feat: support multi agent for ts

.changeset/yellow-jokes-protect.md

@@ -0,0 +1,5 @@
+---
+"create-llama": patch
+---
+
+Add multi agents template for Express

helpers/typescript.ts

@@ -33,7 +33,11 @@ export const installTSTemplate = async ({
    * Copy the template files to the target directory.
    */
   console.log("\nInitializing project with template:", template, "\n");
-  const type = template === "multiagent" ? "streaming" : template; // use nextjs streaming template for multiagent
+  let type = "streaming";
+  if (template === "multiagent" && framework === "express") {
+    // use nextjs streaming template as frontend for express and fastapi
+    type = "multiagent";
+  }
   const templatePath = path.join(templatesDir, "types", type, framework);
   const copySource = ["**"];
 
@@ -145,6 +149,15 @@ export const installTSTemplate = async ({
     cwd: path.join(compPath, "engines", "typescript", engine),
   });
 
+  // copy settings file to engine folder
+  const settingPath = path.join(
+    compPath,
+    "engines",
+    "typescript",
+    "settings.ts",
+  );
+  await copy("settings.ts", enginePath, { cwd: settingPath });
+
   /**
    * Copy the selected UI files to the target directory and reference it.
    */

questions.ts

@@ -410,10 +410,7 @@ 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";
-  } else if (program.template === "extractor") {
+  if (program.template === "extractor") {
     // Extractor template only supports FastAPI, empty data sources, and llamacloud
     // So we just use example file for extractor template, this allows user to choose vector database later
     program.dataSources = [EXAMPLE_FILE];
@@ -424,7 +421,9 @@ export const askQuestions = async (
       program.framework = getPrefOrDefault("framework");
     } else {
       const choices = [
-        { title: "NextJS", value: "nextjs" },
+        ...(program.template === "multiagent"
+          ? []
+          : [{ title: "NextJS", value: "nextjs" }]), // Not supported nextjs for multiagent for now
         { title: "Express", value: "express" },
         { title: "FastAPI (Python)", value: "fastapi" },
       ];

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

@@ -0,0 +1,103 @@
+This is a [LlamaIndex](https://www.llamaindex.ai/) project using [Express](https://expressjs.com/) bootstrapped with [`create-llama`](https://github.com/run-llama/LlamaIndexTS/tree/main/packages/create-llama).
+
+## Getting Started
+
+First, install the dependencies:
+
+```
+npm install
+```
+
+Second, generate the embeddings of the documents in the `./data` directory (if this folder exists - otherwise, skip this step):
+
+```
+npm run generate
+```
+
+Third, run the development server:
+
+```
+npm run dev
+```
+
+The example provides two different API endpoints:
+
+1. `/api/chat` - a streaming chat endpoint (found in `src/controllers/chat.controller.ts`)
+2. `/api/chat/request` - a non-streaming chat endpoint (found in `src/controllers/chat-request.controller.ts`)
+
+You can test the streaming endpoint with the following curl request:
+
+```
+curl --location 'localhost:8000/api/chat' \
+--header 'Content-Type: application/json' \
+--data '{ "messages": [{ "role": "user", "content": "Hello" }] }'
+```
+
+And for the non-streaming endpoint run:
+
+```
+curl --location 'localhost:8000/api/chat/request' \
+--header 'Content-Type: application/json' \
+--data '{ "messages": [{ "role": "user", "content": "Hello" }] }'
+```
+
+You can start editing the API by modifying `src/controllers/chat.controller.ts` or `src/controllers/chat-request.controller.ts`. The endpoint auto-updates as you save the file.
+You can delete the endpoint that you're not using.
+
+## Production
+
+First, build the project:
+
+```
+npm run build
+```
+
+You can then run the production server:
+
+```
+NODE_ENV=production npm run start
+```
+
+> Note that the `NODE_ENV` environment variable is set to `production`. This disables CORS for all origins.
+
+## Using Docker
+
+1. Build an image for the Express API:
+
+```
+docker build -t <your_backend_image_name> .
+```
+
+2. Generate embeddings:
+
+Parse the data and generate the vector embeddings if the `./data` folder exists - otherwise, skip this step:
+
+```
+docker run --rm \
+  -v $(pwd)/.env:/app/.env \ # Use ENV variables and configuration from your file-system
+  -v $(pwd)/config:/app/config \
+  -v $(pwd)/data:/app/data \
+  -v $(pwd)/cache:/app/cache \ # Use your file system to store the vector database
+  <your_backend_image_name>
+  npm run generate
+```
+
+3. Start the API:
+
+```
+docker run \
+  -v $(pwd)/.env:/app/.env \ # Use ENV variables and configuration from your file-system
+  -v $(pwd)/config:/app/config \
+  -v $(pwd)/cache:/app/cache \ # Use your file system to store the vector database
+  -p 8000:8000 \
+  <your_backend_image_name>
+```
+
+## Learn More
+
+To learn more about LlamaIndex, take a look at the following resources:
+
+- [LlamaIndex Documentation](https://docs.llamaindex.ai) - learn about LlamaIndex (Python features).
+- [LlamaIndexTS Documentation](https://ts.llamaindex.ai) - learn about LlamaIndex (Typescript features).
+
+You can check out [the LlamaIndexTS GitHub repository](https://github.com/run-llama/LlamaIndexTS) - your feedback and contributions are welcome!

