Release 0.1.18 (#157)

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

.github/workflows/e2e.yml

@@ -17,7 +17,7 @@ jobs:
       matrix:
         node-version: [18, 20]
         python-version: ["3.11"]
-        os: [macos-latest, windows-latest]
+        os: [macos-latest, windows-latest, ubuntu-22.04]
     defaults:
       run:
         shell: bash

CHANGELOG.md

@@ -1,5 +1,67 @@
 # create-llama
 
+## 0.1.18
+
+### Patch Changes
+
+- d1026ea: support Mistral as llm and embedding
+- a221cfc: Use LlamaParse for all the file types that it supports (if activated)
+
+## 0.1.17
+
+### Patch Changes
+
+- 9ecd061: Add new template for a multi-agents app
+
+## 0.1.16
+
+### Patch Changes
+
+- a0aab03: Add T-System's LLMHUB as a model provider
+
+## 0.1.15
+
+### Patch Changes
+
+- 64732f0: Fix the issue of images not showing with the sandbox URL from OpenAI's models
+- aeb6fef: use llamacloud for chat
+
+## 0.1.14
+
+### Patch Changes
+
+- f2c3389: chore: update to llamaindex 0.4.3
+- 5093b37: Remove non-working file selectors for Linux
+
+## 0.1.13
+
+### Patch Changes
+
+- b3c969d: Add image generator tool
+
+## 0.1.12
+
+### Patch Changes
+
+- aa69014: Fix NextJS for TS 5.2
+
+## 0.1.11
+
+### Patch Changes
+
+- 48b96ff: Add DuckDuckGo search tool
+- 9c9decb: Reuse function tool instances and improve e2b interpreter tool for Python
+- 02ed277: Add Groq as a model provider
+- 0748f2e: Remove hard-coded Gemini supported models
+
+## 0.1.10
+
+### Patch Changes
+
+- 9112d08: Add OpenAPI tool for Typescript
+- 8f03f8d: Add OLLAMA_REQUEST_TIMEOUT variable to config Ollama timeout (Python)
+- 8f03f8d: Apply nest_asyncio for llama parse
+
 ## 0.1.9
 
 ### Patch Changes

create-app.ts

@@ -151,5 +151,19 @@ export async function createApp({
     );
   }
 
+  if (
+    dataSources.some((dataSource) => dataSource.type === "file") &&
+    process.platform === "linux"
+  ) {
+    console.log(
+      yellow(
+        `You can add your own data files to ${terminalLink(
+          "data",
+          `file://${root}/data`,
+        )} folder manually.`,
+      ),
+    );
+  }
+
   console.log();
 }

helpers/env-variables.ts

@@ -5,9 +5,12 @@ import {
   ModelConfig,
   TemplateDataSource,
   TemplateFramework,
+  TemplateType,
   TemplateVectorDB,
 } from "./types";
 
+import { TSYSTEMS_LLMHUB_API_URL } from "./providers/llmhub";
+
 export type EnvVar = {
   name?: string;
   description?: string;
@@ -133,6 +136,26 @@ const getVectorDBEnvs = (
             "Optional API key for authenticating requests to Qdrant.",
         },
       ];
+    case "llamacloud":
+      return [
+        {
+          name: "LLAMA_CLOUD_INDEX_NAME",
+          description:
+            "The name of the LlamaCloud index to use (part of the LlamaCloud project).",
+          value: "test",
+        },
+        {
+          name: "LLAMA_CLOUD_PROJECT_NAME",
+          description: "The name of the LlamaCloud project.",
+          value: "Default",
+        },
+        {
+          name: "LLAMA_CLOUD_BASE_URL",
+          description:
+            "The base URL for the LlamaCloud API. Only change this for non-production environments",
+          value: "https://api.cloud.llamaindex.ai",
+        },
+      ];
     case "chroma":
       const envs = [
         {
@@ -215,6 +238,15 @@ const getModelEnvs = (modelConfig: ModelConfig): EnvVar[] => {
           },
         ]
       : []),
+    ...(modelConfig.provider === "groq"
+      ? [
+          {
+            name: "GROQ_API_KEY",
+            description: "The Groq API key to use.",
+            value: modelConfig.apiKey,
+          },
+        ]
+      : []),
     ...(modelConfig.provider === "gemini"
       ? [
           {
@@ -233,6 +265,30 @@ const getModelEnvs = (modelConfig: ModelConfig): EnvVar[] => {
           },
         ]
       : []),
+    ...(modelConfig.provider === "mistral"
+      ? [
+          {
+            name: "MISTRAL_API_KEY",
+            description: "The Mistral API key to use.",
+            value: modelConfig.apiKey,
+          },
+        ]
+      : []),
+    ...(modelConfig.provider === "t-systems"
+      ? [
+          {
+            name: "T_SYSTEMS_LLMHUB_BASE_URL",
+            description:
+              "The base URL for the T-Systems AI Foundation Model API. Eg: http://localhost:11434",
+            value: TSYSTEMS_LLMHUB_API_URL,
+          },
+          {
+            name: "T_SYSTEMS_LLMHUB_API_KEY",
+            description: "API Key for T-System's AI Foundation Model.",
+            value: modelConfig.apiKey,
+          },
+        ]
+      : []),
   ];
 };
 
@@ -332,6 +388,36 @@ const getSystemPromptEnv = (tools?: Tool[]): EnvVar => {
   };
 };
 
+const getTemplateEnvs = (template?: TemplateType): EnvVar[] => {
+  if (template === "multiagent") {
+    return [
+      {
+        name: "MESSAGE_QUEUE_PORT",
+      },
+      {
+        name: "CONTROL_PLANE_PORT",
+      },
+      {
+        name: "HUMAN_CONSUMER_PORT",
+      },
+      {
+        name: "AGENT_QUERY_ENGINE_PORT",
+        value: "8003",
+      },
+      {
+        name: "AGENT_QUERY_ENGINE_DESCRIPTION",
+        value: "Query information from the provided data",
+      },
+      {
+        name: "AGENT_DUMMY_PORT",
+        value: "8004",
+      },
+    ];
+  } else {
+    return [];
+  }
+};
+
 export const createBackendEnvFile = async (
   root: string,
   opts: {
@@ -340,6 +426,7 @@ export const createBackendEnvFile = async (
     modelConfig: ModelConfig;
     framework: TemplateFramework;
     dataSources?: TemplateDataSource[];
+    template?: TemplateType;
     port?: number;
     tools?: Tool[];
   },
@@ -360,6 +447,8 @@ export const createBackendEnvFile = async (
     ...getVectorDBEnvs(opts.vectorDb, opts.framework),
     ...getFrameworkEnvs(opts.framework, opts.port),
     ...getToolEnvs(opts.tools),
+    // Add template environment variables
+    ...getTemplateEnvs(opts.template),
     getSystemPromptEnv(opts.tools),
   ];
   // Render and write env file

helpers/index.ts

@@ -141,15 +141,18 @@ 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 createBackendEnvFile(props.root, {
-      modelConfig: props.modelConfig,
-      llamaCloudKey: props.llamaCloudKey,
-      vectorDb: props.vectorDb,
-      framework: props.framework,
-      dataSources: props.dataSources,
-      port: props.externalPort,
-      tools: props.tools,
-    });
+    if (props.template === "streaming" || props.template === "multiagent") {
+      await createBackendEnvFile(props.root, {
+        modelConfig: props.modelConfig,
+        llamaCloudKey: props.llamaCloudKey,
+        vectorDb: props.vectorDb,
+        framework: props.framework,
+        dataSources: props.dataSources,
+        port: props.externalPort,
+        tools: props.tools,
+        template: props.template,
+      });
+    }
 
     if (props.dataSources.length > 0) {
       console.log("\nGenerating context data...\n");

helpers/providers/groq.ts

@@ -0,0 +1,99 @@
+import ciInfo from "ci-info";
+import prompts from "prompts";
+import { ModelConfigParams } from ".";
+import { questionHandlers, toChoice } from "../../questions";
+
+const MODELS = ["llama3-8b", "llama3-70b", "mixtral-8x7b"];
+const DEFAULT_MODEL = MODELS[0];
+
+// Use huggingface embedding models for now as Groq doesn't support embedding models
+enum HuggingFaceEmbeddingModelType {
+  XENOVA_ALL_MINILM_L6_V2 = "all-MiniLM-L6-v2",
+  XENOVA_ALL_MPNET_BASE_V2 = "all-mpnet-base-v2",
+}
+type ModelData = {
+  dimensions: number;
+};
+const EMBEDDING_MODELS: Record<HuggingFaceEmbeddingModelType, ModelData> = {
+  [HuggingFaceEmbeddingModelType.XENOVA_ALL_MINILM_L6_V2]: {
+    dimensions: 384,
+  },
+  [HuggingFaceEmbeddingModelType.XENOVA_ALL_MPNET_BASE_V2]: {
+    dimensions: 768,
+  },
+};
+const DEFAULT_EMBEDDING_MODEL = Object.keys(EMBEDDING_MODELS)[0];
+const DEFAULT_DIMENSIONS = Object.values(EMBEDDING_MODELS)[0].dimensions;
+
+type GroqQuestionsParams = {
+  apiKey?: string;
+  askModels: boolean;
+};
+
+export async function askGroqQuestions({
+  askModels,
+  apiKey,
+}: GroqQuestionsParams): Promise<ModelConfigParams> {
+  const config: ModelConfigParams = {
+    apiKey,
+    model: DEFAULT_MODEL,
+    embeddingModel: DEFAULT_EMBEDDING_MODEL,
+    dimensions: DEFAULT_DIMENSIONS,
+    isConfigured(): boolean {
+      if (config.apiKey) {
+        return true;
+      }
+      if (process.env["GROQ_API_KEY"]) {
+        return true;
+      }
+      return false;
+    },
+  };
+
+  if (!config.apiKey) {
+    const { key } = await prompts(
+      {
+        type: "text",
+        name: "key",
+        message:
+          "Please provide your Groq API key (or leave blank to use GROQ_API_KEY env variable):",
+      },
+      questionHandlers,
+    );
+    config.apiKey = key || process.env.GROQ_API_KEY;
+  }
+
+  // use default model values in CI or if user should not be asked
+  const useDefaults = ciInfo.isCI || !askModels;
+  if (!useDefaults) {
+    const { model } = await prompts(
+      {
+        type: "select",
+        name: "model",
+        message: "Which LLM model would you like to use?",
+        choices: MODELS.map(toChoice),
+        initial: 0,
+      },
+      questionHandlers,
+    );
+    config.model = model;
+
+    const { embeddingModel } = await prompts(
+      {
+        type: "select",
+        name: "embeddingModel",
+        message: "Which embedding model would you like to use?",
+        choices: Object.keys(EMBEDDING_MODELS).map(toChoice),
+        initial: 0,
+      },
+      questionHandlers,
+    );
+    config.embeddingModel = embeddingModel;
+    config.dimensions =
+      EMBEDDING_MODELS[
+        embeddingModel as HuggingFaceEmbeddingModelType
+      ].dimensions;
+  }
+
+  return config;
+}

helpers/providers/index.ts

@@ -1,9 +1,12 @@
 import ciInfo from "ci-info";
 import prompts from "prompts";
 import { questionHandlers } from "../../questions";
-import { ModelConfig, ModelProvider } from "../types";
+import { ModelConfig, ModelProvider, TemplateFramework } from "../types";
 import { askAnthropicQuestions } from "./anthropic";
 import { askGeminiQuestions } from "./gemini";
+import { askGroqQuestions } from "./groq";
+import { askLLMHubQuestions } from "./llmhub";
+import { askMistralQuestions } from "./mistral";
 import { askOllamaQuestions } from "./ollama";
 import { askOpenAIQuestions } from "./openai";
 
@@ -12,6 +15,7 @@ const DEFAULT_MODEL_PROVIDER = "openai";
 export type ModelConfigQuestionsParams = {
   openAiKey?: string;
   askModels: boolean;
+  framework?: TemplateFramework;
 };
 
 export type ModelConfigParams = Omit<ModelConfig, "provider">;
@@ -19,23 +23,28 @@ export type ModelConfigParams = Omit<ModelConfig, "provider">;
 export async function askModelConfig({
   askModels,
   openAiKey,
+  framework,
 }: ModelConfigQuestionsParams): Promise<ModelConfig> {
   let modelProvider: ModelProvider = DEFAULT_MODEL_PROVIDER;
   if (askModels && !ciInfo.isCI) {
+    let choices = [
+      { title: "OpenAI", value: "openai" },
+      { title: "Groq", value: "groq" },
+      { title: "Ollama", value: "ollama" },
+      { title: "Anthropic", value: "anthropic" },
+      { title: "Gemini", value: "gemini" },
+      { title: "Mistral", value: "mistral" },
+    ];
+
+    if (framework === "fastapi") {
+      choices.push({ title: "T-Systems", value: "t-systems" });
+    }
     const { provider } = await prompts(
       {
         type: "select",
         name: "provider",
         message: "Which model provider would you like to use",
-        choices: [
-          {
-            title: "OpenAI",
-            value: "openai",
-          },
-          { title: "Ollama", value: "ollama" },
-          { title: "Anthropic", value: "anthropic" },
-          { title: "Gemini", value: "gemini" },
-        ],
+        choices: choices,
         initial: 0,
       },
       questionHandlers,
@@ -48,12 +57,21 @@ export async function askModelConfig({
     case "ollama":
       modelConfig = await askOllamaQuestions({ askModels });
       break;
+    case "groq":
+      modelConfig = await askGroqQuestions({ askModels });
+      break;
     case "anthropic":
       modelConfig = await askAnthropicQuestions({ askModels });
       break;
     case "gemini":
       modelConfig = await askGeminiQuestions({ askModels });
       break;
+    case "mistral":
+      modelConfig = await askMistralQuestions({ askModels });
+      break;
+    case "t-systems":
+      modelConfig = await askLLMHubQuestions({ askModels });
+      break;
     default:
       modelConfig = await askOpenAIQuestions({
         openAiKey,

