QueryEngineParamsNonStreaming,
-  ): Promise<AgentChatResponse> {
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse> {
    // Handle non-streaming query
    const agentResponse = await this.chat({
      message: params.query,
@@ -3,7 +3,7 @@ import { RetrieverQueryEngine } from "../engines/query/RetrieverQueryEngine.js";
 import type { BaseNodePostprocessor } from "../postprocessors/types.js";
 import type { BaseSynthesizer } from "../synthesizers/types.js";
 import type { BaseQueryEngine } from "../types.js";
-import type { RetrieveParams } from "./LlamaCloudRetriever.js";
+import type { CloudRetrieveParams } from "./LlamaCloudRetriever.js";
 import { LlamaCloudRetriever } from "./LlamaCloudRetriever.js";
 import type { CloudConstructorParams } from "./types.js";

@@ -14,7 +14,7 @@ export class LlamaCloudIndex {
    this.params = params;
  }

-  asRetriever(params: RetrieveParams = {}): BaseRetriever {
+  asRetriever(params: CloudRetrieveParams = {}): BaseRetriever {
    return new LlamaCloudRetriever({ ...this.params, ...params });
  }

@@ -23,7 +23,7 @@ export class LlamaCloudIndex {
      responseSynthesizer?: BaseSynthesizer;
      preFilters?: unknown;
      nodePostprocessors?: BaseNodePostprocessor[];
-    } & RetrieveParams,
+    } & CloudRetrieveParams,
  ): BaseQueryEngine {
    const retriever = new LlamaCloudRetriever({
      ...this.params,
@@ -2,15 +2,14 @@ import type { PlatformApi, PlatformApiClient } from "@llamaindex/cloud";
 import { globalsHelper } from "../GlobalsHelper.js";
 import type { NodeWithScore } from "../Node.js";
 import { ObjectType, jsonToNode } from "../Node.js";
-import type { BaseRetriever } from "../Retriever.js";
+import type { BaseRetriever, RetrieveParams } from "../Retriever.js";
 import type { ServiceContext } from "../ServiceContext.js";
 import { serviceContextFromDefaults } from "../ServiceContext.js";
-import type { Event } from "../callbacks/CallbackManager.js";
 import type { ClientParams, CloudConstructorParams } from "./types.js";
 import { DEFAULT_PROJECT_NAME } from "./types.js";
 import { getClient } from "./utils.js";

-export type RetrieveParams = Omit<
+export type CloudRetrieveParams = Omit<
  PlatformApi.RetrievalParams,
  "query" | "searchFilters" | "pipelineId" | "className"
 > & { similarityTopK?: number };
@@ -18,7 +17,7 @@ export type RetrieveParams = Omit<
 export class LlamaCloudRetriever implements BaseRetriever {
  client?: PlatformApiClient;
  clientParams: ClientParams;
-  retrieveParams: RetrieveParams;
+  retrieveParams: CloudRetrieveParams;
  projectName: string = DEFAULT_PROJECT_NAME;
  pipelineName: string;
  serviceContext: ServiceContext;
@@ -35,7 +34,7 @@ export class LlamaCloudRetriever implements BaseRetriever {
    });
  }

-  constructor(params: CloudConstructorParams & RetrieveParams) {
+  constructor(params: CloudConstructorParams & CloudRetrieveParams) {
    this.clientParams = { apiKey: params.apiKey, baseUrl: params.baseUrl };
    if (params.similarityTopK) {
      params.denseSimilarityTopK = params.similarityTopK;
@@ -55,11 +54,11 @@ export class LlamaCloudRetriever implements BaseRetriever {
    return this.client;
  }

-  async retrieve(
-    query: string,
-    parentEvent?: Event | undefined,
-    preFilters?: unknown,
-  ): Promise<NodeWithScore[]> {
+  async retrieve({
+    query,
+    parentEvent,
+    preFilters,
+  }: RetrieveParams): Promise<NodeWithScore[]> {
    const pipelines = await (
      await this.getClient()
    ).pipeline.searchPipelines({
@@ -64,10 +64,10 @@ export class DefaultContextGenerator
        tags: ["final"],
      };
    }
-    const sourceNodesWithScore = await this.retriever.retrieve(
-      message,
+    const sourceNodesWithScore = await this.retriever.retrieve({
+      query: message,
      parentEvent,
-    );
+    });

    const nodes = await this.applyNodePostprocessors(
      sourceNodesWithScore,
@@ -27,6 +27,7 @@ export interface ChatEngineParamsNonStreaming extends ChatEngineParamsBase {

 export interface ChatEngineAgentParams extends ChatEngineParamsBase {
  toolChoice?: string | Record<string, any>;
+  stream?: boolean;
 }

 /**
@@ -86,3 +87,20 @@ export class AgentChatResponse {
    return this.response ?? "";
  }
 }
+
+export class StreamingAgentChatResponse {
+  response: AsyncIterable<Response>;
+
+  sources: ToolOutput[];
+  sourceNodes?: BaseNode[];
+
+  constructor(
+    response: AsyncIterable<Response>,
+    sources?: ToolOutput[],
+    sourceNodes?: BaseNode[],
+  ) {
+    this.response = response;
+    this.sources = sources ?? [];
+    this.sourceNodes = sourceNodes ?? [];
+  }
+}
@@ -63,11 +63,11 @@ export class RetrieverQueryEngine
  }

  private async retrieve(query: string, parentEvent: Event) {
-    const nodes = await this.retriever.retrieve(
+    const nodes = await this.retriever.retrieve({
      query,
      parentEvent,
-      this.preFilters,
-    );
+      preFilters: this.preFilters,
+    });

    return await this.applyNodePostprocessors(nodes, query);
  }
@@ -30,3 +30,4 @@ export * from "./selectors/index.js";
 export * from "./storage/index.js";
 export * from "./synthesizers/index.js";
 export * from "./tools/index.js";
+export * from "./types.js";
@@ -1,4 +1,6 @@
 export * from "./BaseIndex.js";
+export * from "./IndexStruct.js";
+export * from "./json-to-index-struct.js";
 export * from "./keyword/index.js";
 export * from "./summary/index.js";
 export * from "./vectorStore/index.js";
@@ -24,9 +24,15 @@ export class IndexDict extends IndexStruct {
  }

  toJson(): Record<string, unknown> {
+    const nodesDict: Record<string, unknown> = {};
+
+    for (const [key, node] of Object.entries(this.nodesDict)) {
+      nodesDict[key] = node.toJSON();
+    }
+
    return {
      ...super.toJson(),
-      nodesDict: this.nodesDict,
+      nodesDict,
      type: this.type,
    };
  }
@@ -8,7 +8,7 @@ import {
  defaultKeywordExtractPrompt,
  defaultQueryKeywordExtractPrompt,
 } from "../../Prompt.js";
-import type { BaseRetriever } from "../../Retriever.js";
+import type { BaseRetriever, RetrieveParams } from "../../Retriever.js";
 import type { ServiceContext } from "../../ServiceContext.js";
 import { serviceContextFromDefaults } from "../../ServiceContext.js";
 import { RetrieverQueryEngine } from "../../engines/query/index.js";
@@ -79,7 +79,7 @@ abstract class BaseKeywordTableRetriever implements BaseRetriever {

  abstract getKeywords(query: string): Promise<string[]>;