templates/types/multiagent/express/eslintrc.json

@@ -0,0 +1,10 @@
+{
+  "extends": ["eslint:recommended", "prettier"],
+  "rules": {
+    "max-params": ["error", 4],
+    "prefer-const": "error"
+  },
+  "parserOptions": {
+    "sourceType": "module"
+  }
+}

templates/types/multiagent/express/gitignore

@@ -0,0 +1,5 @@
+# local env files
+.env
+node_modules/
+
+output/

templates/types/multiagent/express/index.ts

@@ -0,0 +1,46 @@
+/* eslint-disable turbo/no-undeclared-env-vars */
+import cors from "cors";
+import "dotenv/config";
+import express, { Express, Request, Response } from "express";
+import { initObservability } from "./src/observability";
+import chatRouter from "./src/routes/chat.route";
+
+const app: Express = express();
+const port = parseInt(process.env.PORT || "8000");
+
+const env = process.env["NODE_ENV"];
+const isDevelopment = !env || env === "development";
+const prodCorsOrigin = process.env["PROD_CORS_ORIGIN"];
+
+initObservability();
+
+app.use(express.json({ limit: "50mb" }));
+
+if (isDevelopment) {
+  console.warn("Running in development mode - allowing CORS for all origins");
+  app.use(cors());
+} else if (prodCorsOrigin) {
+  console.log(
+    `Running in production mode - allowing CORS for domain: ${prodCorsOrigin}`,
+  );
+  const corsOptions = {
+    origin: prodCorsOrigin, // Restrict to production domain
+  };
+  app.use(cors(corsOptions));
+} else {
+  console.warn("Production CORS origin not set, defaulting to no CORS.");
+}
+
+app.use("/api/files/data", express.static("data"));
+app.use("/api/files/output", express.static("output"));
+app.use(express.text());
+
+app.get("/", (req: Request, res: Response) => {
+  res.send("LlamaIndex Express Server");
+});
+
+app.use("/api/chat", chatRouter);
+
+app.listen(port, () => {
+  console.log(`⚡️[server]: Server is running at http://localhost:${port}`);
+});

templates/types/multiagent/express/npmrc

@@ -0,0 +1 @@
+node-linker=hoisted

templates/types/multiagent/express/package.json

@@ -0,0 +1,45 @@
+{
+  "name": "llama-index-express-multiagent",
+  "version": "1.0.0",
+  "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 esm --dts",
+    "start": "node dist/index.js",
+    "dev": "concurrently \"tsup index.ts --format esm --dts --watch\" \"nodemon --watch dist/index.js\""
+  },
+  "dependencies": {
+    "ai": "3.3.38",
+    "cors": "^2.8.5",
+    "dotenv": "^16.3.1",
+    "duck-duck-scrape": "^2.2.5",
+    "express": "^4.18.2",
+    "llamaindex": "0.6.2",
+    "pdf2json": "3.0.5",
+    "ajv": "^8.12.0",
+    "@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",
+    "@types/express": "^4.17.21",
+    "@types/node": "^20.9.5",
+    "concurrently": "^8.2.2",
+    "eslint": "^8.54.0",
+    "eslint-config-prettier": "^8.10.0",
+    "nodemon": "^3.0.1",
+    "prettier": "^3.2.5",
+    "prettier-plugin-organize-imports": "^3.2.4",
+    "tsx": "^4.7.2",
+    "tsup": "8.1.0",
+    "typescript": "^5.3.2"
+  }
+}