helpers/providers/llmhub.ts

@@ -0,0 +1,169 @@
+import ciInfo from "ci-info";
+import got from "got";
+import ora from "ora";
+import { red } from "picocolors";
+import prompts from "prompts";
+import { ModelConfigParams } from ".";
+import { questionHandlers } from "../../questions";
+
+export const TSYSTEMS_LLMHUB_API_URL =
+  "https://llm-server.llmhub.t-systems.net/v2";
+
+const DEFAULT_MODEL = "gpt-3.5-turbo";
+const DEFAULT_EMBEDDING_MODEL = "text-embedding-3-large";
+
+const LLMHUB_MODELS = [
+  "gpt-35-turbo",
+  "gpt-4-32k-1",
+  "gpt-4-32k-canada",
+  "gpt-4-32k-france",
+  "gpt-4-turbo-128k-france",
+  "Llama2-70b-Instruct",
+  "Llama-3-70B-Instruct",
+  "Mixtral-8x7B-Instruct-v0.1",
+  "mistral-large-32k-france",
+  "CodeLlama-2",
+];
+const LLMHUB_EMBEDDING_MODELS = [
+  "text-embedding-ada-002",
+  "text-embedding-ada-002-france",
+  "jina-embeddings-v2-base-de",
+  "jina-embeddings-v2-base-code",
+  "text-embedding-bge-m3",
+];
+
+type LLMHubQuestionsParams = {
+  apiKey?: string;
+  askModels: boolean;
+};
+
+export async function askLLMHubQuestions({
+  askModels,
+  apiKey,
+}: LLMHubQuestionsParams): Promise<ModelConfigParams> {
+  const config: ModelConfigParams = {
+    apiKey,
+    model: DEFAULT_MODEL,
+    embeddingModel: DEFAULT_EMBEDDING_MODEL,
+    dimensions: getDimensions(DEFAULT_EMBEDDING_MODEL),
+    isConfigured(): boolean {
+      if (config.apiKey) {
+        return true;
+      }
+      if (process.env["T_SYSTEMS_LLMHUB_API_KEY"]) {
+        return true;
+      }
+      return false;
+    },
+  };
+
+  if (!config.apiKey) {
+    const { key } = await prompts(
+      {
+        type: "text",
+        name: "key",
+        message: askModels
+          ? "Please provide your LLMHub API key (or leave blank to use T_SYSTEMS_LLMHUB_API_KEY env variable):"
+          : "Please provide your LLMHub API key (leave blank to skip):",
+        validate: (value: string) => {
+          if (askModels && !value) {
+            if (process.env.T_SYSTEMS_LLMHUB_API_KEY) {
+              return true;
+            }
+            return "T_SYSTEMS_LLMHUB_API_KEY env variable is not set - key is required";
+          }
+          return true;
+        },
+      },
+      questionHandlers,
+    );
+    config.apiKey = key || process.env.T_SYSTEMS_LLMHUB_API_KEY;
+  }
+
+  // use default model values in CI or if user should not be asked
+  const useDefaults = ciInfo.isCI || !askModels;
+  if (!useDefaults) {
+    const { model } = await prompts(
+      {
+        type: "select",
+        name: "model",
+        message: "Which LLM model would you like to use?",
+        choices: await getAvailableModelChoices(false, config.apiKey),
+        initial: 0,
+      },
+      questionHandlers,
+    );
+    config.model = model;
+
+    const { embeddingModel } = await prompts(
+      {
+        type: "select",
+        name: "embeddingModel",
+        message: "Which embedding model would you like to use?",
+        choices: await getAvailableModelChoices(true, config.apiKey),
+        initial: 0,
+      },
+      questionHandlers,
+    );
+    config.embeddingModel = embeddingModel;
+    config.dimensions = getDimensions(embeddingModel);
+  }
+
+  return config;
+}
+
+async function getAvailableModelChoices(
+  selectEmbedding: boolean,
+  apiKey?: string,
+) {
+  if (!apiKey) {
+    throw new Error("Need LLMHub key to retrieve model choices");
+  }
+  const isLLMModel = (modelId: string) => {
+    return LLMHUB_MODELS.includes(modelId);
+  };
+
+  const isEmbeddingModel = (modelId: string) => {
+    return LLMHUB_EMBEDDING_MODELS.includes(modelId);
+  };
+
+  const spinner = ora("Fetching available models").start();
+  try {
+    const response = await got(`${TSYSTEMS_LLMHUB_API_URL}/models`, {
+      headers: {
+        Authorization: "Bearer " + apiKey,
+      },
+      timeout: 5000,
+      responseType: "json",
+    });
+    const data: any = await response.body;
+    spinner.stop();
+    return data.data
+      .filter((model: any) =>
+        selectEmbedding ? isEmbeddingModel(model.id) : isLLMModel(model.id),
+      )
+      .map((el: any) => {
+        return {
+          title: el.id,
+          value: el.id,
+        };
+      });
+  } catch (error) {
+    spinner.stop();
+    if ((error as any).response?.statusCode === 401) {
+      console.log(
+        red(
+          "Invalid LLMHub API key provided! Please provide a valid key and try again!",
+        ),
+      );
+    } else {
+      console.log(red("Request failed: " + error));
+    }
+    process.exit(1);
+  }
+}
+
+function getDimensions(modelName: string) {
+  // Assuming dimensions similar to OpenAI for simplicity. Update if different.
+  return modelName === "text-embedding-004" ? 768 : 1536;
+}

helpers/providers/mistral.ts

@@ -0,0 +1,86 @@
+import ciInfo from "ci-info";
+import prompts from "prompts";
+import { ModelConfigParams } from ".";
+import { questionHandlers, toChoice } from "../../questions";
+
+const MODELS = ["mistral-tiny", "mistral-small", "mistral-medium"];
+type ModelData = {
+  dimensions: number;
+};
+const EMBEDDING_MODELS: Record<string, ModelData> = {
+  "mistral-embed": { dimensions: 1024 },
+};
+
+const DEFAULT_MODEL = MODELS[0];
+const DEFAULT_EMBEDDING_MODEL = Object.keys(EMBEDDING_MODELS)[0];
+const DEFAULT_DIMENSIONS = Object.values(EMBEDDING_MODELS)[0].dimensions;
+
+type MistralQuestionsParams = {
+  apiKey?: string;
+  askModels: boolean;
+};
+
+export async function askMistralQuestions({
+  askModels,
+  apiKey,
+}: MistralQuestionsParams): Promise<ModelConfigParams> {
+  const config: ModelConfigParams = {
+    apiKey,
+    model: DEFAULT_MODEL,
+    embeddingModel: DEFAULT_EMBEDDING_MODEL,
+    dimensions: DEFAULT_DIMENSIONS,
+    isConfigured(): boolean {
+      if (config.apiKey) {
+        return true;
+      }
+      if (process.env["MISTRAL_API_KEY"]) {
+        return true;
+      }
+      return false;
+    },
+  };
+
+  if (!config.apiKey) {
+    const { key } = await prompts(
+      {
+        type: "text",
+        name: "key",
+        message:
+          "Please provide your Mistral API key (or leave blank to use MISTRAL_API_KEY env variable):",
+      },
+      questionHandlers,
+    );
+    config.apiKey = key || process.env.MISTRAL_API_KEY;
+  }
+
+  // use default model values in CI or if user should not be asked
+  const useDefaults = ciInfo.isCI || !askModels;
+  if (!useDefaults) {
+    const { model } = await prompts(
+      {
+        type: "select",
+        name: "model",
+        message: "Which LLM model would you like to use?",
+        choices: MODELS.map(toChoice),
+        initial: 0,
+      },
+      questionHandlers,
+    );
+    config.model = model;
+
+    const { embeddingModel } = await prompts(
+      {
+        type: "select",
+        name: "embeddingModel",
+        message: "Which embedding model would you like to use?",
+        choices: Object.keys(EMBEDDING_MODELS).map(toChoice),
+        initial: 0,
+      },
+      questionHandlers,
+    );
+    config.embeddingModel = embeddingModel;
+    config.dimensions = EMBEDDING_MODELS[embeddingModel].dimensions;
+  }
+
+  return config;
+}

helpers/python.ts

@@ -118,6 +118,12 @@ const getAdditionalDependencies = (
             version: "^2.9.9",
           });
           break;
+        case "llamacloud":
+          dependencies.push({
+            name: "llama-index-indices-managed-llama-cloud",
+            version: "^0.2.1",
+          });
+          break;
       }
     }
   }
@@ -167,6 +173,26 @@ const getAdditionalDependencies = (
         version: "0.1.6",
       });
       break;
+    case "mistral":
+      dependencies.push({
+        name: "llama-index-llms-mistralai",
+        version: "0.1.17",
+      });
+      dependencies.push({
+        name: "llama-index-embeddings-mistralai",
+        version: "0.1.4",
+      });
+      break;
+    case "t-systems":
+      dependencies.push({
+        name: "llama-index-agent-openai",
+        version: "0.2.2",
+      });
+      dependencies.push({
+        name: "llama-index-llms-openai-like",
+        version: "0.1.3",
+      });
+      break;
   }
 
   return dependencies;
@@ -304,20 +330,27 @@ export const installPythonTemplate = async ({
     cwd: path.join(compPath, "loaders", "python"),
   });
 
-  // Select and copy engine code based on data sources and tools
-  let engine;
-  tools = tools ?? [];
-  if (dataSources.length > 0 && tools.length === 0) {
-    console.log("\nNo tools selected - use optimized context chat engine\n");
-    engine = "chat";
-  } else {
-    engine = "agent";
-  }
-  await copy("**", enginePath, {
-    parents: true,
-    cwd: path.join(compPath, "engines", "python", engine),
+  // Copy settings.py to app
+  await copy("**", path.join(root, "app"), {
+    cwd: path.join(compPath, "settings", "python"),
   });
 
+  if (template === "streaming") {
+    // For the streaming template only:
+    // Select and copy engine code based on data sources and tools
+    let engine;
+    if (dataSources.length > 0 && (!tools || tools.length === 0)) {
+      console.log("\nNo tools selected - use optimized context chat engine\n");
+      engine = "chat";
+    } else {
+      engine = "agent";
+    }
+    await copy("**", enginePath, {
+      parents: true,
+      cwd: path.join(compPath, "engines", "python", engine),
+    });
+  }
+
   console.log("Adding additional dependencies");
 
   const addOnDependencies = getAdditionalDependencies(

helpers/tools.ts

@@ -54,6 +54,29 @@ export const supportedTools: Tool[] = [
       },
     ],
   },
+  {
+    // For python app, we will use a local DuckDuckGo search tool (instead of DuckDuckGo search tool in LlamaHub)
+    // to get the same results as the TS app.
+    display: "DuckDuckGo Search",
+    name: "duckduckgo",
+    dependencies: [
+      {
+        name: "duckduckgo-search",
+        version: "6.1.7",
+      },
+    ],
+    supportedFrameworks: ["fastapi", "nextjs", "express"],
+    type: ToolType.LOCAL,
+    envVars: [
+      {
+        name: TOOL_SYSTEM_PROMPT_ENV_VAR,
+        description: "System prompt for DuckDuckGo search tool.",
+        value: `You are a DuckDuckGo search agent. 
+You can use the duckduckgo search tool to get information from the web to answer user questions.
+For better results, you can specify the region parameter to get results from a specific region but it's optional.`,
+      },
+    ],
+  },
   {
     display: "Wikipedia",
     name: "wikipedia.WikipediaToolSpec",
@@ -107,13 +130,12 @@ export const supportedTools: Tool[] = [
       {
         name: TOOL_SYSTEM_PROMPT_ENV_VAR,
         description: "System prompt for code interpreter tool.",
-        value: `You are a Python interpreter.
-        - You are given tasks to complete and you run python code to solve them.
-        - The python code runs in a Jupyter notebook. Every time you call \`interpreter\` tool, the python code is executed in a separate cell. It's okay to make multiple calls to \`interpreter\`.
-        - Display visualizations using matplotlib or any other visualization library directly in the notebook. Shouldn't save the visualizations to a file, just return the base64 encoded data.
-        - You can install any pip package (if it exists) if you need to but the usual packages for data analysis are already preinstalled.
-        - You can run any python code you want in a secure environment.
-        - Use absolute url from result to display images or any other media.`,
+        value: `-You are a Python interpreter that can run any python code in a secure environment.
+- The python code runs in a Jupyter notebook. Every time you call the 'interpreter' tool, the python code is executed in a separate cell. 
+- You are given tasks to complete and you run python code to solve them.
+- It's okay to make multiple calls to interpreter tool. If you get an error or the result is not what you expected, you can call the tool again. Don't give up too soon!
+- Plot visualizations using matplotlib or any other visualization library directly in the notebook.
+- You can install any pip package (if it exists) by running a cell with pip install.`,
       },
     ],
   },
@@ -137,7 +159,7 @@ export const supportedTools: Tool[] = [
     config: {
       openapi_uri: "The URL or file path of the OpenAPI schema",
     },
-    supportedFrameworks: ["fastapi"],
+    supportedFrameworks: ["fastapi", "express", "nextjs"],
     type: ToolType.LOCAL,
     envVars: [
       {
@@ -148,6 +170,24 @@ export const supportedTools: Tool[] = [
       },
     ],
   },
+  {
+    display: "Image Generator",
+    name: "img_gen",
+    supportedFrameworks: ["fastapi", "express", "nextjs"],
+    type: ToolType.LOCAL,
+    envVars: [
+      {
+        name: "STABILITY_API_KEY",
+        description:
+          "STABILITY_API_KEY key is required to run image generator. Get it here: https://platform.stability.ai/account/keys",
+      },
+      {
+        name: TOOL_SYSTEM_PROMPT_ENV_VAR,
+        description: "System prompt for image generator tool.",
+        value: `You are an image generator agent. You help users to generate images using the Stability API.`,
+      },
+    ],
+  },
 ];
 
 export const getTool = (toolName: string): Tool | undefined => {

helpers/types.ts

@@ -1,7 +1,14 @@
 import { PackageManager } from "../helpers/get-pkg-manager";
 import { Tool } from "./tools";
 
-export type ModelProvider = "openai" | "ollama" | "anthropic" | "gemini";
+export type ModelProvider =
+  | "openai"
+  | "groq"
+  | "ollama"
+  | "anthropic"
+  | "gemini"
+  | "mistral"
+  | "t-systems";
 export type ModelConfig = {
   provider: ModelProvider;
   apiKey?: string;
@@ -10,7 +17,11 @@ export type ModelConfig = {
   dimensions: number;
   isConfigured(): boolean;
 };
-export type TemplateType = "streaming" | "community" | "llamapack";
+export type TemplateType =
+  | "streaming"
+  | "community"
+  | "llamapack"
+  | "multiagent";
 export type TemplateFramework = "nextjs" | "express" | "fastapi";
 export type TemplateUI = "html" | "shadcn";
 export type TemplateVectorDB =
@@ -21,7 +32,8 @@ export type TemplateVectorDB =
   | "milvus"
   | "astra"
   | "qdrant"
-  | "chroma";
+  | "chroma"
+  | "llamacloud";
 export type TemplatePostInstallAction =
   | "none"
   | "VSCode"
@@ -31,7 +43,7 @@ export type TemplateDataSource = {
   type: TemplateDataSourceType;
   config: TemplateDataSourceConfig;
 };
-export type TemplateDataSourceType = "file" | "web" | "db";
+export type TemplateDataSourceType = "file" | "web" | "db" | "llamacloud";
 export type TemplateObservability = "none" | "opentelemetry";
 // Config for both file and folder
 export type FileSourceConfig = {

helpers/typescript.ts

@@ -1,7 +1,7 @@
 import fs from "fs/promises";
 import os from "os";
 import path from "path";
-import { bold, cyan } from "picocolors";
+import { bold, cyan, yellow } from "picocolors";
 import { assetRelocator, copy } from "../helpers/copy";
 import { callPackageManager } from "../helpers/install";
 import { templatesDir } from "./dir";
@@ -105,7 +105,13 @@ export const installTSTemplate = async ({
   const enginePath = path.join(root, relativeEngineDestPath, "engine");
 
   // copy vector db component
-  console.log("\nUsing vector DB:", vectorDb ?? "none", "\n");
+  if (vectorDb === "llamacloud") {
+    console.log(
+      `\nUsing managed index from LlamaCloud. Ensure the ${yellow("LLAMA_CLOUD_* environment variables are set correctly.")}`,
+    );
+  } else {
+    console.log("\nUsing vector DB:", vectorDb ?? "none");
+  }
   await copy("**", enginePath, {
     parents: true,
     cwd: path.join(compPath, "vectordbs", "typescript", vectorDb ?? "none"),

package.json

@@ -1,6 +1,6 @@
 {
   "name": "create-llama",
-  "version": "0.1.9",
+  "version": "0.1.18",
   "description": "Create LlamaIndex-powered apps with one command",
   "keywords": [
     "rag",

questions.ts

@@ -9,6 +9,7 @@ import {
   TemplateDataSource,
   TemplateDataSourceType,
   TemplateFramework,
+  TemplateType,
 } from "./helpers";
 import { COMMUNITY_OWNER, COMMUNITY_REPO } from "./helpers/constant";
 import { EXAMPLE_FILE } from "./helpers/datasources";
@@ -122,8 +123,15 @@ const getVectorDbChoices = (framework: TemplateFramework) => {
 export const getDataSourceChoices = (
   framework: TemplateFramework,
   selectedDataSource: TemplateDataSource[],
+  template?: TemplateType,
 ) => {
+  // If LlamaCloud is already selected, don't show any other options
+  if (selectedDataSource.find((s) => s.type === "llamacloud")) {
+    return [];
+  }
+
   const choices = [];
+
   if (selectedDataSource.length > 0) {
     choices.push({
       title: "No",
@@ -131,29 +139,37 @@ export const getDataSourceChoices = (
     });
   }
   if (selectedDataSource === undefined || selectedDataSource.length === 0) {
+    if (template !== "multiagent") {
+      choices.push({
+        title: "No data, just a simple chat or agent",
+        value: "none",
+      });
+    }
     choices.push({
-      title: "No data, just a simple chat or agent",
-      value: "none",
-    });
-    choices.push({
-      title: "Use an example PDF",
+      title:
+        process.platform !== "linux"
+          ? "Use an example PDF"
+          : "Use an example PDF (you can add your own data files later)",
       value: "exampleFile",
     });
   }
 