-  async retrieve(query: string): Promise<NodeWithScore[]> {
+  async retrieve({ query }: RetrieveParams): Promise<NodeWithScore[]> {
    const keywords = await this.getKeywords(query);
    const chunkIndicesCount: { [key: string]: number } = {};
    const filteredKeywords = keywords.filter((keyword) =>
@@ -3,10 +3,9 @@ import { globalsHelper } from "../../GlobalsHelper.js";
 import type { BaseNode, Document, NodeWithScore } from "../../Node.js";
 import type { ChoiceSelectPrompt } from "../../Prompt.js";
 import { defaultChoiceSelectPrompt } from "../../Prompt.js";
-import type { BaseRetriever } from "../../Retriever.js";
+import type { BaseRetriever, RetrieveParams } from "../../Retriever.js";
 import type { ServiceContext } from "../../ServiceContext.js";
 import { serviceContextFromDefaults } from "../../ServiceContext.js";
-import type { Event } from "../../callbacks/CallbackManager.js";
 import { RetrieverQueryEngine } from "../../engines/query/index.js";
 import type { BaseNodePostprocessor } from "../../postprocessors/index.js";
 import type {
@@ -281,7 +280,10 @@ export class SummaryIndexRetriever implements BaseRetriever {
    this.index = index;
  }

-  async retrieve(query: string, parentEvent?: Event): Promise<NodeWithScore[]> {
+  async retrieve({
+    query,
+    parentEvent,
+  }: RetrieveParams): Promise<NodeWithScore[]> {
    const nodeIds = this.index.indexStruct.nodes;
    const nodes = await this.index.docStore.getNodes(nodeIds);
    const result = nodes.map((node) => ({
@@ -337,7 +339,10 @@ export class SummaryIndexLLMRetriever implements BaseRetriever {
    this.serviceContext = serviceContext || index.serviceContext;
  }

-  async retrieve(query: string, parentEvent?: Event): Promise<NodeWithScore[]> {
+  async retrieve({
+    query,
+    parentEvent,
+  }: RetrieveParams): Promise<NodeWithScore[]> {
    const nodeIds = this.index.indexStruct.nodes;
    const results: NodeWithScore[] = [];

@@ -11,7 +11,7 @@ import {
  ObjectType,
  splitNodesByType,
 } from "../../Node.js";
-import type { BaseRetriever } from "../../Retriever.js";
+import type { BaseRetriever, RetrieveParams } from "../../Retriever.js";
 import type { ServiceContext } from "../../ServiceContext.js";
 import { serviceContextFromDefaults } from "../../ServiceContext.js";
 import type { Event } from "../../callbacks/CallbackManager.js";
@@ -426,14 +426,17 @@ export class VectorIndexRetriever implements BaseRetriever {
    this.imageSimilarityTopK = imageSimilarityTopK ?? DEFAULT_SIMILARITY_TOP_K;
  }

-  async retrieve(
-    query: string,
-    parentEvent?: Event,
-    preFilters?: MetadataFilters,
-  ): Promise<NodeWithScore[]> {
-    let nodesWithScores = await this.textRetrieve(query, preFilters);
+  async retrieve({
+    query,
+    parentEvent,
+    preFilters,
+  }: RetrieveParams): Promise<NodeWithScore[]> {
+    let nodesWithScores = await this.textRetrieve(
+      query,
+      preFilters as MetadataFilters,
+    );
    nodesWithScores = nodesWithScores.concat(
-      await this.textToImageRetrieve(query, preFilters),
+      await this.textToImageRetrieve(query, preFilters as MetadataFilters),
    );
    this.sendEvent(query, nodesWithScores, parentEvent);
    return nodesWithScores;
@@ -69,7 +69,7 @@ export class ObjectRetriever {

  // Translating the retrieve method
  async retrieve(strOrQueryBundle: QueryType): Promise<any> {
-    const nodes = await this.retriever.retrieve(strOrQueryBundle);
+    const nodes = await this.retriever.retrieve({ query: strOrQueryBundle });
    const objs = nodes.map((n) => this._objectNodeMapping.fromNode(n.node));
    return objs;
  }
@@ -0,0 +1,71 @@
+import {
+  IndexDict,
+  IndexList,
+  IndexStruct,
+  IndexStructType,
+  MetadataMode,
+  TextNode,
+  jsonToIndexStruct,
+} from "llamaindex";
+import { describe, expect, it } from "vitest";
+
+describe("jsonToIndexStruct", () => {
+  it("transforms json to IndexDict", () => {
+    function isIndexDict(some: IndexStruct): some is IndexDict {
+      return "type" in some && some.type === IndexStructType.SIMPLE_DICT;
+    }
+
+    const node = new TextNode({ text: "text", id_: "nodeId" });
+    const expected = new IndexDict();
+    expected.addNode(node);
+
+    console.log("expected.toJson()", expected.toJson());
+    const actual = jsonToIndexStruct(expected.toJson());
+
+    expect(isIndexDict(actual)).toBe(true);
+    expect(
+      (actual as IndexDict).nodesDict.nodeId.getContent(MetadataMode.NONE),
+    ).toEqual("text");
+  });
+  it("transforms json to IndexList", () => {
+    function isIndexList(some: IndexStruct): some is IndexList {
+      return "type" in some && some.type === IndexStructType.LIST;
+    }
+
+    const node = new TextNode({ text: "text", id_: "nodeId" });
+    const expected = new IndexList();
+    expected.addNode(node);
+
+    const actual = jsonToIndexStruct(expected.toJson());
+
+    expect(isIndexList(actual)).toBe(true);
+    expect((actual as IndexList).nodes[0]).toEqual("nodeId");
+  });
+  it("fails for unknown index type", () => {
+    expect(() => {
+      const json = {
+        indexId: "dd120b16-8dce-4ce3-9bb6-15ca87fe4a1d",
+        summary: undefined,
+        nodesDict: {},
+        type: "FOO",
+      };
+      return jsonToIndexStruct(json);
+    }).toThrowError("Unknown index struct type: FOO");
+  });
+  it("fails for unknown node type", () => {
+    expect(() => {
+      const json = {
+        indexId: "dd120b16-8dce-4ce3-9bb6-15ca87fe4a1d",
+        summary: undefined,
+        nodesDict: {
+          nodeId: {
+            ...new TextNode({ text: "text", id_: "nodeId" }).toJSON(),
+            type: "BAR",
+          },
+        },
+        type: IndexStructType.SIMPLE_DICT,
+      };
+      return jsonToIndexStruct(json);
+    }).toThrowError("Invalid node type: BAR");
+  });
+});
@@ -0,0 +1,20 @@
+{
+  "root": false,
+  "rules": {
+    "turbo/no-undeclared-env-vars": [
+      "error",
+      {
+        "allowList": [
+          "OPENAI_API_KEY",
+          "LLAMA_CLOUD_API_KEY",
+          "npm_config_user_agent",
+          "http_proxy",
+          "https_proxy",
+          "MODEL",
+          "NEXT_PUBLIC_CHAT_API",
+          "NEXT_PUBLIC_MODEL"
+        ]
+      }
+    ]
+  }
+}
@@ -1,5 +1,20 @@
 # create-llama

