Release 0.11.23 (#2136 )

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: marcusschiesser <17126+marcusschiesser@users.noreply.github.com>
feat: Add toolcall callbacks to agent workflows (#2137 )
2026-07-01 22:14:03 -04:00 · 2025-07-28 14:07:23 +08:00 · 2025-07-24 15:37:14 +08:00 · 2025-07-23 12:55:35 +08:00 · 2025-07-23 12:37:02 +08:00 · 2025-07-23 11:30:04 +08:00
228 changed files with 9631 additions and 776 deletions
@@ -1,5 +1,100 @@
 # @llamaindex/doc

+## 0.2.46
+
+### Patch Changes
+
+- Updated dependencies [f29799e]
+- Updated dependencies [7224c06]
+  - @llamaindex/workflow@1.1.19
+  - @llamaindex/core@0.6.18
+  - llamaindex@0.11.23
+  - @llamaindex/cloud@4.0.27
+  - @llamaindex/node-parser@2.0.18
+  - @llamaindex/openai@0.4.13
+  - @llamaindex/readers@3.1.17
+
+## 0.2.45
+
+### Patch Changes
+
+- Updated dependencies [9ed3195]
+  - @llamaindex/workflow@1.1.18
+  - llamaindex@0.11.22
+
+## 0.2.44
+
+### Patch Changes
+
+- 38da40b: feat: VectoryMemoryBlock
+- Updated dependencies [38da40b]
+  - @llamaindex/core@0.6.17
+  - @llamaindex/cloud@4.0.26
+  - llamaindex@0.11.21
+  - @llamaindex/node-parser@2.0.17
+  - @llamaindex/openai@0.4.12
+  - @llamaindex/readers@3.1.16
+  - @llamaindex/workflow@1.1.17
+
+## 0.2.43
+
+### Patch Changes
+
+- ea15e75: Minor updates in deployment docs
+
+## 0.2.42
+
+### Patch Changes
+
+- a8ec08c: fix: ensure correct message content in agent workflow
+- Updated dependencies [a8ec08c]
+- Updated dependencies [2967d57]
+  - @llamaindex/core@0.6.16
+  - @llamaindex/workflow@1.1.16
+  - @llamaindex/cloud@4.0.25
+  - llamaindex@0.11.20
+  - @llamaindex/node-parser@2.0.16
+  - @llamaindex/openai@0.4.11
+  - @llamaindex/readers@3.1.15
+
+## 0.2.41
+
+### Patch Changes
+
+- Updated dependencies [856dd8c]
+  - @llamaindex/openai@0.4.10
+
+## 0.2.40
+
+### Patch Changes
+
+- Updated dependencies [7ad3411]
+- Updated dependencies [5da5b3c]
+- Updated dependencies [a1fdb07]
+  - @llamaindex/core@0.6.15
+  - @llamaindex/workflow@1.1.15
+  - @llamaindex/openai@0.4.9
+  - @llamaindex/cloud@4.0.24
+  - llamaindex@0.11.19
+  - @llamaindex/node-parser@2.0.15
+  - @llamaindex/readers@3.1.14
+
+## 0.2.39
+
+### Patch Changes
+
+- Updated dependencies [a1b1598]
+  - @llamaindex/cloud@4.0.23
+  - llamaindex@0.11.18
+
+## 0.2.38
+
+### Patch Changes
+
+- Updated dependencies [d2be868]
+  - @llamaindex/cloud@4.0.22
+  - llamaindex@0.11.17
+
 ## 0.2.37