-  choices.push(
-    {
-      title: `Use local files (${supportedContextFileTypes.join(", ")})`,
-      value: "file",
-    },
-    {
-      title:
-        process.platform === "win32"
-          ? "Use a local folder"
-          : "Use local folders",
-      value: "folder",
-    },
-  );
+  // Linux has many distros so we won't support file/folder picker for now
+  if (process.platform !== "linux") {
+    choices.push(
+      {
+        title: `Use local files (${supportedContextFileTypes.join(", ")})`,
+        value: "file",
+      },
+      {
+        title:
+          process.platform === "win32"
+            ? "Use a local folder"
+            : "Use local folders",
+        value: "folder",
+      },
+    );
+  }
 
   if (framework === "fastapi") {
     choices.push({
@@ -165,6 +181,13 @@ export const getDataSourceChoices = (
       value: "db",
     });
   }
+
+  if (!selectedDataSource.length) {
+    choices.push({
+      title: "Use managed index from LlamaCloud",
+      value: "llamacloud",
+    });
+  }
   return choices;
 };
 
@@ -262,25 +285,27 @@ export const askQuestions = async (
           },
         ];
 
-        const modelConfigured =
-          !program.llamapack && program.modelConfig.isConfigured();
-        // If using LlamaParse, require LlamaCloud API key
-        const llamaCloudKeyConfigured = program.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 &&
-          modelConfigured &&
-          llamaCloudKeyConfigured &&
-          !toolsRequireConfig(program.tools)
-        ) {
-          actionChoices.push({
-            title:
-              "Generate code, install dependencies, and run the app (~2 min)",
-            value: "runApp",
-          });
+        if (program.template !== "multiagent") {
+          const modelConfigured =
+            !program.llamapack && program.modelConfig.isConfigured();
+          // If using LlamaParse, require LlamaCloud API key
+          const llamaCloudKeyConfigured = program.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 &&
+            modelConfigured &&
+            llamaCloudKeyConfigured &&
+            !toolsRequireConfig(program.tools)
+          ) {
+            actionChoices.push({
+              title:
+                "Generate code, install dependencies, and run the app (~2 min)",
+              value: "runApp",
+            });
+          }
         }
 
         const { action } = await prompts(
@@ -312,7 +337,11 @@ export const askQuestions = async (
           name: "template",
           message: "Which template would you like to use?",
           choices: [
-            { title: "Chat", value: "streaming" },
+            { title: "Agentic RAG (single agent)", value: "streaming" },
+            {
+              title: "Multi-agent app (using llama-agents)",
+              value: "multiagent",
+            },
             {
               title: `Community template from ${styledRepo}`,
               value: "community",
@@ -376,6 +405,10 @@ export const askQuestions = async (
     return; // early return - no further questions needed for llamapack projects
   }
 
+  if (program.template === "multiagent") {
+    // TODO: multi-agents currently only supports FastAPI
+    program.framework = preferences.framework = "fastapi";
+  }
   if (!program.framework) {
     if (ciInfo.isCI) {
       program.framework = getPrefOrDefault("framework");
@@ -401,7 +434,10 @@ export const askQuestions = async (
     }
   }
 
-  if (program.framework === "express" || program.framework === "fastapi") {
+  if (
+    (program.framework === "express" || program.framework === "fastapi") &&
+    program.template === "streaming"
+  ) {
     // if a backend-only framework is selected, ask whether we should create a frontend
     if (program.frontend === undefined) {
       if (ciInfo.isCI) {
@@ -438,7 +474,7 @@ export const askQuestions = async (
     }
   }
 
-  if (!program.observability) {
+  if (!program.observability && program.template === "streaming") {
     if (ciInfo.isCI) {
       program.observability = getPrefOrDefault("observability");
     } else {
@@ -465,6 +501,7 @@ export const askQuestions = async (
     const modelConfig = await askModelConfig({
       openAiKey,
       askModels: program.askModels ?? false,
+      framework: program.framework,
     });
     program.modelConfig = modelConfig;
     preferences.modelConfig = modelConfig;
@@ -478,6 +515,12 @@ export const askQuestions = async (
       // continue asking user for data sources if none are initially provided
       while (true) {
         const firstQuestion = program.dataSources.length === 0;
+        const choices = getDataSourceChoices(
+          program.framework,
+          program.dataSources,
+          program.template,
+        );
+        if (choices.length === 0) break;
         const { selectedSource } = await prompts(
           {
             type: "select",
@@ -485,10 +528,7 @@ export const askQuestions = async (
             message: firstQuestion
               ? "Which data source would you like to use?"
               : "Would you like to add another data source?",
-            choices: getDataSourceChoices(
-              program.framework,
-              program.dataSources,
-            ),
+            choices,
             initial: firstQuestion ? 1 : 0,
           },
           questionHandlers,
@@ -585,51 +625,76 @@ export const askQuestions = async (
               config: await prompts(dbPrompts, questionHandlers),
             });
           }
+          case "llamacloud": {
+            program.dataSources.push({
+              type: "llamacloud",
+              config: {},
+            });
+            program.dataSources.push(EXAMPLE_FILE);
+            break;
+          }
         }
       }
     }
   }
 
-  // Asking for LlamaParse if user selected file or folder data source
-  if (
-    program.dataSources.some((ds) => ds.type === "file") &&
-    program.useLlamaParse === undefined
-  ) {
-    if (ciInfo.isCI) {
-      program.useLlamaParse = getPrefOrDefault("useLlamaParse");
-      program.llamaCloudKey = getPrefOrDefault("llamaCloudKey");
-    } else {
-      const { useLlamaParse } = await prompts(
-        {
-          type: "toggle",
-          name: "useLlamaParse",
-          message:
-            "Would you like to use LlamaParse (improved parser for RAG - requires API key)?",
-          initial: false,
-          active: "yes",
-          inactive: "no",
-        },
-        questionHandlers,
-      );
-      program.useLlamaParse = useLlamaParse;
+  const isUsingLlamaCloud = program.dataSources.some(
+    (ds) => ds.type === "llamacloud",
+  );
 
-      // Ask for LlamaCloud API key
-      if (useLlamaParse && program.llamaCloudKey === undefined) {
-        const { llamaCloudKey } = await prompts(
+  // Asking for LlamaParse if user selected file data source
+  if (isUsingLlamaCloud) {
+    // default to use LlamaParse if using LlamaCloud
+    program.useLlamaParse = preferences.useLlamaParse = true;
+  } else {
+    if (program.dataSources.some((ds) => ds.type === "file")) {
+      if (ciInfo.isCI) {
+        program.useLlamaParse = getPrefOrDefault("useLlamaParse");
+      } else {
+        const { useLlamaParse } = await prompts(
           {
-            type: "text",
-            name: "llamaCloudKey",
+            type: "toggle",
+            name: "useLlamaParse",
             message:
-              "Please provide your LlamaIndex Cloud API key (leave blank to skip):",
+              "Would you like to use LlamaParse (improved parser for RAG - requires API key)?",
+            initial: false,
+            active: "yes",
+            inactive: "no",
           },
           questionHandlers,
         );
-        program.llamaCloudKey = llamaCloudKey;
+        program.useLlamaParse = useLlamaParse;
+        preferences.useLlamaParse = useLlamaParse;
       }
     }
   }
 
-  if (program.dataSources.length > 0 && !program.vectorDb) {
+  // Ask for LlamaCloud API key when using a LlamaCloud index or LlamaParse
+  if (isUsingLlamaCloud || program.useLlamaParse) {
+    if (ciInfo.isCI) {
+      program.llamaCloudKey = getPrefOrDefault("llamaCloudKey");
+    } else {
+      // Ask for LlamaCloud API key
+      const { llamaCloudKey } = await prompts(
+        {
+          type: "text",
+          name: "llamaCloudKey",
+          message:
+            "Please provide your LlamaCloud API key (leave blank to skip):",
+        },
+        questionHandlers,
+      );
+      program.llamaCloudKey = preferences.llamaCloudKey =
+        llamaCloudKey || process.env.LLAMA_CLOUD_API_KEY;
+    }
+  }
+
+  if (isUsingLlamaCloud) {
+    // When using a LlamaCloud index, don't ask for vector database and use code in `llamacloud` folder for vector database
+    const vectorDb = "llamacloud";
+    program.vectorDb = vectorDb;
+    preferences.vectorDb = vectorDb;
+  } else if (program.dataSources.length > 0 && !program.vectorDb) {
     if (ciInfo.isCI) {
       program.vectorDb = getPrefOrDefault("vectorDb");
     } else {
@@ -648,7 +713,8 @@ export const askQuestions = async (
     }
   }
 
-  if (!program.tools) {
+  if (!program.tools && program.template === "streaming") {
+    // TODO: allow to select tools also for multi-agent framework
     if (ciInfo.isCI) {
       program.tools = getPrefOrDefault("tools");
     } else {
@@ -656,7 +722,7 @@ export const askQuestions = async (
         t.supportedFrameworks?.includes(program.framework),
       );
       const toolChoices = options.map((tool) => ({
-        title: `${tool.display}${toolRequiresConfig(tool) ? "" : " (no config needed)"}`,
+        title: `${tool.display}${toolRequiresConfig(tool) ? " (needs configuration)" : ""}`,
         value: tool.name,
       }));
       const { toolsName } = await prompts({

templates/components/engines/python/agent/tools/__init__.py

@@ -19,15 +19,6 @@ class ToolFactory:
         ToolType.LOCAL: "app.engine.tools",
     }
 
-    @staticmethod
-    @cached(
-        LRUCache(maxsize=100),
-        key=lambda tool_type, tool_name, config: (
-            tool_type,
-            tool_name,
-            json.dumps(config, sort_keys=True),
-        ),
-    )
     def load_tools(tool_type: str, tool_name: str, config: dict) -> list[FunctionTool]:
         source_package = ToolFactory.TOOL_SOURCE_PACKAGE_MAP[tool_type]
         try:
@@ -40,7 +31,7 @@ class ToolFactory:
                 return tool_spec.to_tool_list()
             else:
                 module = importlib.import_module(f"{source_package}.{tool_name}")
-                tools = getattr(module, "tools")
+                tools = module.get_tools(**config)
                 if not all(isinstance(tool, FunctionTool) for tool in tools):
                     raise ValueError(
                         f"The module {module} does not contain valid tools"

templates/components/engines/python/agent/tools/duckduckgo.py

@@ -0,0 +1,36 @@
+from llama_index.core.tools.function_tool import FunctionTool
+
+
+def duckduckgo_search(
+    query: str,
+    region: str = "wt-wt",
+    max_results: int = 10,
+):
+    """
+    Use this function to search for any query in DuckDuckGo.
+    Args:
+        query (str): The query to search in DuckDuckGo.
+        region Optional(str): The region to be used for the search in [country-language] convention, ex us-en, uk-en, ru-ru, etc...
+        max_results Optional(int): The maximum number of results to be returned. Default is 10.
+    """
+    try:
+        from duckduckgo_search import DDGS
+    except ImportError:
+        raise ImportError(
+            "duckduckgo_search package is required to use this function."
+            "Please install it by running: `poetry add duckduckgo_search` or `pip install duckduckgo_search`"
+        )
+
+    params = {
+        "keywords": query,
+        "region": region,
+        "max_results": max_results,
+    }
+    results = []
+    with DDGS() as ddg:
+        results = list(ddg.text(**params))
+    return results
+
+
+def get_tools(**kwargs):
+    return [FunctionTool.from_defaults(duckduckgo_search)]

templates/components/engines/python/agent/tools/img_gen.py

@@ -0,0 +1,108 @@
+import os
+import uuid
+import logging
+import requests
+from typing import Optional
+from pydantic import BaseModel, Field
+from llama_index.core.tools import FunctionTool
+
+logger = logging.getLogger(__name__)
+
+
+class ImageGeneratorToolOutput(BaseModel):
+    is_success: bool = Field(
+        ...,
+        description="Whether the image generation was successful.",
+    )
+    image_url: Optional[str] = Field(
+        None,
+        description="The URL of the generated image.",
+    )
+    error_message: Optional[str] = Field(
+        None,
+        description="The error message if the image generation failed.",
+    )
+
+
+class ImageGeneratorTool:
+    _IMG_OUTPUT_FORMAT = "webp"
+    _IMG_OUTPUT_DIR = "tool-output"
+    _IMG_GEN_API = "https://api.stability.ai/v2beta/stable-image/generate/core"
+
+    def __init__(self, api_key: str = None):
+        if not api_key:
+            api_key = os.getenv("STABILITY_API_KEY")
+        self._api_key = api_key
+        self.fileserver_url_prefix = os.getenv("FILESERVER_URL_PREFIX")
+        if self._api_key is None:
+            raise ValueError(
+                "STABILITY_API_KEY key is required to run image generator. Get it here: https://platform.stability.ai/account/keys"
+            )
+        if self.fileserver_url_prefix is None:
+            raise ValueError("FILESERVER_URL_PREFIX is required.")
+
+    def _prepare_output_dir(self):
+        """
+        Create the output directory if it doesn't exist
+        """
+        if not os.path.exists(self._IMG_OUTPUT_DIR):
+            os.makedirs(self._IMG_OUTPUT_DIR, exist_ok=True)
+
+    def _save_image(self, image_data: bytes):
+        self._prepare_output_dir()
+        filename = f"{uuid.uuid4()}.{self._IMG_OUTPUT_FORMAT}"
+        output_path = os.path.join(self._IMG_OUTPUT_DIR, filename)
+        with open(output_path, "wb") as f:
+            f.write(image_data)
+        url = f"{os.getenv('FILESERVER_URL_PREFIX')}/{self._IMG_OUTPUT_DIR}/{filename}"
+        logger.info(f"Saved image to {output_path}.\nURL: {url}")
+        return url
+
+    def _call_stability_api(self, prompt: str):
+        headers = {
+            "authorization": f"Bearer {self._api_key}",
+            "accept": "image/*",
+        }
+        data = {
+            "prompt": prompt,
+            "output_format": self._IMG_OUTPUT_FORMAT,
+        }
+
+        response = requests.post(
+            self._IMG_GEN_API,
+            headers=headers,
+            files={"none": ""},
+            data=data,
+        )
+        response.raise_for_status()
+
+        return response
+
+    def generate_image(self, prompt: str) -> ImageGeneratorToolOutput:
+        """
+        Use this tool to generate an image based on the prompt.
+        Args:
+            prompt (str): The prompt to generate the image from.
+        """
+
+        try:
+            # Call the Stability API
+            response = self._call_stability_api(prompt)
+
+            # Save the image and get the URL
+            image_url = self._save_image(response.content)
+
+            return ImageGeneratorToolOutput(
+                is_success=True,
+                image_url=image_url,
+            )
+        except Exception as e:
+            logger.exception(e, exc_info=True)
+            return ImageGeneratorToolOutput(
+                is_success=False,
+                error_message=str(e),
+            )
+
+
+def get_tools(**kwargs):
+    return [FunctionTool.from_defaults(ImageGeneratorTool(**kwargs).generate_image)]

templates/components/engines/python/agent/tools/interpreter.py

@@ -29,9 +29,24 @@ class E2BCodeInterpreter:
 
     output_dir = "tool-output"
 
-    def __init__(self, api_key: str, filesever_url_prefix: str):
-        self.api_key = api_key
+    def __init__(self, api_key: str = None):
+        if api_key is None:
+            api_key = os.getenv("E2B_API_KEY")
+        filesever_url_prefix = os.getenv("FILESERVER_URL_PREFIX")
+        if not api_key:
+            raise ValueError(
+                "E2B_API_KEY key is required to run code interpreter. Get it here: https://e2b.dev/docs/getting-started/api-key"
+            )
+        if not filesever_url_prefix:
+            raise ValueError(
+                "FILESERVER_URL_PREFIX is required to display file output from sandbox"
+            )
+
         self.filesever_url_prefix = filesever_url_prefix
+        self.interpreter = CodeInterpreter(api_key=api_key)
+
+    def __del__(self):
+        self.interpreter.close()
 
     def get_output_path(self, filename: str) -> str:
         # if output directory doesn't exist, create it
@@ -101,50 +116,28 @@ class E2BCodeInterpreter:
         return output
 
     def interpret(self, code: str) -> E2BToolOutput:
-        with CodeInterpreter(api_key=self.api_key) as interpreter:
-            logger.info(
-                f"\n{'='*50}\n> Running following AI-generated code:\n{code}\n{'='*50}"
-            )
-            exec = interpreter.notebook.exec_cell(code)
+        """
+        Execute python code in a Jupyter notebook cell, the toll will return result, stdout, stderr, display_data, and error.
 