+## 0.0.28
+
+### Patch Changes
+
+- 89a49f4: Add more config variables to .env file
+- fdf48dd: Add "Start in VSCode" option to postInstallAction
+- fdf48dd: Add devcontainers to generated code
+
+## 0.0.27
+
+### Patch Changes
+
+- 2d29350: Add LlamaParse option when selecting a pdf file or a folder (FastAPI only)
+- b354f23: Add embedding model option to create-llama (FastAPI only)
+
 ## 0.0.26

 ### Patch Changes
@@ -11,6 +11,7 @@ import fs from "fs";
 import terminalLink from "terminal-link";
 import type { InstallTemplateArgs } from "./helpers";
 import { installTemplate } from "./helpers";
+import { writeDevcontainer } from "./helpers/devcontainer";
 import { templatesDir } from "./helpers/dir";
 import { toolsRequireConfig } from "./helpers/tools";

@@ -34,6 +35,7 @@ export async function createApp({
  openAiKey,
  llamaCloudKey,
  model,
+  embeddingModel,
  communityProjectPath,
  llamapack,
  vectorDb,
@@ -80,6 +82,7 @@ export async function createApp({
    openAiKey,
    llamaCloudKey,
    model,
+    embeddingModel,
    communityProjectPath,
    llamapack,
    vectorDb,
@@ -110,7 +113,7 @@ export async function createApp({
      path.join(root, "README.md"),
    );
  } else {
-    await installTemplate({ ...args, backend: true, forBackend: framework });
+    await installTemplate({ ...args, backend: true });
  }

  process.chdir(root);
@@ -119,6 +122,8 @@ export async function createApp({
    console.log();
  }

+  await writeDevcontainer(root, templatesDir, framework, frontend);
+
  if (toolsRequireConfig(tools)) {
    console.log(
      yellow(
@@ -91,17 +91,19 @@ for (const templateType of templateTypes) {
              test.skip(appType === "--no-frontend");
              await page.goto(`http://localhost:${port}`);
              await page.fill("form input", "hello");
-              await page.click("form button[type=submit]");
-              const response = await page.waitForResponse(
-                (res) => {
-                  return (
-                    res.url().includes("/api/chat") && res.status() === 200
-                  );
-                },
-                {
-                  timeout: 1000 * 60,
-                },
-              );
+              const [response] = await Promise.all([
+                page.waitForResponse(
+                  (res) => {
+                    return (
+                      res.url().includes("/api/chat") && res.status() === 200
+                    );
+                  },
+                  {
+                    timeout: 1000 * 60,
+                  },
+                ),
+                page.click("form button[type=submit]"),
+              ]);
              const text = await response.text();
              console.log("AI response when submitting message: ", text);
              expect(response.ok()).toBeTruthy();
@@ -14,6 +14,7 @@ import {

 export type AppType = "--frontend" | "--no-frontend" | "";
 const MODEL = "gpt-3.5-turbo";
+const EMBEDDING_MODEL = "text-embedding-ada-002";
 export type CreateLlamaResult = {
  projectName: string;
  appProcess: ChildProcess;
@@ -106,6 +107,8 @@ export async function runCreateLlama(
    vectorDb,
    "--model",
    MODEL,
+    "--embedding-model",
+    EMBEDDING_MODEL,
    "--open-ai-key",
    process.env.OPENAI_API_KEY || "testKey",
    appType,
@@ -119,6 +122,7 @@ export async function runCreateLlama(
    postInstallAction,
    "--tools",
    "none",
+    "--no-llama-parse",
  ].join(" ");
  console.log(`running command '${command}' in ${cwd}`);
  const appProcess = exec(command, {
@@ -171,7 +175,7 @@ export async function runCreateLlama(
 }