templates/types/multiagent/express/prettier.config.cjs

@@ -0,0 +1,3 @@
+module.exports = {
+  plugins: ["prettier-plugin-organize-imports"],
+};

templates/types/multiagent/express/src/controllers/chat-config.controller.ts

@@ -0,0 +1,31 @@
+import { Request, Response } from "express";
+import { LLamaCloudFileService } from "llamaindex";
+
+export const chatConfig = async (_req: Request, res: Response) => {
+  let starterQuestions = undefined;
+  if (
+    process.env.CONVERSATION_STARTERS &&
+    process.env.CONVERSATION_STARTERS.trim()
+  ) {
+    starterQuestions = process.env.CONVERSATION_STARTERS.trim().split("\n");
+  }
+  return res.status(200).json({
+    starterQuestions,
+  });
+};
+
+export const chatLlamaCloudConfig = async (_req: Request, res: Response) => {
+  if (!process.env.LLAMA_CLOUD_API_KEY) {
+    return res.status(500).json({
+      error: "env variable LLAMA_CLOUD_API_KEY is required to use LlamaCloud",
+    });
+  }
+  const config = {
+    projects: await LLamaCloudFileService.getAllProjectsWithPipelines(),
+    pipeline: {
+      pipeline: process.env.LLAMA_CLOUD_INDEX_NAME,
+      project: process.env.LLAMA_CLOUD_PROJECT_NAME,
+    },
+  };
+  return res.status(200).json(config);
+};

templates/types/multiagent/express/src/controllers/chat.controller.ts

@@ -0,0 +1,34 @@
+import { Message, StreamData, streamToResponse } from "ai";
+import { Request, Response } from "express";
+import { ChatMessage } from "llamaindex";
+import { createWorkflow } from "../workflow";
+import { toDataStream } from "../workflow/stream";
+import { createStreamTimeout } from "./llamaindex/streaming/events";
+
+export const chat = async (req: Request, res: Response) => {
+  const vercelStreamData = new StreamData();
+  const streamTimeout = createStreamTimeout(vercelStreamData);
+  try {
+    const { messages }: { messages: Message[] } = req.body;
+    const userMessage = messages.pop();
+    if (!messages || !userMessage || userMessage.role !== "user") {
+      return res.status(400).json({
+        error:
+          "messages are required in the request body and the last message must be from the user",
+      });
+    }
+
+    const chatHistory = messages as ChatMessage[];
+    const agent = await createWorkflow(chatHistory, vercelStreamData);
+    agent.run(userMessage.content);
+    const stream = toDataStream(agent.streamEvents(), vercelStreamData);
+    return streamToResponse(stream, res, {}, vercelStreamData);
+  } catch (error) {
+    console.error("[LlamaIndex]", error);
+    return res.status(500).json({
+      detail: (error as Error).message,
+    });
+  } finally {
+    clearTimeout(streamTimeout);
+  }
+};

templates/types/multiagent/express/src/observability/index.ts

@@ -0,0 +1 @@
+export const initObservability = () => {};

templates/types/multiagent/express/src/routes/chat.route.ts

@@ -0,0 +1,12 @@
+import express, { Router } from "express";
+import { chatConfig } from "../controllers/chat-config.controller";
+import { chat } from "../controllers/chat.controller";
+import { initSettings } from "../controllers/engine/settings";
+
+const llmRouter: Router = express.Router();
+
+initSettings();
+llmRouter.route("/").post(chat);
+llmRouter.route("/config").get(chatConfig);
+
+export default llmRouter;