-            if exec.error:
-                logger.error("Error when executing code", exec.error)
-                output = E2BToolOutput(is_error=True, logs=exec.logs, results=[])
+        Parameters:
+            code (str): The python code to be executed in a single cell.
+        """
+        logger.info(
+            f"\n{'='*50}\n> Running following AI-generated code:\n{code}\n{'='*50}"
+        )
+        exec = self.interpreter.notebook.exec_cell(code)
+
+        if exec.error:
+            logger.error("Error when executing code", exec.error)
+            output = E2BToolOutput(is_error=True, logs=exec.logs, results=[])
+        else:
+            if len(exec.results) == 0:
+                output = E2BToolOutput(is_error=False, logs=exec.logs, results=[])
             else:
-                if len(exec.results) == 0:
-                    output = E2BToolOutput(is_error=False, logs=exec.logs, results=[])
-                else:
-                    results = self.parse_result(exec.results[0])
-                    output = E2BToolOutput(
-                        is_error=False, logs=exec.logs, results=results
-                    )
-            return output
+                results = self.parse_result(exec.results[0])
+                output = E2BToolOutput(is_error=False, logs=exec.logs, results=results)
+        return output
 
 
-def code_interpret(code: str) -> Dict:
-    """
-    Execute python code in a Jupyter notebook cell and return any result, stdout, stderr, display_data, and error.
-
-    Parameters:
-        code (str): The python code to be executed in a single cell.
-    """
-    api_key = os.getenv("E2B_API_KEY")
-    filesever_url_prefix = os.getenv("FILESERVER_URL_PREFIX")
-    if not api_key:
-        raise ValueError(
-            "E2B_API_KEY key is required to run code interpreter. Get it here: https://e2b.dev/docs/getting-started/api-key"
-        )
-    if not filesever_url_prefix:
-        raise ValueError(
-            "FILESERVER_URL_PREFIX is required to display file output from sandbox"
-        )
-
-    interpreter = E2BCodeInterpreter(
-        api_key=api_key, filesever_url_prefix=filesever_url_prefix
-    )
-    output = interpreter.interpret(code)
-    return output.dict()
-
-
-# Specify as functions tools to be loaded by the ToolFactory
-tools = [FunctionTool.from_defaults(code_interpret)]
+def get_tools(**kwargs):
+    return [FunctionTool.from_defaults(E2BCodeInterpreter(**kwargs).interpret)]

templates/components/engines/python/agent/tools/openapi_action.py

@@ -12,10 +12,17 @@ class OpenAPIActionToolSpec(OpenAPIToolSpec, RequestsToolSpec):
     """
 
     spec_functions = OpenAPIToolSpec.spec_functions + RequestsToolSpec.spec_functions
+    # Cached parsed specs by URI
+    _specs: Dict[str, Tuple[Dict, List[str]]] = {}
 
-    def __init__(self, openapi_uri: str, domain_headers: dict = {}, **kwargs):
-        # Load the OpenAPI spec
-        openapi_spec, servers = self.load_openapi_spec(openapi_uri)
+    def __init__(self, openapi_uri: str, domain_headers: dict = None, **kwargs):
+        if domain_headers is None:
+            domain_headers = {}
+        if openapi_uri not in self._specs:
+            openapi_spec, servers = self._load_openapi_spec(openapi_uri)
+            self._specs[openapi_uri] = (openapi_spec, servers)
+        else:
+            openapi_spec, servers = self._specs[openapi_uri]
 
         # Add the servers to the domain headers if they are not already present
         for server in servers:
@@ -26,7 +33,7 @@ class OpenAPIActionToolSpec(OpenAPIToolSpec, RequestsToolSpec):
         RequestsToolSpec.__init__(self, domain_headers)
 
     @staticmethod
-    def load_openapi_spec(uri: str) -> Tuple[Dict, List[str]]:
+    def _load_openapi_spec(uri: str) -> Tuple[Dict, List[str]]:
         """
         Load an OpenAPI spec from a URI.
 

templates/components/engines/python/agent/tools/weather.py

@@ -69,4 +69,5 @@ class OpenMeteoWeather:
         return response.json()
 
 
-tools = [FunctionTool.from_defaults(OpenMeteoWeather.get_weather_information)]
+def get_tools(**kwargs):
+    return [FunctionTool.from_defaults(OpenMeteoWeather.get_weather_information)]

templates/components/engines/typescript/agent/chat.ts

@@ -2,7 +2,6 @@ import { BaseToolWithCall, OpenAIAgent, QueryEngineTool } from "llamaindex";
 import fs from "node:fs/promises";
 import path from "node:path";
 import { getDataSource } from "./index";
-import { STORAGE_CACHE_DIR } from "./shared";
 import { createTools } from "./tools";
 
 export async function createChatEngine() {
@@ -17,7 +16,7 @@ export async function createChatEngine() {
         queryEngine: index.asQueryEngine(),
         metadata: {
           name: "data_query_engine",
-          description: `A query engine for documents in storage folder: ${STORAGE_CACHE_DIR}`,
+          description: `A query engine for documents from your data source.`,
         },
       }),
     );

templates/components/engines/typescript/agent/tools/duckduckgo.ts

@@ -0,0 +1,61 @@
+import { JSONSchemaType } from "ajv";
+import { search } from "duck-duck-scrape";
+import { BaseTool, ToolMetadata } from "llamaindex";
+
+export type DuckDuckGoParameter = {
+  query: string;
+  region?: string;
+};
+
+export type DuckDuckGoToolParams = {
+  metadata?: ToolMetadata<JSONSchemaType<DuckDuckGoParameter>>;
+};
+
+const DEFAULT_META_DATA: ToolMetadata<JSONSchemaType<DuckDuckGoParameter>> = {
+  name: "duckduckgo",
+  description: "Use this function to search for any query in DuckDuckGo.",
+  parameters: {
+    type: "object",
+    properties: {
+      query: {
+        type: "string",
+        description: "The query to search in DuckDuckGo.",
+      },
+      region: {
+        type: "string",
+        description:
+          "Optional, The region to be used for the search in [country-language] convention, ex us-en, uk-en, ru-ru, etc...",
+        nullable: true,
+      },
+    },
+    required: ["query"],
+  },
+};
+
+type DuckDuckGoSearchResult = {
+  title: string;
+  description: string;
+  url: string;
+};
+
+export class DuckDuckGoSearchTool implements BaseTool<DuckDuckGoParameter> {
+  metadata: ToolMetadata<JSONSchemaType<DuckDuckGoParameter>>;
+
+  constructor(params: DuckDuckGoToolParams) {
+    this.metadata = params.metadata ?? DEFAULT_META_DATA;
+  }
+
+  async call(input: DuckDuckGoParameter) {
+    const { query, region } = input;
+    const options = region ? { region } : {};
+    const searchResults = await search(query, options);
+
+    return searchResults.results.map((result) => {
+      return {
+        title: result.title,
+        description: result.description,
+        url: result.url,
+      } as DuckDuckGoSearchResult;
+    });
+  }
+}

templates/components/engines/typescript/agent/tools/img-gen.ts

@@ -0,0 +1,112 @@
+import type { JSONSchemaType } from "ajv";
+import { FormData } from "formdata-node";
+import fs from "fs";
+import got from "got";
+import { BaseTool, ToolMetadata } from "llamaindex";
+import path from "node:path";
+import { Readable } from "stream";
+
+export type ImgGeneratorParameter = {
+  prompt: string;
+};
+
+export type ImgGeneratorToolParams = {
+  metadata?: ToolMetadata<JSONSchemaType<ImgGeneratorParameter>>;
+};
+
+export type ImgGeneratorToolOutput = {
+  isSuccess: boolean;
+  imageUrl?: string;
+  errorMessage?: string;
+};
+
+const DEFAULT_META_DATA: ToolMetadata<JSONSchemaType<ImgGeneratorParameter>> = {
+  name: "image_generator",
+  description: `Use this function to generate an image based on the prompt.`,
+  parameters: {
+    type: "object",
+    properties: {
+      prompt: {
+        type: "string",
+        description: "The prompt to generate the image",
+      },
+    },
+    required: ["prompt"],
+  },
+};
+
+export class ImgGeneratorTool implements BaseTool<ImgGeneratorParameter> {
+  readonly IMG_OUTPUT_FORMAT = "webp";
+  readonly IMG_OUTPUT_DIR = "tool-output";
+  readonly IMG_GEN_API =
+    "https://api.stability.ai/v2beta/stable-image/generate/core";
+
+  metadata: ToolMetadata<JSONSchemaType<ImgGeneratorParameter>>;
+
+  constructor(params?: ImgGeneratorToolParams) {
+    this.checkRequiredEnvVars();
+    this.metadata = params?.metadata || DEFAULT_META_DATA;
+  }
+
+  async call(input: ImgGeneratorParameter): Promise<ImgGeneratorToolOutput> {
+    return await this.generateImage(input.prompt);
+  }
+
+  private generateImage = async (
+    prompt: string,
+  ): Promise<ImgGeneratorToolOutput> => {
+    try {
+      const buffer = await this.promptToImgBuffer(prompt);
+      const imageUrl = this.saveImage(buffer);
+      return { isSuccess: true, imageUrl };
+    } catch (error) {
+      console.error(error);
+      return {
+        isSuccess: false,
+        errorMessage: "Failed to generate image. Please try again.",
+      };
+    }
+  };
+
+  private promptToImgBuffer = async (prompt: string) => {
+    const form = new FormData();
+    form.append("prompt", prompt);
+    form.append("output_format", this.IMG_OUTPUT_FORMAT);
+    const buffer = await got
+      .post(this.IMG_GEN_API, {
+        // Not sure why it shows an type error when passing form to body
+        // Although I follow document: https://github.com/sindresorhus/got/blob/main/documentation/2-options.md#body
+        // Tt still works fine, so I make casting to unknown to avoid the typescript warning
+        // Found a similar issue: https://github.com/sindresorhus/got/discussions/1877
+        body: form as unknown as Buffer | Readable | string,
+        headers: {
+          Authorization: `Bearer ${process.env.STABILITY_API_KEY}`,
+          Accept: "image/*",
+        },
+      })
+      .buffer();
+    return buffer;
+  };
+
+  private saveImage = (buffer: Buffer) => {
+    const filename = `${crypto.randomUUID()}.${this.IMG_OUTPUT_FORMAT}`;
+    const outputPath = path.join(this.IMG_OUTPUT_DIR, filename);
+    fs.writeFileSync(outputPath, buffer);
+    const url = `${process.env.FILESERVER_URL_PREFIX}/${this.IMG_OUTPUT_DIR}/${filename}`;
+    console.log(`Saved image to ${outputPath}.\nURL: ${url}`);
+    return url;
+  };
+
+  private checkRequiredEnvVars = () => {
+    if (!process.env.STABILITY_API_KEY) {
+      throw new Error(
+        "STABILITY_API_KEY key is required to run image generator. Get it here: https://platform.stability.ai/account/keys",
+      );
+    }
+    if (!process.env.FILESERVER_URL_PREFIX) {
+      throw new Error(
+        "FILESERVER_URL_PREFIX is required to display file output after generation",
+      );
+    }
+  };
+}

templates/components/engines/typescript/agent/tools/index.ts

@@ -1,42 +1,61 @@
 import { BaseToolWithCall } from "llamaindex";
 import { ToolsFactory } from "llamaindex/tools/ToolsFactory";
+import { DuckDuckGoSearchTool, DuckDuckGoToolParams } from "./duckduckgo";
+import { ImgGeneratorTool, ImgGeneratorToolParams } from "./img-gen";
 import { InterpreterTool, InterpreterToolParams } from "./interpreter";
+import { OpenAPIActionTool } from "./openapi-action";
 import { WeatherTool, WeatherToolParams } from "./weather";
 
-type ToolCreator = (config: unknown) => BaseToolWithCall;
+type ToolCreator = (config: unknown) => Promise<BaseToolWithCall[]>;
 
 export async function createTools(toolConfig: {
   local: Record<string, unknown>;
   llamahub: any;
 }): Promise<BaseToolWithCall[]> {
   // add local tools from the 'tools' folder (if configured)
-  const tools = createLocalTools(toolConfig.local);
+  const tools = await createLocalTools(toolConfig.local);
   // add tools from LlamaIndexTS (if configured)
   tools.push(...(await ToolsFactory.createTools(toolConfig.llamahub)));
   return tools;
 }
 
 const toolFactory: Record<string, ToolCreator> = {
-  weather: (config: unknown) => {
-    return new WeatherTool(config as WeatherToolParams);
+  weather: async (config: unknown) => {
+    return [new WeatherTool(config as WeatherToolParams)];
   },
-  interpreter: (config: unknown) => {
-    return new InterpreterTool(config as InterpreterToolParams);
+  interpreter: async (config: unknown) => {
+    return [new InterpreterTool(config as InterpreterToolParams)];
+  },
+  "openapi_action.OpenAPIActionToolSpec": async (config: unknown) => {
+    const { openapi_uri, domain_headers } = config as {
+      openapi_uri: string;
+      domain_headers: Record<string, Record<string, string>>;
+    };
+    const openAPIActionTool = new OpenAPIActionTool(
+      openapi_uri,
+      domain_headers,
+    );
+    return await openAPIActionTool.toToolFunctions();
+  },
+  duckduckgo: async (config: unknown) => {
+    return [new DuckDuckGoSearchTool(config as DuckDuckGoToolParams)];
+  },
+  img_gen: async (config: unknown) => {
+    return [new ImgGeneratorTool(config as ImgGeneratorToolParams)];
   },
 };
 
-function createLocalTools(
+async function createLocalTools(
   localConfig: Record<string, unknown>,
-): BaseToolWithCall[] {
+): Promise<BaseToolWithCall[]> {
   const tools: BaseToolWithCall[] = [];
 