 export async function createTestDir() {
-  const cwd = path.join(__dirname, ".cache", crypto.randomUUID());
+  const cwd = path.join(__dirname, "cache", crypto.randomUUID());
  await mkdir(cwd, { recursive: true });
  return cwd;
 }
@@ -0,0 +1,61 @@
+import fs from "fs";
+import path from "path";
+import { TemplateFramework } from "./types";
+
+function renderDevcontainerContent(
+  templatesDir: string,
+  framework: TemplateFramework,
+  frontend: boolean,
+) {
+  const devcontainerJson: any = JSON.parse(
+    fs.readFileSync(path.join(templatesDir, "devcontainer.json"), "utf8"),
+  );
+
+  // Modify postCreateCommand
+  if (frontend) {
+    devcontainerJson.postCreateCommand =
+      framework === "fastapi"
+        ? "cd backend && poetry install && cd ../frontend && npm install"
+        : "cd backend && npm install && cd ../frontend && npm install";
+  } else {
+    devcontainerJson.postCreateCommand =
+      framework === "fastapi" ? "poetry install" : "npm install";
+  }
+
+  // Modify containerEnv
+  if (framework === "fastapi") {
+    if (frontend) {
+      devcontainerJson.containerEnv = {
+        ...devcontainerJson.containerEnv,
+        PYTHONPATH: "${PYTHONPATH}:${workspaceFolder}/backend",
+      };
+    } else {
+      devcontainerJson.containerEnv = {
+        ...devcontainerJson.containerEnv,
+        PYTHONPATH: "${PYTHONPATH}:${workspaceFolder}",
+      };
+    }
+  }
+
+  return JSON.stringify(devcontainerJson, null, 2);
+}
+
+export const writeDevcontainer = async (
+  root: string,
+  templatesDir: string,
+  framework: TemplateFramework,
+  frontend: boolean,
+) => {
+  console.log("Adding .devcontainer");
+  const devcontainerContent = renderDevcontainerContent(
+    templatesDir,
+    framework,
+    frontend,
+  );
+  const devcontainerDir = path.join(root, ".devcontainer");
+  fs.mkdirSync(devcontainerDir);
+  await fs.promises.writeFile(
+    path.join(devcontainerDir, "devcontainer.json"),
+    devcontainerContent,
+  );
+};
@@ -0,0 +1,241 @@
+import fs from "fs/promises";
+import path from "path";
+import {
+  FileSourceConfig,
+  TemplateDataSource,
+  TemplateFramework,
+  TemplateVectorDB,
+} from "./types";
+
+type EnvVar = {
+  name?: string;
+  description?: string;
+  value?: string;
+};
+
+const renderEnvVar = (envVars: EnvVar[]): string => {
+  return envVars.reduce(
+    (prev, env) =>
+      prev +
+      (env.description
+        ? `# ${env.description.replaceAll("\n", "\n# ")}\n`
+        : "") +
+      (env.name
+        ? env.value
+          ? `${env.name}=${env.value}\n\n`
+          : `# ${env.name}=\n\n`
+        : ""),
+    "",
+  );
+};
+
+const getVectorDBEnvs = (vectorDb: TemplateVectorDB) => {
+  switch (vectorDb) {
+    case "mongo":
+      return [
+        {
+          name: "MONGO_URI",
+          description:
+            "For generating a connection URI, see https://docs.timescale.com/use-timescale/latest/services/create-a-service\nThe MongoDB connection URI.",
+        },
+        {
+          name: "MONGODB_DATABASE",
+        },
+        {
+          name: "MONGODB_VECTORS",
+        },
+        {
+          name: "MONGODB_VECTOR_INDEX",
+        },
+      ];
+    case "pg":
+      return [
+        {
+          name: "PG_CONNECTION_STRING",
+          description:
+            "For generating a connection URI, see https://docs.timescale.com/use-timescale/latest/services/create-a-service\nThe PostgreSQL connection string.",
+        },
+      ];
+
+    case "pinecone":
+      return [
+        {
+          name: "PINECONE_API_KEY",
+          description:
+            "Configuration for Pinecone vector store\nThe Pinecone API key.",
+        },
+        {
+          name: "PINECONE_ENVIRONMENT",
+        },
+        {
+          name: "PINECONE_INDEX_NAME",
+        },
+      ];
+    default:
+      return [];
+  }
+};
+
+const getDataSourceEnvs = (dataSource: TemplateDataSource) => {
+  switch (dataSource.type) {
+    case "web":
+      return [
+        {
+          name: "BASE_URL",
+          description: "The base URL to start web scraping.",
+        },
+        {
+          name: "URL_PREFIX",
+          description: "The prefix of the URL to start web scraping.",
+        },
+        {
+          name: "MAX_DEPTH",
+          description: "The maximum depth to scrape.",
+        },
+      ];
+    default:
+      return [];
+  }
+};
+
+export const createBackendEnvFile = async (
+  root: string,
+  opts: {
+    openAiKey?: string;
+    llamaCloudKey?: string;
+    vectorDb?: TemplateVectorDB;
+    model?: string;
+    embeddingModel?: string;
+    framework?: TemplateFramework;
+    dataSource?: TemplateDataSource;
+    port?: number;
+  },
+) => {
+  // Init env values
+  const envFileName = ".env";
+  const defaultEnvs = [
+    {
+      render: true,
+      name: "MODEL",
+      description: "The name of LLM model to use.",
+      value: opts.model || "gpt-3.5-turbo",
+    },
+    {
+      render: true,
+      name: "OPENAI_API_KEY",
+      description: "The OpenAI API key to use.",
+      value: opts.openAiKey,
+    },
+    // Add vector database environment variables
+    ...(opts.vectorDb ? getVectorDBEnvs(opts.vectorDb) : []),
+    // Add data source environment variables
+    ...(opts.dataSource ? getDataSourceEnvs(opts.dataSource) : []),
+  ];
+  let envVars: EnvVar[] = [];
+  if (opts.framework === "fastapi") {
+    envVars = [
+      ...defaultEnvs,
+      ...[
+        {
+          name: "APP_HOST",
+          description: "The address to start the backend app.",
+          value: "0.0.0.0",
+        },
+        {
+          name: "APP_PORT",
+          description: "The port to start the backend app.",
+          value: opts.port?.toString() || "8000",
+        },
+        {
+          name: "EMBEDDING_MODEL",
+          description: "Name of the embedding model to use.",
+          value: opts.embeddingModel,
+        },
+        {
+          name: "EMBEDDING_DIM",
+          description: "Dimension of the embedding model to use.",
+        },
+        {
+          name: "LLM_TEMPERATURE",
+          description: "Temperature for sampling from the model.",
+        },
+        {
+          name: "LLM_MAX_TOKENS",
+          description: "Maximum number of tokens to generate.",
+        },
+        {
+          name: "TOP_K",
+          description:
+            "The number of similar embeddings to return when retrieving documents.",
+          value: "3",
+        },
+        {
+          name: "SYSTEM_PROMPT",
+          description: `Custom system prompt.
+Example:
+SYSTEM_PROMPT="
+We have provided context information below.
+---------------------
+{context_str}
+---------------------
+Given this information, please answer the question: {query_str}
+"`,
+        },
+        (opts?.dataSource?.config as FileSourceConfig).useLlamaParse
+          ? {
+              name: "LLAMA_CLOUD_API_KEY",
+              description: `The Llama Cloud API key.`,
+              value: opts.llamaCloudKey,
+            }
+          : {},
+      ],
+    ];
+  } else {
+    envVars = [
+      ...defaultEnvs,
+      ...[
+        opts.framework === "nextjs"
+          ? {
+              name: "NEXT_PUBLIC_MODEL",
+              description:
+                "The LLM model to use (hardcode to front-end artifact).",
+            }
+          : {},
+      ],
+    ];
+  }
+  // Render and write env file
+  const content = renderEnvVar(envVars);
+  await fs.writeFile(path.join(root, envFileName), content);
+  console.log(`Created '${envFileName}' file. Please check the settings.`);
+};
+
+export const createFrontendEnvFile = async (
+  root: string,
+  opts: {
+    customApiPath?: string;
+    model?: string;
+  },
+) => {
+  const defaultFrontendEnvs = [
+    {
+      name: "MODEL",
+      description: "The OpenAI model to use.",
+      value: opts.model,
+    },
+    {
+      name: "NEXT_PUBLIC_MODEL",
+      description: "The OpenAI model to use (hardcode to front-end artifact).",
+      value: opts.model,
+    },
+    {
+      name: "NEXT_PUBLIC_CHAT_API",
+      description: "The backend API for chat endpoint.",
+      value: opts.customApiPath
+        ? opts.customApiPath
+        : "http://localhost:8000/api/chat",
+    },
+  ];
+  const content = renderEnvVar(defaultFrontendEnvs);
+  await fs.writeFile(path.join(root, ".env"), content);
+};
@@ -7,6 +7,7 @@ import { cyan } from "picocolors";

 import { COMMUNITY_OWNER, COMMUNITY_REPO } from "./constant";
 import { templatesDir } from "./dir";
+import { createBackendEnvFile, createFrontendEnvFile } from "./env-variables";
 import { PackageManager } from "./get-pkg-manager";
 import { installLlamapackProject } from "./llama-pack";
 import { isHavingPoetryLockFile, tryPoetryRun } from "./poetry";
@@ -18,94 +19,37 @@ import {
  TemplateDataSource,
  TemplateFramework,
  TemplateVectorDB,
-  WebSourceConfig,
 } from "./types";
 import { installTSTemplate } from "./typescript";