-  Object.keys(localConfig).forEach((key) => {
+  for (const [key, toolConfig] of Object.entries(localConfig)) {
     if (key in toolFactory) {
-      const toolConfig = localConfig[key];
-      const tool = toolFactory[key](toolConfig);
-      tools.push(tool);
+      const newTools = await toolFactory[key](toolConfig);
+      tools.push(...newTools);
     }
-  });
+  }
 
   return tools;
 }

templates/components/engines/typescript/agent/tools/openapi-action.ts

@@ -0,0 +1,164 @@
+import SwaggerParser from "@apidevtools/swagger-parser";
+import { JSONSchemaType } from "ajv";
+import got from "got";
+import { FunctionTool, JSONValue, ToolMetadata } from "llamaindex";
+
+interface DomainHeaders {
+  [key: string]: { [header: string]: string };
+}
+
+type Input = {
+  url: string;
+  params: object;
+};
+
+type APIInfo = {
+  description: string;
+  title: string;
+};
+
+export class OpenAPIActionTool {
+  // cache the loaded specs by URL
+  private static specs: Record<string, any> = {};
+
+  private readonly INVALID_URL_PROMPT =
+    "This url did not include a hostname or scheme. Please determine the complete URL and try again.";
+
+  private createLoadSpecMetaData = (info: APIInfo) => {
+    return {
+      name: "load_openapi_spec",
+      description: `Use this to retrieve the OpenAPI spec for the API named ${info.title} with the following description: ${info.description}. Call it before making any requests to the API.`,
+    };
+  };
+
+  private readonly createMethodCallMetaData = (
+    method: "POST" | "PATCH" | "GET",
+    info: APIInfo,
+  ) => {
+    return {
+      name: `${method.toLowerCase()}_request`,
+      description: `Use this to call the ${method} method on the API named ${info.title}`,
+      parameters: {
+        type: "object",
+        properties: {
+          url: {
+            type: "string",
+            description: `The url to make the ${method} request against`,
+          },
+          params: {
+            type: "object",
+            description:
+              method === "GET"
+                ? "the URL parameters to provide with the get request"
+                : `the key-value pairs to provide with the ${method} request`,
+          },
+        },
+        required: ["url"],
+      },
+    } as ToolMetadata<JSONSchemaType<Input>>;
+  };
+
+  constructor(
+    public openapi_uri: string,
+    public domainHeaders: DomainHeaders = {},
+  ) {}
+
+  async loadOpenapiSpec(url: string): Promise<any> {
+    const api = await SwaggerParser.validate(url);
+    return {
+      servers: "servers" in api ? api.servers : "",
+      info: { description: api.info.description, title: api.info.title },
+      endpoints: api.paths,
+    };
+  }
+
+  async getRequest(input: Input): Promise<JSONValue> {
+    if (!this.validUrl(input.url)) {
+      return this.INVALID_URL_PROMPT;
+    }
+    try {
+      const data = await got
+        .get(input.url, {
+          headers: this.getHeadersForUrl(input.url),
+          searchParams: input.params as URLSearchParams,
+        })
+        .json();
+      return data as JSONValue;
+    } catch (error) {
+      return error as JSONValue;
+    }
+  }
+
+  async postRequest(input: Input): Promise<JSONValue> {
+    if (!this.validUrl(input.url)) {
+      return this.INVALID_URL_PROMPT;
+    }
+    try {
+      const res = await got.post(input.url, {
+        headers: this.getHeadersForUrl(input.url),
+        json: input.params,
+      });
+      return res.body as JSONValue;
+    } catch (error) {
+      return error as JSONValue;
+    }
+  }
+
+  async patchRequest(input: Input): Promise<JSONValue> {
+    if (!this.validUrl(input.url)) {
+      return this.INVALID_URL_PROMPT;
+    }
+    try {
+      const res = await got.patch(input.url, {
+        headers: this.getHeadersForUrl(input.url),
+        json: input.params,
+      });
+      return res.body as JSONValue;
+    } catch (error) {
+      return error as JSONValue;
+    }
+  }
+
+  public async toToolFunctions() {
+    if (!OpenAPIActionTool.specs[this.openapi_uri]) {
+      console.log(`Loading spec for URL: ${this.openapi_uri}`);
+      const spec = await this.loadOpenapiSpec(this.openapi_uri);
+      OpenAPIActionTool.specs[this.openapi_uri] = spec;
+    }
+    const spec = OpenAPIActionTool.specs[this.openapi_uri];
+    // TODO: read endpoints with parameters from spec and create one tool for each endpoint
+    // For now, we just create a tool for each HTTP method which does not work well for passing parameters
+    return [
+      FunctionTool.from(() => {
+        return spec;
+      }, this.createLoadSpecMetaData(spec.info)),
+      FunctionTool.from(
+        this.getRequest.bind(this),
+        this.createMethodCallMetaData("GET", spec.info),
+      ),
+      FunctionTool.from(
+        this.postRequest.bind(this),
+        this.createMethodCallMetaData("POST", spec.info),
+      ),
+      FunctionTool.from(
+        this.patchRequest.bind(this),
+        this.createMethodCallMetaData("PATCH", spec.info),
+      ),
+    ];
+  }
+
+  private validUrl(url: string): boolean {
+    const parsed = new URL(url);
+    return !!parsed.protocol && !!parsed.hostname;
+  }
+
+  private getDomain(url: string): string {
+    const parsed = new URL(url);
+    return parsed.hostname;
+  }
+
+  private getHeadersForUrl(url: string): { [header: string]: string } {
+    const domain = this.getDomain(url);
+    return this.domainHeaders[domain] || {};
+  }
+}

templates/components/engines/typescript/chat/chat.ts

@@ -8,14 +8,14 @@ export async function createChatEngine() {
       `StorageContext is empty - call 'npm run generate' to generate the storage first`,
     );
   }
-  const retriever = index.asRetriever();
-  retriever.similarityTopK = process.env.TOP_K
-    ? parseInt(process.env.TOP_K)
-    : 3;
+  const retriever = index.asRetriever({
+    similarityTopK: process.env.TOP_K ? parseInt(process.env.TOP_K) : 3,
+  });
 
   return new ContextChatEngine({
     chatModel: Settings.llm,
     retriever,
-    systemPrompt: process.env.SYSTEM_PROMPT,
+    // disable as a custom system prompt disables the generated context
+    // systemPrompt: process.env.SYSTEM_PROMPT,
   });
 }

templates/components/loaders/python/file.py

@@ -1,5 +1,6 @@
 import os
 import logging
+from typing import Dict
 from llama_parse import LlamaParse
 from pydantic import BaseModel, validator
 
@@ -23,24 +24,44 @@ def llama_parse_parser():
             "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")
+    parser = LlamaParse(
+        result_type="markdown",
+        verbose=True,
+        language="en",
+        ignore_errors=False,
+    )
     return parser
 
 
+def llama_parse_extractor() -> Dict[str, LlamaParse]:
+    from llama_parse.utils import SUPPORTED_FILE_TYPES
+
+    parser = llama_parse_parser()
+    return {file_type: parser for file_type in SUPPORTED_FILE_TYPES}
+
+
 def get_file_documents(config: FileLoaderConfig):
     from llama_index.core.readers import SimpleDirectoryReader
 
     try:
+        file_extractor = None
+        if config.use_llama_parse:
+            # LlamaParse is async first,
+            # so we need to use nest_asyncio to run it in sync mode
+            import nest_asyncio
+
+            nest_asyncio.apply()
+
+            file_extractor = llama_parse_extractor()
         reader = SimpleDirectoryReader(
             config.data_dir,
             recursive=True,
             filename_as_id=True,
+            raise_on_error=True,
+            file_extractor=file_extractor,
         )
-        if config.use_llama_parse:
-            parser = llama_parse_parser()
-            reader.file_extractor = {".pdf": parser}
         return reader.load_data()
-    except ValueError as e:
+    except Exception as e:
         import sys, traceback
 
         # Catch the error if the data dir is empty

templates/components/settings/python/llmhub.py

@@ -0,0 +1,61 @@
+from llama_index.embeddings.openai import OpenAIEmbedding
+from llama_index.core.settings import Settings
+from typing import Dict
+import os
+
+DEFAULT_MODEL = "gpt-3.5-turbo"
+DEFAULT_EMBEDDING_MODEL = "text-embedding-3-large"
+
+class TSIEmbedding(OpenAIEmbedding):
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        self._query_engine = self._text_engine = self.model_name
+
+def llm_config_from_env() -> Dict:
+    from llama_index.core.constants import DEFAULT_TEMPERATURE
+
+    model = os.getenv("MODEL", DEFAULT_MODEL)
+    temperature = os.getenv("LLM_TEMPERATURE", DEFAULT_TEMPERATURE)
+    max_tokens = os.getenv("LLM_MAX_TOKENS")
+    api_key = os.getenv("T_SYSTEMS_LLMHUB_API_KEY")
+    api_base = os.getenv("T_SYSTEMS_LLMHUB_BASE_URL")
+
+    config = {
+        "model": model,
+        "api_key": api_key,
+        "api_base": api_base,
+        "temperature": float(temperature),
+        "max_tokens": int(max_tokens) if max_tokens is not None else None,
+    }
+    return config
+
+
+def embedding_config_from_env() -> Dict:
+    from llama_index.core.constants import DEFAULT_EMBEDDING_DIM
+    
+    model = os.getenv("EMBEDDING_MODEL", DEFAULT_EMBEDDING_MODEL)
+    dimension = os.getenv("EMBEDDING_DIM", DEFAULT_EMBEDDING_DIM)
+    api_key = os.getenv("T_SYSTEMS_LLMHUB_API_KEY")
+    api_base = os.getenv("T_SYSTEMS_LLMHUB_BASE_URL")
+
+    config = {
+        "model_name": model,
+        "dimension": int(dimension) if dimension is not None else None,
+        "api_key": api_key,
+        "api_base": api_base,
+    }
+    return config
+
+def init_llmhub():
+    from llama_index.llms.openai_like import OpenAILike
+
+    llm_configs = llm_config_from_env()
+    embedding_configs = embedding_config_from_env()
+
+    Settings.embed_model = TSIEmbedding(**embedding_configs)
+    Settings.llm = OpenAILike(
+        **llm_configs,
+        is_chat_model=True,
+        is_function_calling_model=False,
+        context_window=4096,
+    )

templates/components/settings/python/settings.py

@@ -1,5 +1,6 @@
 import os
 from typing import Dict
+
 from llama_index.core.settings import Settings
 
 
@@ -8,36 +9,50 @@ def init_settings():
     match model_provider:
         case "openai":
             init_openai()
+        case "groq":
+            init_groq()
         case "ollama":
             init_ollama()
         case "anthropic":
             init_anthropic()
         case "gemini":
             init_gemini()
+        case "mistral":
+            init_mistral()
         case "azure-openai":
             init_azure_openai()
+        case "t-systems":
+            from .llmhub import init_llmhub
+
+            init_llmhub()
         case _:
             raise ValueError(f"Invalid model provider: {model_provider}")
+
     Settings.chunk_size = int(os.getenv("CHUNK_SIZE", "1024"))
     Settings.chunk_overlap = int(os.getenv("CHUNK_OVERLAP", "20"))
 
 
 def init_ollama():
-    from llama_index.llms.ollama import Ollama
     from llama_index.embeddings.ollama import OllamaEmbedding
+    from llama_index.llms.ollama.base import DEFAULT_REQUEST_TIMEOUT, Ollama
 
     base_url = os.getenv("OLLAMA_BASE_URL") or "http://127.0.0.1:11434"
+    request_timeout = float(
+        os.getenv("OLLAMA_REQUEST_TIMEOUT", DEFAULT_REQUEST_TIMEOUT)
+    )
     Settings.embed_model = OllamaEmbedding(
         base_url=base_url,
         model_name=os.getenv("EMBEDDING_MODEL"),
     )
-    Settings.llm = Ollama(base_url=base_url, model=os.getenv("MODEL"))
+    Settings.llm = Ollama(
+        base_url=base_url, model=os.getenv("MODEL"), request_timeout=request_timeout
+    )
 
 
 def init_openai():
-    from llama_index.llms.openai import OpenAI
-    from llama_index.embeddings.openai import OpenAIEmbedding
     from llama_index.core.constants import DEFAULT_TEMPERATURE
+    from llama_index.embeddings.openai import OpenAIEmbedding
+    from llama_index.llms.openai import OpenAI
 
     max_tokens = os.getenv("LLM_MAX_TOKENS")
     config = {
@@ -56,9 +71,9 @@ def init_openai():
 
 
 def init_azure_openai():
-    from llama_index.llms.azure_openai import AzureOpenAI
-    from llama_index.embeddings.azure_openai import AzureOpenAIEmbedding
     from llama_index.core.constants import DEFAULT_TEMPERATURE
+    from llama_index.embeddings.azure_openai import AzureOpenAIEmbedding
+    from llama_index.llms.azure_openai import AzureOpenAI
 
     llm_deployment = os.getenv("AZURE_OPENAI_LLM_DEPLOYMENT")
     embedding_deployment = os.getenv("AZURE_OPENAI_EMBEDDING_DEPLOYMENT")
@@ -83,9 +98,30 @@ def init_azure_openai():
     Settings.embed_model = AzureOpenAIEmbedding(**embedding_config)
 
 
-def init_anthropic():
-    from llama_index.llms.anthropic import Anthropic
+def init_groq():
     from llama_index.embeddings.huggingface import HuggingFaceEmbedding
+    from llama_index.llms.groq import Groq
+
+    model_map: Dict[str, str] = {
+        "llama3-8b": "llama3-8b-8192",
+        "llama3-70b": "llama3-70b-8192",
+        "mixtral-8x7b": "mixtral-8x7b-32768",
+    }
+
+    embed_model_map: Dict[str, str] = {
+        "all-MiniLM-L6-v2": "sentence-transformers/all-MiniLM-L6-v2",
+        "all-mpnet-base-v2": "sentence-transformers/all-mpnet-base-v2",
+    }
+
+    Settings.llm = Groq(model=model_map[os.getenv("MODEL")])
+    Settings.embed_model = HuggingFaceEmbedding(
+        model_name=embed_model_map[os.getenv("EMBEDDING_MODEL")]
+    )
+
+
+def init_anthropic():
+    from llama_index.embeddings.huggingface import HuggingFaceEmbedding
+    from llama_index.llms.anthropic import Anthropic
 
     model_map: Dict[str, str] = {
         "claude-3-opus": "claude-3-opus-20240229",
@@ -107,21 +143,19 @@ def init_anthropic():
 
 
 def init_gemini():
-    from llama_index.llms.gemini import Gemini
     from llama_index.embeddings.gemini import GeminiEmbedding
+    from llama_index.llms.gemini import Gemini
 
-    model_map: Dict[str, str] = {
-        "gemini-1.5-pro-latest": "models/gemini-1.5-pro-latest",
-        "gemini-pro": "models/gemini-pro",
-        "gemini-pro-vision": "models/gemini-pro-vision",
-    }
+    model_name = f"models/{os.getenv('MODEL')}"
+    embed_model_name = f"models/{os.getenv('EMBEDDING_MODEL')}"
 
-    embed_model_map: Dict[str, str] = {
-        "embedding-001": "models/embedding-001",
-        "text-embedding-004": "models/text-embedding-004",
-    }
+    Settings.llm = Gemini(model=model_name)
+    Settings.embed_model = GeminiEmbedding(model_name=embed_model_name)
 
-    Settings.llm = Gemini(model=model_map[os.getenv("MODEL")])
-    Settings.embed_model = GeminiEmbedding(
-        model_name=embed_model_map[os.getenv("EMBEDDING_MODEL")]
-    )
+
+def init_mistral():
+    from llama_index.embeddings.mistralai import MistralAIEmbedding
+    from llama_index.llms.mistralai import MistralAI
+
+    Settings.llm = MistralAI(model=os.getenv("MODEL"))
+    Settings.embed_model = MistralAIEmbedding(model_name=os.getenv("EMBEDDING_MODEL"))

templates/components/ui/html/chat/chat-messages.tsx

@@ -24,6 +24,7 @@ export default function ChatMessages({
   ) => Promise<string | null | undefined>;
 }) {
   const scrollableChatContainerRef = useRef<HTMLDivElement>(null);
+  const lastMessage = messages[messages.length - 1];
 
   const scrollToBottom = () => {
     if (scrollableChatContainerRef.current) {
@@ -34,14 +35,14 @@ export default function ChatMessages({
 
   useEffect(() => {
     scrollToBottom();
-  }, [messages.length]);
+  }, [messages.length, lastMessage]);
 
   return (
-    <div className="w-full max-w-5xl p-4 bg-white rounded-xl shadow-xl">
-      <div
-        className="flex flex-col gap-5 divide-y h-[50vh] overflow-auto"
-        ref={scrollableChatContainerRef}
-      >
+    <div
+      className="flex-1 w-full max-w-5xl p-4 bg-white rounded-xl shadow-xl overflow-auto"
+      ref={scrollableChatContainerRef}
+    >
+      <div className="flex flex-col gap-5 divide-y">
         {messages.map((m: Message) => (
           <ChatItem key={m.id} {...m} />
         ))}

templates/components/vectordbs/python/llamacloud/generate.py

@@ -0,0 +1,45 @@
+from dotenv import load_dotenv
+
+load_dotenv()
+
+import os
+import logging
+from app.settings import init_settings
+from app.engine.loaders import get_documents
+from llama_index.indices.managed.llama_cloud import LlamaCloudIndex
+
+
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger()
+
+
+def generate_datasource():
+    init_settings()
+    logger.info("Generate index for the provided data")
+
+    name = os.getenv("LLAMA_CLOUD_INDEX_NAME")
+    project_name = os.getenv("LLAMA_CLOUD_PROJECT_NAME")
+    api_key = os.getenv("LLAMA_CLOUD_API_KEY")
+    base_url = os.getenv("LLAMA_CLOUD_BASE_URL")
+
+    if name is None or project_name is None or api_key is None:
+        raise ValueError(
+            "Please set LLAMA_CLOUD_INDEX_NAME, LLAMA_CLOUD_PROJECT_NAME and LLAMA_CLOUD_API_KEY"
+            " to your environment variables or config them in .env file"
+        )
+
+    documents = get_documents()
+
+    LlamaCloudIndex.from_documents(
+        documents=documents,
+        name=name,
+        project_name=project_name,
+        api_key=api_key,
+        base_url=base_url,
+    )
+
+    logger.info("Finished generating the index")
+
+
+if __name__ == "__main__":
+    generate_datasource()

templates/components/vectordbs/python/llamacloud/index.py

@@ -0,0 +1,28 @@
+import logging
+import os
+from llama_index.indices.managed.llama_cloud import LlamaCloudIndex
+
+
+logger = logging.getLogger("uvicorn")
+
+
+def get_index():
+    name = os.getenv("LLAMA_CLOUD_INDEX_NAME")
+    project_name = os.getenv("LLAMA_CLOUD_PROJECT_NAME")
+    api_key = os.getenv("LLAMA_CLOUD_API_KEY")
+    base_url = os.getenv("LLAMA_CLOUD_BASE_URL")
+
+    if name is None or project_name is None or api_key is None:
+        raise ValueError(
+            "Please set LLAMA_CLOUD_INDEX_NAME, LLAMA_CLOUD_PROJECT_NAME and LLAMA_CLOUD_API_KEY"
+            " to your environment variables or config them in .env file"
+        )
+
+    index = LlamaCloudIndex(
+        name=name,
+        project_name=project_name,
+        api_key=api_key,
+        base_url=base_url,
+    )
+
+    return index

templates/components/vectordbs/typescript/llamacloud/generate.ts

@@ -0,0 +1,26 @@
+import * as dotenv from "dotenv";
+import { LlamaCloudIndex } from "llamaindex";
+import { getDocuments } from "./loader";
+import { initSettings } from "./settings";
+import { checkRequiredEnvVars } from "./shared";
+
+dotenv.config();
+
+async function loadAndIndex() {
+  const documents = await getDocuments();
+  await LlamaCloudIndex.fromDocuments({
+    documents,
+    name: process.env.LLAMA_CLOUD_INDEX_NAME!,
+    projectName: process.env.LLAMA_CLOUD_PROJECT_NAME!,
+    apiKey: process.env.LLAMA_CLOUD_API_KEY,
+    baseUrl: process.env.LLAMA_CLOUD_BASE_URL,
+  });
+  console.log(`Successfully created embeddings!`);
+}
+
+(async () => {
+  checkRequiredEnvVars();
+  initSettings();
+  await loadAndIndex();
+  console.log("Finished generating storage.");
+})();

templates/components/vectordbs/typescript/llamacloud/index.ts

@@ -0,0 +1,13 @@
+import { LlamaCloudIndex } from "llamaindex/cloud/LlamaCloudIndex";
+import { checkRequiredEnvVars } from "./shared";
+
+export async function getDataSource() {
+  checkRequiredEnvVars();
+  const index = new LlamaCloudIndex({
+    name: process.env.LLAMA_CLOUD_INDEX_NAME!,
+    projectName: process.env.LLAMA_CLOUD_PROJECT_NAME!,
+    apiKey: process.env.LLAMA_CLOUD_API_KEY,
+    baseUrl: process.env.LLAMA_CLOUD_BASE_URL,
+  });
+  return index;
+}

templates/components/vectordbs/typescript/llamacloud/shared.ts

@@ -0,0 +1,22 @@
+const REQUIRED_ENV_VARS = [
+  "LLAMA_CLOUD_INDEX_NAME",
+  "LLAMA_CLOUD_PROJECT_NAME",
+  "LLAMA_CLOUD_API_KEY",
+];
+
+export function checkRequiredEnvVars() {
+  const missingEnvVars = REQUIRED_ENV_VARS.filter((envVar) => {
+    return !process.env[envVar];
+  });
+
+  if (missingEnvVars.length > 0) {
+    console.log(
+      `The following environment variables are required but missing: ${missingEnvVars.join(
+        ", ",
+      )}`,
+    );
+    throw new Error(
+      `Missing environment variables: ${missingEnvVars.join(", ")}`,
+    );
+  }
+}

templates/types/multiagent/fastapi/README-template.md

@@ -0,0 +1,50 @@
+This is a [LlamaIndex](https://www.llamaindex.ai/) project using [FastAPI](https://fastapi.tiangolo.com/) bootstrapped with [`create-llama`](https://github.com/run-llama/LlamaIndexTS/tree/main/packages/create-llama).
+
+## Getting Started
+
+First, setup the environment with poetry:
+
+> **_Note:_** This step is not needed if you are using the dev-container.
+
+```shell
+poetry install
+poetry shell
+```
+
+Then check the parameters that have been pre-configured in the `.env` file in this directory. (E.g. you might need to configure an `OPENAI_API_KEY` if you're using OpenAI as model provider).
+
+Second, generate the embeddings of the documents in the `./data` directory (if this folder exists - otherwise, skip this step):
+
+```shell
+poetry run generate
+```
+
+Third, run all the services in one command:
+
+```shell
+poetry run python main.py
+```
+
+You can monitor and test the agent services with `llama-agents` monitor TUI:
+
+```shell
+poetry run llama-agents monitor --control-plane-url http://127.0.0.1:8001
+```
+
+## Services:
+
+- Message queue (port 8000): To exchange the message between services
+- Control plane (port 8001): A gateway to manage the tasks and services.
+- Human consumer (port 8002): To handle result when the task is completed.
+- Agent service `query_engine` (port 8003): Agent that can query information from the configured LlamaIndex index.
+- Agent service `dummy_agent` (port 8004): A dummy agent that does nothing. Good starting point to add more agents.
+
+The ports listed above are set by default, but you can change them in the `.env` file.
+
+## Learn More
+
+To learn more about LlamaIndex, take a look at the following resources:
+
+- [LlamaIndex Documentation](https://docs.llamaindex.ai) - learn about LlamaIndex.
+
+You can check out [the LlamaIndex GitHub repository](https://github.com/run-llama/llama_index) - your feedback and contributions are welcome!

templates/types/multiagent/fastapi/app/agents/dummy/agent.py

@@ -0,0 +1,33 @@
+from llama_agents import AgentService, SimpleMessageQueue
+from llama_index.core.agent import FunctionCallingAgentWorker
+from llama_index.core.tools import FunctionTool
+from llama_index.core.settings import Settings
+from app.utils import load_from_env
+
+
+DEFAULT_DUMMY_AGENT_DESCRIPTION = "I'm a dummy agent which does nothing."
+
+
+def dummy_function():
+    """
+    This function does nothing.
+    """
+    return ""
+
+
+def init_dummy_agent(message_queue: SimpleMessageQueue) -> AgentService:
+    agent = FunctionCallingAgentWorker(
+        tools=[FunctionTool.from_defaults(fn=dummy_function)],
+        llm=Settings.llm,
+        prefix_messages=[],
+    ).as_agent()
+
+    return AgentService(
+        service_name="dummy_agent",
+        agent=agent,
+        message_queue=message_queue.client,
+        description=load_from_env("AGENT_DUMMY_DESCRIPTION", throw_error=False)
+        or DEFAULT_DUMMY_AGENT_DESCRIPTION,
+        host=load_from_env("AGENT_DUMMY_HOST", throw_error=False) or "127.0.0.1",
+        port=int(load_from_env("AGENT_DUMMY_PORT")),
+    )

templates/types/multiagent/fastapi/app/agents/query_engine/agent.py

@@ -0,0 +1,52 @@
+import os
+from llama_agents import AgentService, SimpleMessageQueue
+from llama_index.core.agent import FunctionCallingAgentWorker
+from llama_index.core.tools import QueryEngineTool, ToolMetadata
+from llama_index.core.settings import Settings
+from app.engine.index import get_index
+from app.utils import load_from_env
+
+
+DEFAULT_QUERY_ENGINE_AGENT_DESCRIPTION = (
+    "Used to answer the questions using the provided context data."
+)
+
+
+def get_query_engine_tool() -> QueryEngineTool:
+    """
+    Provide an agent worker that can be used to query the index.
+    """
+    index = get_index()
+    if index is None:
+        raise ValueError("Index not found. Please create an index first.")
+    query_engine = index.as_query_engine(similarity_top_k=int(os.getenv("TOP_K", 3)))
+    return QueryEngineTool(
+        query_engine=query_engine,
+        metadata=ToolMetadata(
+            name="context_data",
+            description="""
+                Provide the provided context information. 
+                Use a detailed plain text question as input to the tool.
+            """,
+        ),
+    )
+
+
+def init_query_engine_agent(
+    message_queue: SimpleMessageQueue,
+) -> AgentService:
+    """
+    Initialize the agent service.
+    """
+    agent = FunctionCallingAgentWorker(
+        tools=[get_query_engine_tool()], llm=Settings.llm, prefix_messages=[]
+    ).as_agent()
+    return AgentService(
+        service_name="context_query_agent",
+        agent=agent,
+        message_queue=message_queue.client,
+        description=load_from_env("AGENT_QUERY_ENGINE_DESCRIPTION", throw_error=False)
+        or DEFAULT_QUERY_ENGINE_AGENT_DESCRIPTION,
+        host=load_from_env("AGENT_QUERY_ENGINE_HOST", throw_error=False) or "127.0.0.1",
+        port=int(load_from_env("AGENT_QUERY_ENGINE_PORT")),
+    )

templates/types/multiagent/fastapi/app/core/control_plane.py

@@ -0,0 +1,19 @@
+from llama_index.llms.openai import OpenAI
+from llama_agents import AgentOrchestrator, ControlPlaneServer
+from app.core.message_queue import message_queue
+from app.utils import load_from_env
+
+
+control_plane_host = (
+    load_from_env("CONTROL_PLANE_HOST", throw_error=False) or "127.0.0.1"
+)
+control_plane_port = load_from_env("CONTROL_PLANE_PORT", throw_error=False) or "8001"
+
+
+# setup control plane
+control_plane = ControlPlaneServer(
+    message_queue=message_queue,
+    orchestrator=AgentOrchestrator(llm=OpenAI()),
+    host=control_plane_host,
+    port=int(control_plane_port) if control_plane_port else None,
+)

templates/types/multiagent/fastapi/app/core/message_queue.py

@@ -0,0 +1,12 @@
+from llama_agents import SimpleMessageQueue
+from app.utils import load_from_env
+
+message_queue_host = (
+    load_from_env("MESSAGE_QUEUE_HOST", throw_error=False) or "127.0.0.1"
+)
+message_queue_port = load_from_env("MESSAGE_QUEUE_PORT", throw_error=False) or "8000"
+
+message_queue = SimpleMessageQueue(
+    host=message_queue_host,
+    port=int(message_queue_port) if message_queue_port else None,
+)

templates/types/multiagent/fastapi/app/core/task_result.py

@@ -0,0 +1,88 @@
+import json
+from logging import getLogger
+from pathlib import Path
+from fastapi import FastAPI
+from typing import Dict, Optional
+from llama_agents import CallableMessageConsumer, QueueMessage
+from llama_agents.message_queues.base import BaseMessageQueue
+from llama_agents.message_consumers.base import BaseMessageQueueConsumer
+from llama_agents.message_consumers.remote import RemoteMessageConsumer
+from app.utils import load_from_env
+from app.core.message_queue import message_queue
+
+
+logger = getLogger(__name__)
+
+
+class TaskResultService:
+    def __init__(
+        self,
+        message_queue: BaseMessageQueue,
+        name: str = "human",
+        host: str = "127.0.0.1",
+        port: Optional[int] = 8002,
+    ) -> None:
+        self.name = name
+        self.host = host
+        self.port = port
+
+        self._message_queue = message_queue
+
+        # app
+        self._app = FastAPI()
+        self._app.add_api_route(
+            "/", self.home, methods=["GET"], tags=["Human Consumer"]
+        )
+        self._app.add_api_route(
+            "/process_message",
+            self.process_message,
+            methods=["POST"],
+            tags=["Human Consumer"],
+        )
+
+    @property
+    def message_queue(self) -> BaseMessageQueue:
+        return self._message_queue
+
+    def as_consumer(self, remote: bool = False) -> BaseMessageQueueConsumer:
+        if remote:
+            return RemoteMessageConsumer(
+                url=(
+                    f"http://{self.host}:{self.port}/process_message"
+                    if self.port
+                    else f"http://{self.host}/process_message"
+                ),
+                message_type=self.name,
+            )
+
+        return CallableMessageConsumer(
+            message_type=self.name,
+            handler=self.process_message,
+        )
+
+    async def process_message(self, message: QueueMessage) -> None:
+        Path("task_results").mkdir(exist_ok=True)
+        with open("task_results/task_results.json", "+a") as f:
+            json.dump(message.model_dump(), f)
+            f.write("\n")
+
+    async def home(self) -> Dict[str, str]:
+        return {"message": "hello, human."}
+
+    async def register_to_message_queue(self) -> None:
+        """Register to the message queue."""
+        await self.message_queue.register_consumer(self.as_consumer(remote=True))
+
+
+human_consumer_host = (
+    load_from_env("HUMAN_CONSUMER_HOST", throw_error=False) or "127.0.0.1"
+)
+human_consumer_port = load_from_env("HUMAN_CONSUMER_PORT", throw_error=False) or "8002"
+
+
+human_consumer_server = TaskResultService(
+    message_queue=message_queue,
+    host=human_consumer_host,
+    port=int(human_consumer_port) if human_consumer_port else None,
+    name="human",
+)

templates/types/multiagent/fastapi/app/utils.py

@@ -0,0 +1,8 @@
+import os
+
+
+def load_from_env(var: str, throw_error: bool = True) -> str:
+    res = os.getenv(var)
+    if res is None and throw_error:
+        raise ValueError(f"Missing environment variable: {var}")
+    return res