-const createEnvLocalFile = async (
-  root: string,
-  opts?: {
-    openAiKey?: string;
-    llamaCloudKey?: string;
-    vectorDb?: TemplateVectorDB;
-    model?: string;
-    framework?: TemplateFramework;
-    dataSource?: TemplateDataSource;
-  },
-) => {
-  const envFileName = ".env";
-  let content = "";
-
-  const model = opts?.model || "gpt-3.5-turbo";
-  content += `MODEL=${model}\n`;
-  if (opts?.framework === "nextjs") {
-    content += `NEXT_PUBLIC_MODEL=${model}\n`;
-  }
-  console.log("\nUsing OpenAI model: ", model, "\n");
-
-  if (opts?.openAiKey) {
-    content += `OPENAI_API_KEY=${opts?.openAiKey}\n`;
-  }
-
-  if (opts?.llamaCloudKey) {
-    content += `LLAMA_CLOUD_API_KEY=${opts?.llamaCloudKey}\n`;
-  }
-
-  switch (opts?.vectorDb) {
-    case "mongo": {
-      content += `# For generating a connection URI, see https://www.mongodb.com/docs/guides/atlas/connection-string\n`;
-      content += `MONGO_URI=\n`;
-      content += `MONGODB_DATABASE=\n`;
-      content += `MONGODB_VECTORS=\n`;
-      content += `MONGODB_VECTOR_INDEX=\n`;
-      break;
-    }
-    case "pg": {
-      content += `# For generating a connection URI, see https://docs.timescale.com/use-timescale/latest/services/create-a-service\n`;
-      content += `PG_CONNECTION_STRING=\n`;
-      break;
-    }
-    case "pinecone": {
-      content += `PINECONE_API_KEY=\n`;
-      content += `PINECONE_ENVIRONMENT=\n`;
-      content += `PINECONE_INDEX_NAME=\n`;
-      break;
-    }
-  }
-
-  switch (opts?.dataSource?.type) {
-    case "web": {
-      const webConfig = opts?.dataSource.config as WebSourceConfig;
-      content += `# web loader config\n`;
-      content += `BASE_URL=${webConfig.baseUrl}\n`;
-      content += `URL_PREFIX=${webConfig.baseUrl}\n`;
-      content += `MAX_DEPTH=${webConfig.depth}\n`;
-      break;
-    }
-  }
-
-  if (content) {
-    await fs.writeFile(path.join(root, envFileName), content);
-    console.log(`Created '${envFileName}' file. Please check the settings.`);
-  }
-};
-
-const generateContextData = async (
+// eslint-disable-next-line max-params
+async function generateContextData(
  framework: TemplateFramework,
  packageManager?: PackageManager,
  openAiKey?: string,
  vectorDb?: TemplateVectorDB,
-) => {
+  dataSource?: TemplateDataSource,
+  llamaCloudKey?: string,
+) {
  if (packageManager) {
    const runGenerate = `${cyan(
      framework === "fastapi"
        ? "poetry run python app/engine/generate.py"
        : `${packageManager} run generate`,
    )}`;
-    const hasOpenAiKey = openAiKey || process.env["OPENAI_API_KEY"];
+    const openAiKeyConfigured = openAiKey || process.env["OPENAI_API_KEY"];
+    const llamaCloudKeyConfigured = (dataSource?.config as FileSourceConfig)
+      ?.useLlamaParse
+      ? llamaCloudKey || process.env["LLAMA_CLOUD_API_KEY"]
+      : true;
    const hasVectorDb = vectorDb && vectorDb !== "none";
    if (framework === "fastapi") {
-      if (hasOpenAiKey && !hasVectorDb && isHavingPoetryLockFile()) {
+      if (
+        openAiKeyConfigured &&
+        llamaCloudKeyConfigured &&
+        !hasVectorDb &&
+        isHavingPoetryLockFile()
+      ) {
        console.log(`Running ${runGenerate} to generate the context data.`);
        const result = tryPoetryRun("python app/engine/generate.py");
        if (!result) {
@@ -116,7 +60,7 @@ const generateContextData = async (
        return;
      }
    } else {
-      if (hasOpenAiKey && vectorDb === "none") {
+      if (openAiKeyConfigured && vectorDb === "none") {
        console.log(`Running ${runGenerate} to generate the context data.`);
        await callPackageManager(packageManager, true, ["run", "generate"]);
        return;
@@ -124,14 +68,15 @@ const generateContextData = async (
    }

    const settings = [];
-    if (!hasOpenAiKey) settings.push("your OpenAI key");
+    if (!openAiKeyConfigured) settings.push("your OpenAI key");
+    if (!llamaCloudKeyConfigured) settings.push("your Llama Cloud key");
    if (hasVectorDb) settings.push("your Vector DB environment variables");
    const settingsMessage =
      settings.length > 0 ? `After setting ${settings.join(" and ")}, ` : "";
    const generateMessage = `run ${runGenerate} to generate the context data.`;
    console.log(`\n${settingsMessage}${generateMessage}\n\n`);
  }
-};
+}

 const copyContextData = async (
  root: string,
@@ -208,13 +153,15 @@ export const installTemplate = async (
    // This is a backend, so we need to copy the test data and create the env file.

    // Copy the environment file to the target directory.
-    await createEnvLocalFile(props.root, {
+    await createBackendEnvFile(props.root, {
      openAiKey: props.openAiKey,
      llamaCloudKey: props.llamaCloudKey,
      vectorDb: props.vectorDb,
      model: props.model,
+      embeddingModel: props.embeddingModel,
      framework: props.framework,
      dataSource: props.dataSource,
+      port: props.externalPort,
    });

    if (props.engine === "context") {
@@ -228,13 +175,17 @@ export const installTemplate = async (
          props.packageManager,
          props.openAiKey,
          props.vectorDb,
+          props.dataSource,
+          props.llamaCloudKey,
        );
      }
    }
  } else {
    // this is a frontend for a full-stack app, create .env file with model information
-    const content = `MODEL=${props.model}\nNEXT_PUBLIC_MODEL=${props.model}\n`;
-    await fs.writeFile(path.join(props.root, ".env"), content);
+    createFrontendEnvFile(props.root, {
+      model: props.model,
+      customApiPath: props.customApiPath,
+    });
  }
 };

@@ -142,7 +142,7 @@ export const installLlamapackProject = async ({
  await copyLlamapackEmptyProject({ root });
  await copyData({ root });
  await installLlamapackExample({ root, llamapack });
-  if (postInstallAction !== "none") {
+  if (postInstallAction === "runApp" || postInstallAction === "dependencies") {
    installPythonDependencies({ noRoot: true });
  }
 };
@@ -266,7 +266,7 @@ export const installPythonTemplate = async ({
  );
  await addDependencies(root, addOnDependencies);

-  if (postInstallAction !== "none") {
+  if (postInstallAction === "runApp" || postInstallAction === "dependencies") {
    installPythonDependencies();
  }
 };
@@ -6,7 +6,11 @@ export type TemplateFramework = "nextjs" | "express" | "fastapi";
 export type TemplateEngine = "simple" | "context";
 export type TemplateUI = "html" | "shadcn";
 export type TemplateVectorDB = "none" | "mongo" | "pg" | "pinecone";
-export type TemplatePostInstallAction = "none" | "dependencies" | "runApp";
+export type TemplatePostInstallAction =
+  | "none"
+  | "VSCode"
+  | "dependencies"
+  | "runApp";
 export type TemplateDataSource = {
  type: TemplateDataSourceType;
  config: TemplateDataSourceConfig;
@@ -37,8 +41,8 @@ export interface InstallTemplateArgs {
  customApiPath?: string;
  openAiKey?: string;
  llamaCloudKey?: string;
-  forBackend?: string;
  model: string;
+  embeddingModel: string;
  communityProjectPath?: string;
  llamapack?: string;
  vectorDb?: TemplateVectorDB;
@@ -61,10 +61,10 @@ export const installTSTemplate = async ({
  ui,
  eslint,
  customApiPath,
-  forBackend,
  vectorDb,
  postInstallAction,
-}: InstallTemplateArgs) => {
+  backend,
+}: InstallTemplateArgs & { backend: boolean }) => {
  console.log(bold(`Using ${packageManager}.`));