templates/types/multiagent/fastapi/main.py

@@ -0,0 +1,27 @@
+from dotenv import load_dotenv
+from app.settings import init_settings
+
+load_dotenv()
+init_settings()
+
+from llama_agents import ServerLauncher
+from app.core.message_queue import message_queue
+from app.core.control_plane import control_plane
+from app.core.task_result import human_consumer_server
+from app.agents.query_engine.agent import init_query_engine_agent
+from app.agents.dummy.agent import init_dummy_agent
+
+agents = [
+    init_query_engine_agent(message_queue),
+    init_dummy_agent(message_queue),
+]
+
+launcher = ServerLauncher(
+    agents,
+    control_plane,
+    message_queue,
+    additional_consumers=[human_consumer_server.as_consumer()],
+)
+
+if __name__ == "__main__":
+    launcher.launch_servers()

templates/types/multiagent/fastapi/pyproject.toml

@@ -0,0 +1,20 @@
+[tool.poetry]
+name = "app"
+version = "0.1.0"
+description = ""
+authors = ["Marcus Schiesser <mail@marcusschiesser.de>"]
+readme = "README.md"
+
+[tool.poetry.scripts]
+generate = "app.engine.generate:generate_datasource"
+
+[tool.poetry.dependencies]
+python = "^3.11"
+llama-agents = "^0.0.3"
+llama-index-agent-openai = "^0.2.7"
+llama-index-embeddings-openai = "^0.1.10"
+llama-index-llms-openai = "^0.1.23"
+
+[build-system]
+requires = ["poetry-core"]
+build-backend = "poetry.core.masonry.api"

templates/types/streaming/express/eslintrc.json

@@ -3,5 +3,8 @@
   "rules": {
     "max-params": ["error", 4],
     "prefer-const": "error"
+  },
+  "parserOptions": {
+    "sourceType": "module"
   }
 }

templates/types/streaming/express/package.json

@@ -1,23 +1,32 @@
 {
   "name": "llama-index-express-streaming",
   "version": "1.0.0",
-  "main": "dist/index.js",
+  "exports": "./index.js",
+  "types": "./index.d.ts",
+  "type": "module",
+  "engines": {
+    "node": ">=18"
+  },
   "scripts": {
     "format": "prettier --ignore-unknown --cache --check .",
     "format:write": "prettier --ignore-unknown --write .",
-    "build": "tsup index.ts --format cjs --dts",
+    "build": "tsup index.ts --format esm --dts",
     "start": "node dist/index.js",
-    "dev": "concurrently \"tsup index.ts --format cjs --dts --watch\" \"nodemon -q dist/index.js\""
+    "dev": "concurrently \"tsup index.ts --format esm --dts --watch\" \"nodemon --watch dist/index.js\""
   },
   "dependencies": {
     "ai": "^3.0.21",
     "cors": "^2.8.5",
     "dotenv": "^16.3.1",
+    "duck-duck-scrape": "^2.2.5",
     "express": "^4.18.2",
-    "llamaindex": "0.3.16",
+    "llamaindex": "0.4.6",
     "pdf2json": "3.0.5",
     "ajv": "^8.12.0",
-    "@e2b/code-interpreter": "^0.0.5"
+    "@e2b/code-interpreter": "^0.0.5",
+    "got": "^14.4.1",
+    "@apidevtools/swagger-parser": "^10.1.0",
+    "formdata-node": "^6.0.3"
   },
   "devDependencies": {
     "@types/cors": "^2.8.16",

templates/types/streaming/express/src/controllers/engine/settings.ts

@@ -1,9 +1,14 @@
 import {
+  ALL_AVAILABLE_MISTRAL_MODELS,
   Anthropic,
   GEMINI_EMBEDDING_MODEL,
   GEMINI_MODEL,
   Gemini,
   GeminiEmbedding,
+  Groq,
+  MistralAI,
+  MistralAIEmbedding,
+  MistralAIEmbeddingModelType,
   OpenAI,
   OpenAIEmbedding,
   Settings,
@@ -28,12 +33,18 @@ export const initSettings = async () => {
     case "ollama":
       initOllama();
       break;
+    case "groq":
+      initGroq();
+      break;
     case "anthropic":
       initAnthropic();
       break;
     case "gemini":
       initGemini();
       break;
+    case "mistral":
+      initMistralAI();
+      break;
     default:
       initOpenAI();
       break;
@@ -61,7 +72,6 @@ function initOllama() {
   const config = {
     host: process.env.OLLAMA_BASE_URL ?? "http://127.0.0.1:11434",
   };
-
   Settings.llm = new Ollama({
     model: process.env.MODEL ?? "",
     config,
@@ -72,6 +82,27 @@ function initOllama() {
   });
 }
 
+function initGroq() {
+  const embedModelMap: Record<string, string> = {
+    "all-MiniLM-L6-v2": "Xenova/all-MiniLM-L6-v2",
+    "all-mpnet-base-v2": "Xenova/all-mpnet-base-v2",
+  };
+
+  const modelMap: Record<string, string> = {
+    "llama3-8b": "llama3-8b-8192",
+    "llama3-70b": "llama3-70b-8192",
+    "mixtral-8x7b": "mixtral-8x7b-32768",
+  };
+
+  Settings.llm = new Groq({
+    model: modelMap[process.env.MODEL!],
+  });
+
+  Settings.embedModel = new HuggingFaceEmbedding({
+    modelType: embedModelMap[process.env.EMBEDDING_MODEL!],
+  });
+}
+
 function initAnthropic() {
   const embedModelMap: Record<string, string> = {
     "all-MiniLM-L6-v2": "Xenova/all-MiniLM-L6-v2",
@@ -93,3 +124,12 @@ function initGemini() {
     model: process.env.EMBEDDING_MODEL as GEMINI_EMBEDDING_MODEL,
   });
 }
+
+function initMistralAI() {
+  Settings.llm = new MistralAI({
+    model: process.env.MODEL as keyof typeof ALL_AVAILABLE_MISTRAL_MODELS,
+  });
+  Settings.embedModel = new MistralAIEmbedding({
+    model: process.env.EMBEDDING_MODEL as MistralAIEmbeddingModelType,
+  });
+}

templates/types/streaming/express/src/controllers/llamaindex-stream.ts

@@ -7,20 +7,12 @@ import {
   type AIStreamCallbacksAndOptions,
 } from "ai";
 import {
+  EngineResponse,
   MessageContent,
   MessageContentDetail,
-  Metadata,
-  NodeWithScore,
-  Response,
-  ToolCallLLMMessageOptions,
 } from "llamaindex";
 
-import { AgentStreamChatResponse } from "llamaindex/agent/base";
-import { CsvFile, appendSourceData } from "./stream-helper";
-
-type LlamaIndexResponse =
-  | AgentStreamChatResponse<ToolCallLLMMessageOptions>
-  | Response;
+import { CsvFile } from "./stream-helper";
 
 export const convertMessageContent = (
   content: string,
@@ -86,39 +78,19 @@ const convertAnnotations = (
   return content;
 };
 
-function createParser(
-  res: AsyncIterable<LlamaIndexResponse>,
-  data: StreamData,
-) {
+function createParser(res: AsyncIterable<EngineResponse>, data: StreamData) {
   const it = res[Symbol.asyncIterator]();
   const trimStartOfStream = trimStartOfStreamHelper();
 
-  let sourceNodes: NodeWithScore<Metadata>[] | undefined;
   return new ReadableStream<string>({
     async pull(controller): Promise<void> {
       const { value, done } = await it.next();
       if (done) {
-        if (sourceNodes) {
-          appendSourceData(data, sourceNodes);
-        }
         controller.close();
         data.close();
         return;
       }
-
-      let delta;
-      if (value instanceof Response) {
-        // handle Response type
-        if (value.sourceNodes) {
-          // get source nodes from the first response
-          sourceNodes = value.sourceNodes;
-        }
-        delta = value.response ?? "";
-      } else {
-        // handle other types
-        delta = value.response.delta;
-      }
-      const text = trimStartOfStream(delta ?? "");
+      const text = trimStartOfStream(value.delta ?? "");
       if (text) {
         controller.enqueue(text);
       }
@@ -127,7 +99,7 @@ function createParser(
 }
 
 export function LlamaIndexStream(
-  response: AsyncIterable<LlamaIndexResponse>,
+  response: AsyncIterable<EngineResponse>,
   data: StreamData,
   opts?: {
     callbacks?: AIStreamCallbacksAndOptions;

templates/types/streaming/express/src/controllers/stream-helper.ts

@@ -84,8 +84,9 @@ export function createStreamTimeout(stream: StreamData) {
 export function createCallbackManager(stream: StreamData) {
   const callbackManager = new CallbackManager();
 
-  callbackManager.on("retrieve", (data) => {
-    const { nodes, query } = data.detail;
+  callbackManager.on("retrieve-end", (data) => {
+    const { nodes, query } = data.detail.payload;
+    appendSourceData(stream, nodes);
     appendEventData(stream, `Retrieving context for query: '${query}'`);
     appendEventData(
       stream,

templates/types/streaming/fastapi/app/llmhub.py

@@ -0,0 +1,61 @@
+from llama_index.embeddings.openai import OpenAIEmbedding
+from llama_index.core.settings import Settings
+from typing import Dict
+import os
+
+DEFAULT_MODEL = "gpt-3.5-turbo"
+DEFAULT_EMBEDDING_MODEL = "text-embedding-3-large"
+
+class TSIEmbedding(OpenAIEmbedding):
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        self._query_engine = self._text_engine = self.model_name
+
+def llm_config_from_env() -> Dict:
+    from llama_index.core.constants import DEFAULT_TEMPERATURE
+
+    model = os.getenv("MODEL", DEFAULT_MODEL)
+    temperature = os.getenv("LLM_TEMPERATURE", DEFAULT_TEMPERATURE)
+    max_tokens = os.getenv("LLM_MAX_TOKENS")
+    api_key = os.getenv("T_SYSTEMS_LLMHUB_API_KEY")
+    api_base = os.getenv("T_SYSTEMS_LLMHUB_BASE_URL")
+
+    config = {
+        "model": model,
+        "api_key": api_key,
+        "api_base": api_base,
+        "temperature": float(temperature),
+        "max_tokens": int(max_tokens) if max_tokens is not None else None,
+    }
+    return config
+
+
+def embedding_config_from_env() -> Dict:
+    from llama_index.core.constants import DEFAULT_EMBEDDING_DIM
+    
+    model = os.getenv("EMBEDDING_MODEL", DEFAULT_EMBEDDING_MODEL)
+    dimension = os.getenv("EMBEDDING_DIM", DEFAULT_EMBEDDING_DIM)
+    api_key = os.getenv("T_SYSTEMS_LLMHUB_API_KEY")
+    api_base = os.getenv("T_SYSTEMS_LLMHUB_BASE_URL")
+
+    config = {
+        "model_name": model,
+        "dimension": int(dimension) if dimension is not None else None,
+        "api_key": api_key,
+        "api_base": api_base,
+    }
+    return config
+
+def init_llmhub():
+    from llama_index.llms.openai_like import OpenAILike
+
+    llm_configs = llm_config_from_env()
+    embedding_configs = embedding_config_from_env()
+
+    Settings.embed_model = TSIEmbedding(**embedding_configs)
+    Settings.llm = OpenAILike(
+        **llm_configs,
+        is_chat_model=True,
+        is_function_calling_model=False,
+        context_window=4096,
+    )

templates/types/streaming/fastapi/pyproject.toml

@@ -14,8 +14,7 @@ fastapi = "^0.109.1"
 uvicorn = { extras = ["standard"], version = "^0.23.2" }
 python-dotenv = "^1.0.0"
 aiostream = "^0.5.2"
-llama-index = "0.10.41"
-llama-index-core = "0.10.41"
+llama-index = "0.10.53"
 cachetools = "^5.3.3"
 
 [build-system]

templates/types/streaming/nextjs/app/api/chat/engine/settings.ts

@@ -1,9 +1,14 @@
 import {
+  ALL_AVAILABLE_MISTRAL_MODELS,
   Anthropic,
   GEMINI_EMBEDDING_MODEL,
   GEMINI_MODEL,
   Gemini,
   GeminiEmbedding,
+  Groq,
+  MistralAI,
+  MistralAIEmbedding,
+  MistralAIEmbeddingModelType,
   OpenAI,
   OpenAIEmbedding,
   Settings,
@@ -28,12 +33,18 @@ export const initSettings = async () => {
     case "ollama":
       initOllama();
       break;
+    case "groq":
+      initGroq();
+      break;
     case "anthropic":
       initAnthropic();
       break;
     case "gemini":
       initGemini();
       break;
+    case "mistral":
+      initMistralAI();
+      break;
     default:
       initOpenAI();
       break;
@@ -71,6 +82,27 @@ function initOllama() {
   });
 }
 
+function initGroq() {
+  const embedModelMap: Record<string, string> = {
+    "all-MiniLM-L6-v2": "Xenova/all-MiniLM-L6-v2",
+    "all-mpnet-base-v2": "Xenova/all-mpnet-base-v2",
+  };
+
+  const modelMap: Record<string, string> = {
+    "llama3-8b": "llama3-8b-8192",
+    "llama3-70b": "llama3-70b-8192",
+    "mixtral-8x7b": "mixtral-8x7b-32768",
+  };
+
+  Settings.llm = new Groq({
+    model: modelMap[process.env.MODEL!],
+  });
+
+  Settings.embedModel = new HuggingFaceEmbedding({
+    modelType: embedModelMap[process.env.EMBEDDING_MODEL!],
+  });
+}
+
 function initAnthropic() {
   const embedModelMap: Record<string, string> = {
     "all-MiniLM-L6-v2": "Xenova/all-MiniLM-L6-v2",
@@ -92,3 +124,12 @@ function initGemini() {
     model: process.env.EMBEDDING_MODEL as GEMINI_EMBEDDING_MODEL,
   });
 }
+
+function initMistralAI() {
+  Settings.llm = new MistralAI({
+    model: process.env.MODEL as keyof typeof ALL_AVAILABLE_MISTRAL_MODELS,
+  });
+  Settings.embedModel = new MistralAIEmbedding({
+    model: process.env.EMBEDDING_MODEL as MistralAIEmbeddingModelType,
+  });
+}

templates/types/streaming/nextjs/app/api/chat/llamaindex-stream.ts

@@ -7,20 +7,12 @@ import {
   type AIStreamCallbacksAndOptions,
 } from "ai";
 import {
+  EngineResponse,
   MessageContent,
   MessageContentDetail,
-  Metadata,
-  NodeWithScore,
-  Response,
-  ToolCallLLMMessageOptions,
 } from "llamaindex";
 
-import { AgentStreamChatResponse } from "llamaindex/agent/base";
-import { CsvFile, appendSourceData } from "./stream-helper";
-
-type LlamaIndexResponse =
-  | AgentStreamChatResponse<ToolCallLLMMessageOptions>
-  | Response;
+import { CsvFile } from "./stream-helper";
 
 export const convertMessageContent = (
   content: string,
@@ -86,39 +78,19 @@ const convertAnnotations = (
   return content;
 };
 
-function createParser(
-  res: AsyncIterable<LlamaIndexResponse>,
-  data: StreamData,
-) {
+function createParser(res: AsyncIterable<EngineResponse>, data: StreamData) {
   const it = res[Symbol.asyncIterator]();
   const trimStartOfStream = trimStartOfStreamHelper();
 
-  let sourceNodes: NodeWithScore<Metadata>[] | undefined;
   return new ReadableStream<string>({
     async pull(controller): Promise<void> {
       const { value, done } = await it.next();
       if (done) {
-        if (sourceNodes) {
-          appendSourceData(data, sourceNodes);
-        }
         controller.close();
         data.close();
         return;
       }
-
-      let delta;
-      if (value instanceof Response) {
-        // handle Response type
-        if (value.sourceNodes) {
-          // get source nodes from the first response
-          sourceNodes = value.sourceNodes;
-        }
-        delta = value.response ?? "";
-      } else {
-        // handle other types
-        delta = value.response.delta;
-      }
-      const text = trimStartOfStream(delta ?? "");
+      const text = trimStartOfStream(value.delta ?? "");
       if (text) {
         controller.enqueue(text);
       }
@@ -127,7 +99,7 @@ function createParser(
 }
 
 export function LlamaIndexStream(
-  response: AsyncIterable<LlamaIndexResponse>,
+  response: AsyncIterable<EngineResponse>,
   data: StreamData,
   opts?: {
     callbacks?: AIStreamCallbacksAndOptions;

templates/types/streaming/nextjs/app/api/chat/stream-helper.ts

@@ -84,8 +84,9 @@ export function createStreamTimeout(stream: StreamData) {
 export function createCallbackManager(stream: StreamData) {
   const callbackManager = new CallbackManager();
 