  /**
@@ -82,23 +82,20 @@ export const installTSTemplate = async ({
  });

  /**
-   * If the backend is next.js, rename next.config.app.js to next.config.js
-   * If not, rename next.config.static.js to next.config.js
+   * If next.js is not used as a backend, update next.config.js to use static site generation.
   */
-  if (framework == "nextjs" && forBackend === "nextjs") {
-    const nextConfigAppPath = path.join(root, "next.config.app.js");
-    const nextConfigPath = path.join(root, "next.config.js");
-    await fs.rename(nextConfigAppPath, nextConfigPath);
-    // delete next.config.static.js
-    const nextConfigStaticPath = path.join(root, "next.config.static.js");
-    await fs.rm(nextConfigStaticPath);
-  } else if (framework == "nextjs" && typeof forBackend === "undefined") {
-    const nextConfigStaticPath = path.join(root, "next.config.static.js");
-    const nextConfigPath = path.join(root, "next.config.js");
-    await fs.rename(nextConfigStaticPath, nextConfigPath);
-    // delete next.config.app.js
-    const nextConfigAppPath = path.join(root, "next.config.app.js");
-    await fs.rm(nextConfigAppPath);
+  if (framework === "nextjs" && !backend) {
+    // update next.config.json for static site generation
+    const nextConfigJsonFile = path.join(root, "next.config.json");
+    const nextConfigJson: any = JSON.parse(
+      await fs.readFile(nextConfigJsonFile, "utf8"),
+    );
+    nextConfigJson.output = "export";
+    nextConfigJson.images = { unoptimized: true };
+    await fs.writeFile(
+      nextConfigJsonFile,
+      JSON.stringify(nextConfigJson, null, 2) + os.EOL,
+    );
  }

  /**
@@ -174,10 +171,6 @@ export const installTSTemplate = async ({
    const apiPath = path.join(root, "app", "api");
    await fs.rm(apiPath, { recursive: true });
    // modify the dev script to use the custom api path
-    packageJson.scripts = {
-      ...packageJson.scripts,
-      dev: `cross-env NEXT_PUBLIC_CHAT_API=${customApiPath} next dev`,
-    };
  }

  if (engine === "context" && relativeEngineDestPath) {
@@ -228,7 +221,7 @@ export const installTSTemplate = async ({
    JSON.stringify(packageJson, null, 2) + os.EOL,
  );

-  if (postInstallAction !== "none") {
+  if (postInstallAction === "runApp" || postInstallAction === "dependencies") {
    await installTSDependencies(packageJson, packageManager, isOnline);
  }
 };
@@ -1,11 +1,13 @@
 #!/usr/bin/env node
 /* eslint-disable import/no-extraneous-dependencies */
+import { execSync } from "child_process";
 import Commander from "commander";
 import Conf from "conf";
 import fs from "fs";
 import path from "path";
 import { bold, cyan, green, red, yellow } from "picocolors";
 import prompts from "prompts";
+import terminalLink from "terminal-link";
 import checkForUpdate from "update-check";
 import { createApp } from "./create-app";
 import { getPkgManager } from "./helpers/get-pkg-manager";
@@ -119,6 +121,12 @@ const program = new Commander.Command(packageJson.name)
    `

  Select OpenAI model to use. E.g. gpt-3.5-turbo.
+`,
+  )
+  .option(
+    "--embedding-model <embeddingModel>",
+    `
+  Select OpenAI embedding model to use. E.g. text-embedding-ada-002.
 `,
  )
  .option(
@@ -281,6 +289,7 @@ async function run(): Promise<void> {
    openAiKey: program.openAiKey,
    llamaCloudKey: program.llamaCloudKey,
    model: program.model,
+    embeddingModel: program.embeddingModel,
    communityProjectPath: program.communityProjectPath,
    llamapack: program.llamapack,
    vectorDb: program.vectorDb,
@@ -291,7 +300,31 @@ async function run(): Promise<void> {
  });
  conf.set("preferences", preferences);

-  if (program.postInstallAction === "runApp") {
+  if (program.postInstallAction === "VSCode") {
+    console.log(`Starting VSCode in ${root}...`);
+    try {
+      execSync(`code . --new-window --goto README.md`, {
+        stdio: "inherit",
+        cwd: root,
+      });
+    } catch (error) {
+      console.log(
+        red(
+          `Failed to start VSCode in ${root}. 
+Got error: ${(error as Error).message}.\n`,
+        ),
+      );
+      console.log(
+        `Make sure you have VSCode installed and added to your PATH. 
+Please check ${cyan(
+          terminalLink(
+            "This documentation",
+            `https://code.visualstudio.com/docs/setup/setup-overview`,
+          ),
+        )} for more information.`,
+      );
+    }
+  } else if (program.postInstallAction === "runApp") {
    console.log(`Running app in ${root}...`);
    await runApp(
      root,
@@ -1,6 +1,6 @@
 {
  "name": "create-llama",
-  "version": "0.0.26",
+  "version": "0.0.28",
  "keywords": [
    "rag",
    "llamaindex",
@@ -23,7 +23,7 @@
    "clean": "rimraf --glob ./dist ./templates/**/__pycache__ ./templates/**/node_modules ./templates/**/poetry.lock",
    "dev": "ncc build ./index.ts -w -o dist/",
    "build": "npm run clean && ncc build ./index.ts -o ./dist/ --minify --no-cache --no-source-map-register",
-    "lint": "eslint . --ignore-pattern dist",
+    "lint": "eslint . --ignore-pattern dist --ignore-pattern e2e/cache",
    "e2e": "playwright test",
    "prepublishOnly": "cd ../../ && pnpm run build:release"
  },
@@ -69,6 +69,7 @@ const defaults: QuestionArgs = {
  openAiKey: "",
  llamaCloudKey: "",
  model: "gpt-3.5-turbo",
+  embeddingModel: "text-embedding-ada-002",
  communityProjectPath: "",
  llamapack: "",
  postInstallAction: "dependencies",
@@ -214,18 +215,30 @@ export const askQuestions = async (
            title: "Just generate code (~1 sec)",
            value: "none",
          },
+          {
+            title: "Start in VSCode (~1 sec)",
+            value: "VSCode",
+          },
          {
            title: "Generate code and install dependencies (~2 min)",
            value: "dependencies",
          },
        ];

-        const hasOpenAiKey = program.openAiKey || process.env["OPENAI_API_KEY"];
+        const openAiKeyConfigured =
+          program.openAiKey || process.env["OPENAI_API_KEY"];
+        // If using LlamaParse, require LlamaCloud API key
+        const llamaCloudKeyConfigured = (
+          program.dataSource?.config as FileSourceConfig
+        )?.useLlamaParse
+          ? program.llamaCloudKey || process.env["LLAMA_CLOUD_API_KEY"]
+          : true;
        const hasVectorDb = program.vectorDb && program.vectorDb !== "none";
        // Can run the app if all tools do not require configuration
        if (
          !hasVectorDb &&
-          hasOpenAiKey &&
+          openAiKeyConfigured &&
+          llamaCloudKeyConfigured &&
          !toolsRequireConfig(program.tools) &&
          !program.llamapack
        ) {
@@ -443,6 +456,38 @@ export const askQuestions = async (
    }
  }

+  if (!program.embeddingModel && program.framework === "fastapi") {
+    if (ciInfo.isCI) {
+      program.embeddingModel = getPrefOrDefault("embeddingModel");
+    } else {
+      const { embeddingModel } = await prompts(
+        {
+          type: "select",
+          name: "embeddingModel",
+          message: "Which embedding model would you like to use?",
+          choices: [
+            {
+              title: "text-embedding-ada-002",
+              value: "text-embedding-ada-002",
+            },
+            {
+              title: "text-embedding-3-small",
+              value: "text-embedding-3-small",
+            },
+            {
+              title: "text-embedding-3-large",
+              value: "text-embedding-3-large",
+            },
+          ],
+          initial: 0,
+        },
+        handlers,
+      );
+      program.embeddingModel = embeddingModel;
+      preferences.embeddingModel = embeddingModel;
+    }
+  }
+
  if (program.files) {
    // If user specified files option, then the program should use context engine
    program.engine == "context";
@@ -527,8 +572,9 @@ export const askQuestions = async (
  }

  if (
-    program.dataSource?.type === "file" ||
-    (program.dataSource?.type === "folder" && program.framework === "fastapi")
+    (program.dataSource?.type === "file" ||
+      program.dataSource?.type === "folder") &&
+    program.framework === "fastapi"
  ) {
    if (ciInfo.isCI) {
      program.llamaCloudKey = getPrefOrDefault("llamaCloudKey");
@@ -571,11 +617,8 @@ export const askQuestions = async (
          {
            type: "text",
            name: "llamaCloudKey",
-            message: "Please provide your LlamaIndex Cloud API key:",
-            validate: (value) =>
-              value
-                ? true
-                : "LlamaIndex Cloud API key is required. You can get it from: https://cloud.llamaindex.ai/api-key",
+            message:
+              "Please provide your LlamaIndex Cloud API key (leave blank to skip):",
          },
          handlers,
        );
@@ -1,3 +1,4 @@
+import os
 from llama_index.core.settings import Settings
 from llama_index.core.agent import AgentRunner
 from llama_index.core.tools.query_engine import QueryEngineTool
@@ -6,11 +7,13 @@ from app.engine.index import get_index


 def get_chat_engine():
+    system_prompt = os.getenv("SYSTEM_PROMPT")
+    top_k = os.getenv("TOP_K", "3")
    tools = []

    # Add query tool
    index = get_index()
-    query_engine = index.as_query_engine(similarity_top_k=3)
+    query_engine = index.as_query_engine(similarity_top_k=int(top_k))
    query_engine_tool = QueryEngineTool.from_defaults(query_engine=query_engine)
    tools.append(query_engine_tool)

@@ -20,5 +23,6 @@ def get_chat_engine():
    return AgentRunner.from_llm(
        llm=Settings.llm,
        tools=tools,
+        system_prompt=system_prompt,
        verbose=True,
    )
@@ -1,7 +1,13 @@
+import os
 from app.engine.index import get_index


 def get_chat_engine():
+    system_prompt = os.getenv("SYSTEM_PROMPT")
+    top_k = os.getenv("TOP_K", 3)
+
    return get_index().as_chat_engine(
-        similarity_top_k=3, chat_mode="condense_plus_context"
+        similarity_top_k=int(top_k),
+        system_prompt=system_prompt,
+        chat_mode="condense_plus_context",
    )
@@ -1,3 +1,4 @@
+import os
 from llama_parse import LlamaParse
 from llama_index.core import SimpleDirectoryReader

@@ -5,10 +6,12 @@ DATA_DIR = "data"  # directory containing the documents


 def get_documents():
-    parser = LlamaParse(
-        result_type="markdown",
-        verbose=True,
-    )
+    if os.getenv("LLAMA_CLOUD_API_KEY") is None:
+        raise ValueError(
+            "LLAMA_CLOUD_API_KEY environment variable is not set. "
+            "Please set it in .env file or in your shell environment then run again!"
+        )
+    parser = LlamaParse(result_type="markdown", verbose=True, language="en")

    reader = SimpleDirectoryReader(DATA_DIR, file_extractor={".pdf": parser})
    return reader.load_data()
@@ -0,0 +1,35 @@
+{
+  "image": "mcr.microsoft.com/vscode/devcontainers/typescript-node:dev-20-bullseye",
+  "features": {
+    "ghcr.io/devcontainers-contrib/features/turborepo-npm:1": {},
+    "ghcr.io/devcontainers-contrib/features/typescript:2": {},
+    "ghcr.io/devcontainers/features/python:1": {
+      "version": "3.11",
+      "toolsToInstall": ["flake8", "black", "mypy", "poetry"]
+    }
+  },
+  "customizations": {
+    "codespaces": {
+      "openFiles": ["README.md"]
+    },
+    "vscode": {
+      "extensions": [
+        "ms-vscode.typescript-language-features",
+        "esbenp.prettier-vscode",
+        "ms-python.python",
+        "ms-python.black-formatter",
+        "ms-python.vscode-flake8",
+        "ms-python.vscode-pylance"
+      ],
+      "settings": {
+        "python.formatting.provider": "black",
+        "python.languageServer": "Pylance",
+        "python.analysis.typeCheckingMode": "basic"
+      }
+    }
+  },
+  "containerEnv": {
+    "POETRY_VIRTUALENVS_CREATE": "false"
+  },
+  "forwardPorts": [3000, 8000]
+}
@@ -1,10 +1,41 @@
 import os
-from llama_index.llms.openai import OpenAI
+from typing import Dict
 from llama_index.core.settings import Settings
+from llama_index.llms.openai import OpenAI
+from llama_index.embeddings.openai import OpenAIEmbedding
+
+
+def llm_config_from_env() -> Dict:
+    from llama_index.core.constants import DEFAULT_TEMPERATURE
+
+    model = os.getenv("MODEL")
+    temperature = os.getenv("LLM_TEMPERATURE", DEFAULT_TEMPERATURE)
+    max_tokens = os.getenv("LLM_MAX_TOKENS")
+
+    config = {
+        "model": model,
+        "temperature": float(temperature),
+        "max_tokens": int(max_tokens) if max_tokens is not None else None,
+    }
+    return config
+
+
+def embedding_config_from_env() -> Dict:
+    model = os.getenv("EMBEDDING_MODEL")
+    dimension = os.getenv("EMBEDDING_DIM")
+
+    config = {
+        "model": model,
+        "dimension": int(dimension) if dimension is not None else None,
+    }
+    return config


 def init_settings():
-    model = os.getenv("MODEL", "gpt-3.5-turbo")
-    Settings.llm = OpenAI(model=model)
-    Settings.chunk_size = 1024
-    Settings.chunk_overlap = 20
+    llm_configs = llm_config_from_env()
+    embedding_configs = embedding_config_from_env()
+
+    Settings.llm = OpenAI(**llm_configs)
+    Settings.embed_model = OpenAIEmbedding(**embedding_configs)
+    Settings.chunk_size = int(os.getenv("CHUNK_SIZE", "1024"))
+    Settings.chunk_overlap = int(os.getenv("CHUNK_OVERLAP", "20"))
@@ -32,4 +32,7 @@ app.include_router(chat_router, prefix="/api/chat")