-  callbackManager.on("retrieve", (data) => {
-    const { nodes, query } = data.detail;
+  callbackManager.on("retrieve-end", (data) => {
+    const { nodes, query } = data.detail.payload;
+    appendSourceData(stream, nodes);
     appendEventData(stream, `Retrieving context for query: '${query}'`);
     appendEventData(
       stream,

templates/types/streaming/nextjs/app/components/chat-section.tsx

@@ -2,7 +2,7 @@
 
 import { useChat } from "ai/react";
 import { ChatInput, ChatMessages } from "./ui/chat";
-import { useClientConfig } from "./ui/chat/use-config";
+import { useClientConfig } from "./ui/chat/hooks/use-config";
 
 export default function ChatSection() {
   const { chatAPI } = useClientConfig();
@@ -29,7 +29,7 @@ export default function ChatSection() {
   });
 
   return (
-    <div className="space-y-4 max-w-5xl w-full">
+    <div className="space-y-4 w-full h-full flex flex-col">
       <ChatMessages
         messages={messages}
         isLoading={isLoading}

templates/types/streaming/nextjs/app/components/ui/chat/chat-input.tsx

@@ -8,7 +8,7 @@ import { Input } from "../input";
 import UploadCsvPreview from "../upload-csv-preview";
 import UploadImagePreview from "../upload-image-preview";
 import { ChatHandler } from "./chat.interface";
-import { useCsv } from "./use-csv";
+import { useCsv } from "./hooks/use-csv";
 
 export default function ChatInput(
   props: Pick<
@@ -81,23 +81,27 @@ export default function ChatInput(
 
   const onRemovePreviewImage = () => setImageUrl(null);
 
-  const handleUploadImageFile = async (file: File) => {
-    const base64 = await new Promise<string>((resolve, reject) => {
+  const readContent = async (file: File): Promise<string> => {
+    const content = await new Promise<string>((resolve, reject) => {
       const reader = new FileReader();
-      reader.readAsDataURL(file);
+      if (file.type.startsWith("image/")) {
+        reader.readAsDataURL(file);
+      } else {
+        reader.readAsText(file);
+      }
       reader.onload = () => resolve(reader.result as string);
       reader.onerror = (error) => reject(error);
     });
+    return content;
+  };
+
+  const handleUploadImageFile = async (file: File) => {
+    const base64 = await readContent(file);
     setImageUrl(base64);
   };
 
   const handleUploadCsvFile = async (file: File) => {
-    const content = await new Promise<string>((resolve, reject) => {
-      const reader = new FileReader();
-      reader.readAsText(file);
-      reader.onload = () => resolve(reader.result as string);
-      reader.onerror = (error) => reject(error);
-    });
+    const content = await readContent(file);
     const isSuccess = upload({
       id: uuidv4(),
       content,
@@ -130,7 +134,7 @@ export default function ChatInput(
   return (
     <form
       onSubmit={onSubmit}
-      className="rounded-xl bg-white p-4 shadow-xl space-y-4"
+      className="rounded-xl bg-white p-4 shadow-xl space-y-4 shrink-0"
     >
       {imageUrl && (
         <UploadImagePreview url={imageUrl} onRemove={onRemovePreviewImage} />

templates/types/streaming/nextjs/app/components/ui/chat/chat-message/chat-events.tsx

@@ -1,12 +1,12 @@
 import { ChevronDown, ChevronRight, Loader2 } from "lucide-react";
 import { useState } from "react";
-import { Button } from "../button";
+import { Button } from "../../button";
 import {
   Collapsible,
   CollapsibleContent,
   CollapsibleTrigger,
-} from "../collapsible";
-import { EventData } from "./index";
+} from "../../collapsible";
+import { EventData } from "../index";
 
 export function ChatEvents({
   data,

templates/types/streaming/nextjs/app/components/ui/chat/chat-message/chat-image.tsx

@@ -1,5 +1,5 @@
 import Image from "next/image";
-import { type ImageData } from "./index";
+import { type ImageData } from "../index";
 
 export function ChatImage({ data }: { data: ImageData }) {
   return (

templates/types/streaming/nextjs/app/components/ui/chat/chat-message/chat-sources.tsx

@@ -1,10 +1,14 @@
 import { Check, Copy } from "lucide-react";
 import { useMemo } from "react";
-import { Button } from "../button";
-import { HoverCard, HoverCardContent, HoverCardTrigger } from "../hover-card";
-import { SourceData } from "./index";
-import { useCopyToClipboard } from "./use-copy-to-clipboard";
-import PdfDialog from "./widgets/PdfDialog";
+import { Button } from "../../button";
+import {
+  HoverCard,
+  HoverCardContent,
+  HoverCardTrigger,
+} from "../../hover-card";
+import { useCopyToClipboard } from "../hooks/use-copy-to-clipboard";
+import { SourceData } from "../index";
+import PdfDialog from "../widgets/PdfDialog";
 
 const SCORE_THRESHOLD = 0.3;
 

templates/types/streaming/nextjs/app/components/ui/chat/chat-message/chat-tools.tsx

@@ -1,5 +1,5 @@
-import { ToolData } from "./index";
-import { WeatherCard, WeatherData } from "./widgets/WeatherCard";
+import { ToolData } from "../index";
+import { WeatherCard, WeatherData } from "../widgets/WeatherCard";
 
 // TODO: If needed, add displaying more tool outputs here
 export default function ChatTools({ data }: { data: ToolData }) {

templates/types/streaming/nextjs/app/components/ui/chat/chat-message/codeblock.tsx

@@ -5,8 +5,8 @@ import { FC, memo } from "react";
 import { Prism, SyntaxHighlighterProps } from "react-syntax-highlighter";
 import { coldarkDark } from "react-syntax-highlighter/dist/cjs/styles/prism";
 
-import { Button } from "../button";
-import { useCopyToClipboard } from "./use-copy-to-clipboard";
+import { Button } from "../../button";
+import { useCopyToClipboard } from "../hooks/use-copy-to-clipboard";
 
 // TODO: Remove this when @type/react-syntax-highlighter is updated
 const SyntaxHighlighter = Prism as unknown as FC<SyntaxHighlighterProps>;

templates/types/streaming/nextjs/app/components/ui/chat/chat-message/csv-content.tsx

@@ -1,5 +1,5 @@
-import { CsvData } from ".";
-import UploadCsvPreview from "../upload-csv-preview";
+import UploadCsvPreview from "../../upload-csv-preview";
+import { CsvData } from "../index";
 
 export default function CsvContent({ data }: { data: CsvData }) {
   if (!data.csvFiles.length) return null;

templates/types/streaming/nextjs/app/components/ui/chat/chat-message/index.tsx

@@ -2,13 +2,8 @@ import { Check, Copy } from "lucide-react";
 
 import { Message } from "ai";
 import { Fragment } from "react";
-import { Button } from "../button";
-import ChatAvatar from "./chat-avatar";
-import { ChatEvents } from "./chat-events";
-import { ChatImage } from "./chat-image";
-import { ChatSources } from "./chat-sources";
-import ChatTools from "./chat-tools";
-import CsvContent from "./csv-content";
+import { Button } from "../../button";
+import { useCopyToClipboard } from "../hooks/use-copy-to-clipboard";
 import {
   CsvData,
   EventData,
@@ -18,9 +13,14 @@ import {
   SourceData,
   ToolData,
   getAnnotationData,
-} from "./index";
+} from "../index";
+import ChatAvatar from "./chat-avatar";
+import { ChatEvents } from "./chat-events";
+import { ChatImage } from "./chat-image";
+import { ChatSources } from "./chat-sources";
+import ChatTools from "./chat-tools";
+import CsvContent from "./csv-content";
 import Markdown from "./markdown";
-import { useCopyToClipboard } from "./use-copy-to-clipboard";
 
 type ContentDisplayConfig = {
   order: number;

templates/types/streaming/nextjs/app/components/ui/chat/chat-message/markdown.tsx

@@ -17,19 +17,30 @@ const MemoizedReactMarkdown: FC<Options> = memo(
 const preprocessLaTeX = (content: string) => {
   // Replace block-level LaTeX delimiters \[ \] with $$ $$
   const blockProcessedContent = content.replace(
-    /\\\[(.*?)\\\]/gs,
+    /\\\[([\s\S]*?)\\\]/g,
     (_, equation) => `$$${equation}$$`,
   );
   // Replace inline LaTeX delimiters \( \) with $ $
   const inlineProcessedContent = blockProcessedContent.replace(
-    /\\\((.*?)\\\)/gs,
+    /\\\[([\s\S]*?)\\\]/g,
     (_, equation) => `$${equation}$`,
   );
   return inlineProcessedContent;
 };
 
+const preprocessMedia = (content: string) => {
+  // Remove `sandbox:` from the beginning of the URL
+  // to fix OpenAI's models issue appending `sandbox:` to the relative URL
+  return content.replace(/(sandbox|attachment|snt):/g, "");
+};
+
+const preprocessContent = (content: string) => {
+  return preprocessMedia(preprocessLaTeX(content));
+};
+
 export default function Markdown({ content }: { content: string }) {
-  const processedContent = preprocessLaTeX(content);
+  const processedContent = preprocessContent(content);
+
   return (
     <MemoizedReactMarkdown
       className="prose dark:prose-invert prose-p:leading-relaxed prose-pre:p-0 break-words custom-markdown"

templates/types/streaming/nextjs/app/components/ui/chat/chat-messages.tsx

@@ -5,7 +5,7 @@ import { Button } from "../button";
 import ChatActions from "./chat-actions";
 import ChatMessage from "./chat-message";
 import { ChatHandler } from "./chat.interface";
-import { useClientConfig } from "./use-config";
+import { useClientConfig } from "./hooks/use-config";
 
 export default function ChatMessages(
   props: Pick<
@@ -41,11 +41,11 @@ export default function ChatMessages(
   }, [messageLength, lastMessage]);
 
   return (
-    <div className="w-full rounded-xl bg-white p-4 shadow-xl pb-0 relative">
-      <div
-        className="flex h-[50vh] flex-col gap-5 divide-y overflow-y-auto pb-4"
-        ref={scrollableChatContainerRef}
-      >
+    <div
+      className="flex-1 w-full rounded-xl bg-white p-4 shadow-xl relative overflow-y-auto"
+      ref={scrollableChatContainerRef}
+    >
+      <div className="flex flex-col gap-5 divide-y">
         {props.messages.map((m, i) => {
           const isLoadingMessage = i === messageLength - 1 && props.isLoading;
           return (
@@ -62,14 +62,16 @@ export default function ChatMessages(
           </div>
         )}
       </div>
-      <div className="flex justify-end py-4">
-        <ChatActions
-          reload={props.reload}
-          stop={props.stop}
-          showReload={showReload}
-          showStop={showStop}
-        />
-      </div>
+      {(showReload || showStop) && (
+        <div className="flex justify-end py-4">
+          <ChatActions
+            reload={props.reload}
+            stop={props.stop}
+            showReload={showReload}
+            showStop={showStop}
+          />
+        </div>
+      )}
       {!messageLength && starterQuestions?.length && props.append && (
         <div className="absolute bottom-6 left-0 w-full">
           <div className="grid grid-cols-2 gap-2 mx-20">

templates/types/streaming/nextjs/app/components/ui/chat/hooks/use-csv.ts

@@ -1,7 +1,7 @@
 "use client";
 
 import { useState } from "react";
-import { CsvFile } from ".";
+import { CsvFile } from "../index";
 
 export function useCsv() {
   const [files, setFiles] = useState<CsvFile[]>([]);

templates/types/streaming/nextjs/app/globals.css

@@ -74,8 +74,11 @@
   * {
     @apply border-border;
   }
+  html {
+    @apply h-full;
+  }
   body {
-    @apply bg-background text-foreground;
+    @apply bg-background text-foreground h-full;
     font-feature-settings:
       "rlig" 1,
       "calt" 1;

templates/types/streaming/nextjs/app/page.tsx

@@ -3,9 +3,13 @@ import ChatSection from "./components/chat-section";
 
 export default function Home() {
   return (
-    <main className="flex min-h-screen flex-col items-center gap-10 p-24 background-gradient">
-      <Header />
-      <ChatSection />
+    <main className="h-full w-full flex justify-center items-center background-gradient">
+      <div className="space-y-2 lg:space-y-10 w-[90%] lg:w-[60rem]">
+        <Header />
+        <div className="h-[65vh] flex">
+          <ChatSection />
+        </div>
+      </div>
     </main>
   );
 }

templates/types/streaming/nextjs/package.json

@@ -10,6 +10,9 @@
     "lint": "next lint"
   },
   "dependencies": {
+    "@apidevtools/swagger-parser": "^10.1.0",
+    "@e2b/code-interpreter": "^0.0.5",
+    "@llamaindex/pdf-viewer": "^1.1.1",
     "@radix-ui/react-collapsible": "^1.0.3",
     "@radix-ui/react-hover-card": "^1.0.7",
     "@radix-ui/react-slot": "^1.0.2",
@@ -18,42 +21,44 @@
     "class-variance-authority": "^0.7.0",
     "clsx": "^2.1.1",
     "dotenv": "^16.3.1",
-    "llamaindex": "0.3.16",
+    "duck-duck-scrape": "^2.2.5",
+    "formdata-node": "^6.0.3",
+    "got": "^14.4.1",
+    "llamaindex": "0.4.6",
     "lucide-react": "^0.294.0",
-    "next": "^14.0.3",
+    "next": "^14.2.4",
     "pdf2json": "3.0.5",
     "react": "^18.2.0",
     "react-dom": "^18.2.0",
     "react-markdown": "^8.0.7",
     "react-syntax-highlighter": "^15.5.0",
+    "rehype-katex": "^7.0.0",
     "remark": "^14.0.3",
     "remark-code-import": "^1.2.0",
     "remark-gfm": "^3.0.1",
     "remark-math": "^5.1.1",
-    "rehype-katex": "^7.0.0",
     "supports-color": "^8.1.1",
     "tailwind-merge": "^2.1.0",
-    "vaul": "^0.9.1",
-    "@llamaindex/pdf-viewer": "^1.1.1",
-    "@e2b/code-interpreter": "^0.0.5",
-    "uuid": "^9.0.1"
+    "tiktoken": "^1.0.15",
+    "uuid": "^9.0.1",
+    "vaul": "^0.9.1"
   },
   "devDependencies": {
     "@types/node": "^20.10.3",
     "@types/react": "^18.2.42",
     "@types/react-dom": "^18.2.17",
     "@types/react-syntax-highlighter": "^15.5.11",
+    "@types/uuid": "^9.0.8",
     "autoprefixer": "^10.4.16",
     "cross-env": "^7.0.3",
     "eslint": "^8.55.0",
-    "eslint-config-next": "^14.0.3",
+    "eslint-config-next": "^14.2.4",
     "eslint-config-prettier": "^8.10.0",
     "postcss": "^8.4.32",
     "prettier": "^3.2.5",
     "prettier-plugin-organize-imports": "^3.2.4",
     "tailwindcss": "^3.3.6",
     "tsx": "^4.7.2",
-    "typescript": "^5.3.2",
-    "@types/uuid": "^9.0.8"
+    "typescript": "^5.3.2"
   }
 }