 if __name__ == "__main__":
-    uvicorn.run(app="main:app", host="0.0.0.0", reload=True)
+    app_host = os.getenv("APP_HOST", "0.0.0.0")
+    app_port = int(os.getenv("APP_PORT", "8000"))
+
+    uvicorn.run(app="main:app", host=app_host, port=app_port, reload=True)
@@ -1,10 +1,41 @@
 import os
-from llama_index.llms.openai import OpenAI
+from typing import Dict
 from llama_index.core.settings import Settings
+from llama_index.llms.openai import OpenAI
+from llama_index.embeddings.openai import OpenAIEmbedding
+
+
+def llm_config_from_env() -> Dict:
+    from llama_index.core.constants import DEFAULT_TEMPERATURE
+
+    model = os.getenv("MODEL")
+    temperature = os.getenv("LLM_TEMPERATURE", DEFAULT_TEMPERATURE)
+    max_tokens = os.getenv("LLM_MAX_TOKENS")
+
+    config = {
+        "model": model,
+        "temperature": float(temperature),
+        "max_tokens": int(max_tokens) if max_tokens is not None else None,
+    }
+    return config
+
+
+def embedding_config_from_env() -> Dict:
+    model = os.getenv("EMBEDDING_MODEL")
+    dimension = os.getenv("EMBEDDING_DIM")
+
+    config = {
+        "model": model,
+        "dimension": int(dimension) if dimension is not None else None,
+    }
+    return config


 def init_settings():
-    model = os.getenv("MODEL", "gpt-3.5-turbo")
-    Settings.llm = OpenAI(model=model)
-    Settings.chunk_size = 1024
-    Settings.chunk_overlap = 20
+    llm_configs = llm_config_from_env()
+    embedding_configs = embedding_config_from_env()
+
+    Settings.llm = OpenAI(**llm_configs)
+    Settings.embed_model = OpenAIEmbedding(**embedding_configs)
+    Settings.chunk_size = int(os.getenv("CHUNK_SIZE", "1024"))
+    Settings.chunk_overlap = int(os.getenv("CHUNK_OVERLAP", "20"))
@@ -33,4 +33,7 @@ app.include_router(chat_router, prefix="/api/chat")


 if __name__ == "__main__":
-    uvicorn.run(app="main:app", host="0.0.0.0", reload=True)
+    app_host = os.getenv("APP_HOST", "0.0.0.0")
+    app_port = int(os.getenv("APP_PORT", "8000"))
+
+    uvicorn.run(app="main:app", host=app_host, port=app_port, reload=True)
@@ -1,19 +0,0 @@
-/** @type {import('next').NextConfig} */
-const nextConfig = {
-  webpack: (config) => {
-    // See https://webpack.js.org/configuration/resolve/#resolvealias
-    config.resolve.alias = {
-      ...config.resolve.alias,
-      sharp$: false,
-      "onnxruntime-node$": false,
-    };
-    return config;
-  },
-  experimental: {
-    outputFileTracingIncludes: {
-      "/*": ["./cache/**/*"],
-    },
-  },
-};
-
-module.exports = nextConfig;
@@ -0,0 +1,7 @@
+{
+  "experimental": {
+    "outputFileTracingIncludes": {
+      "/*": ["./cache/**/*"]
+    }
+  }
+}
@@ -0,0 +1,8 @@
+/** @type {import('next').NextConfig} */
+import fs from "fs";
+import webpack from "./webpack.config.mjs";
+
+const nextConfig = JSON.parse(fs.readFileSync("./next.config.json", "utf-8"));
+nextConfig.webpack = webpack;
+
+export default nextConfig;
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Marcus Schiesser	4589a84643	RELEASING: Releasing 1 package(s) Releases: create-llama@0.0.28 [skip ci]	2024-03-12 13:41:36 +07:00
Huu Le (Lee)	e6b7f52d3e	fix: add missing check env logic (#636 )	2024-03-12 12:29:00 +07:00
Marcus Schiesser	b169db617a	refactor: use a function for webpack config (#634 )	2024-03-12 11:10:31 +07:00
Huu Le (Lee)	89a49f4f4f	feat: Add more. env variables to config host, port, llm and embedding (#630 )	2024-03-12 09:22:21 +07:00
Marcus Schiesser	58490715fe	refactor: clean nextjs config generation (use JSON) (#631 )	2024-03-11 14:16:15 +07:00
Huu Le (Lee)	4c2283c4e5	fix: Rename folder e2e/.cache to e2e/cache (#632 )	2024-03-11 14:15:13 +07:00
Eka Prasetia	a059070dec	docs: Fix typo in transformations.md (#625 )	2024-03-11 12:16:23 +07:00
Emanuel Ferreira	20dfeb4cfa	chore: remove comment (#624 )	2024-03-08 15:54:24 -03:00
Emanuel Ferreira	aefc3266c1	feat: experimental package + json query engine (#613 )	2024-03-07 14:34:55 -03:00
Huu Le (Lee)	fdf48dd459	feat: Add start in VSCode option and support python for dev container (#619 )	2024-03-07 17:19:08 +07:00
Alex Yang	66525346a2	build: use single swc config (#620 )	2024-03-06 23:41:42 -06:00
Alex Yang	c9b2ec4a2b	fix: release	2024-03-06 23:32:02 -06:00
Marcus Schiesser	bf583a7266	Use parameter object for retrieve function of Retriever (#616 )	2024-03-06 21:15:22 -08:00
Marcus Schiesser	de194d1c73	fix: running new-create-llama	2024-03-06 15:13:20 +07:00
Marcus Schiesser	ecdc289df1	RELEASING: Releasing 1 package(s) Releases: create-llama@0.0.27 [skip ci]	2024-03-06 15:11:47 +07:00
Marcus Schiesser	9e198ac40d	fix: build types for core locally (#615 )	2024-03-06 14:35:31 +07:00
Huu Le (Lee)	0a06998690	fix: hardcode "en" as default language for llama-parse and use llama cloud key from env (#614 )	2024-03-06 14:31:21 +07:00
Wojciech Grzebieniowski	484a7105a9	fix: restore missing exports (#610 )	2024-03-05 14:56:25 -06:00
Alex Yang	8d18ea167b	fix: publish.yml	2024-03-05 14:37:54 -06:00
Alex Yang	a2ca89bfe0	fix: config (#611 )	2024-03-05 14:20:58 -06:00
Alex Yang	edeea40898	ci: add publish.yml	2024-03-05 13:49:49 -06:00
Alex Yang	2a7080b094	build: fix version	2024-03-05 12:26:47 -06:00
Huu Le (Lee)	b354f2386b	feat: add embedding model option to create-llama (#608 )	2024-03-05 16:59:51 +07:00
Emanuel Ferreira	d766bd03d2	feat: OpenAI Agent Stream (#597 )	2024-03-05 15:46:44 +07:00
Huu Le (Lee)	6a69148356	fix: add --no-llama-parse and improve e2e test (#607 )	2024-03-05 14:57:38 +07:00