 ### Patch Changes
@@ -27,6 +27,33 @@ const config = {
        destination: "/docs/workflows/:path*",
        permanent: true,
      },
+      {
+        source: "/docs/llamaindex/getting_started/installation/node.mdx",
+        destination:
+          "/docs/llamaindex/getting_started/installation/server-apis.mdx",
+        permanent: true,
+      },
+      {
+        source: "/docs/llamaindex/getting_started/installation/typescript.mdx",
+        destination: "/docs/llamaindex/getting_started/installation/index.mdx",
+        permanent: true,
+      },
+      {
+        source: "/docs/llamaindex/getting_started/installation/next.mdx",
+        destination: "/docs/llamaindex/getting_started/installation/nextjs.mdx",
+        permanent: true,
+      },
+      {
+        source: "/docs/llamaindex/getting_started/installation/vite.mdx",
+        destination: "/docs/llamaindex/getting_started/installation/index.mdx",
+        permanent: true,
+      },
+      {
+        source: "/docs/llamaindex/getting_started/installation/cloudflare.mdx",
+        destination:
+          "/docs/llamaindex/getting_started/installation/serverless.mdx",
+        permanent: true,
+      },
    ];
  },
  turbopack: {
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/doc",
-  "version": "0.2.37",
+  "version": "0.2.46",
  "private": true,
  "scripts": {
    "postinstall": "fumadocs-mdx",
@@ -1,4 +1,4 @@
-import { MockLLM } from "@llamaindex/core/utils";
+import { MockLLM } from "@llamaindex/core/llms/mock";
 import { LlamaIndexAdapter, type Message } from "ai";
 import { Settings, SimpleChatEngine, type ChatMessage } from "llamaindex";
 import { NextResponse, type NextRequest } from "next/server";
@@ -1,6 +1,6 @@
 import { AIProvider } from "@/actions";
 import { TooltipProvider } from "@/components/ui/tooltip";
-import { GoogleAnalytics } from "@next/third-parties/google";
+import { GoogleAnalytics, GoogleTagManager } from "@next/third-parties/google";
 import { RootProvider } from "fumadocs-ui/provider";
 import { Inter } from "next/font/google";
 import type { ReactNode } from "react";
@@ -36,6 +36,7 @@ export default function Layout({ children }: { children: ReactNode }) {
          LlamaIndex.TS - Build LLM-powered document agents and workflows
        </title>
      </head>
+      <GoogleTagManager gtmId="GTM-WWRFB36R" />
      <body className="flex min-h-screen flex-col">
        <TooltipProvider>
          <AIProvider>
@@ -19,3 +19,8 @@ npm run dev
 to start the development server. You can then visit [http://localhost:3000](http://localhost:3000) to see your app, which should look something like this:

 ![create-llama interface](/images/create_llama.png)
+
+## Learn more
+
+- [Learn more about `create-llama`](https://github.com/run-llama/create-llama)
+- [Want to use the same UI components? You can use our React components](https://ui.llamaindex.ai/)
@@ -17,7 +17,8 @@ npm i
 Then you can run any example in the folder with `tsx`, e.g.:

 ```bash npm2yarn
-npx tsx ./vectorIndex.ts
+export OPENAI_API_KEY=your-api-key
+npx tsx ./agents/agent/openai.ts
 ```

 ## Try examples online
@@ -1,70 +0,0 @@
---
-title: With Cloudflare Worker
-description: In this guide, you'll learn how to use LlamaIndex with CloudFlare Worker
---
-
-Before you start, make sure you have try LlamaIndex.TS in Node.js to make sure you understand the basics.
-
-<Card
-  title="Getting Started with LlamaIndex.TS in Node.js"
-  href="/docs/llamaindex/getting_started/installation/node"
-/>
-
-Also, you need have the basic understanding of <a href='https://developers.cloudflare.com/workers/'><SiCloudflareworkers className="inline mr-2" color="#F38020" />Cloudflare Worker</a>.
-
-## Adding environment variables
-
-```ts
-export default {
-  async fetch(request: Request, env: Env): Promise<Response> {
-    const { setEnvs } = await import("@llamaindex/env");
-    setEnvs(env);
-    const { OpenAIAgent } = await import("@llamaindex/openai");
-    // Start your code here
-    return new Response("Hello, world!");
-  },
-};
-```
-
-Then, you need create `.dev.vars` and add LLM api keys for the local development, such as `OPENAI_API_KEY` for OpenAI API key.
-
-<Callout type="warn">Do not commit the api key to git repository.</Callout>
-
-## Integrating with Hono
-
-```ts
-import { Hono } from "hono";
-
-type Bindings = {
-  OPENAI_API_KEY: string;
-};
-
-const app = new Hono<{
-  Bindings: Bindings;
-}>();
-
-app.post("/llm", async (c) => {
-  const { setEnvs } = await import("@llamaindex/env");
-  setEnvs(c.env);
-
-  // ...
-
-  return new Response('Hello, world!');
-})
-
-export default {
-  fetch: app.fetch,
-};
-```
-
-## Difference between Node.js and Cloudflare Worker
-
-In Cloudflare Worker and similar serverless JS environment, you need to be aware of the following differences:
-
- Some Node.js modules are not available in Cloudflare Worker, such as `node:fs`, `node:child_process`, `node:cluster`...
- You are recommend to design your code using network request, such as use `fetch` API to communicate with database, instead of a long-running process in Node.js.
- Some of LlamaIndex.TS packages are not available in Cloudflare Worker, for example `@llamaindex/readers` and `@llamaindex/huggingface`.
- The main `llamaindex` is designed to work in all JavaScript environment, including Cloudflare Worker. If you find any issue, please report to us.
- `@llamaindex/env` is a JS environment binding module, which polyfill some Node.js/Modern Web API (for example, we have a memory based `fs` module, and Crypto API polyfill). It is designed to work in all JavaScript environment, including Cloudflare Worker.
-
-
@@ -1,69 +1,177 @@
 ---
 title: Installation
-description: How to install llamaindex packages.
+description: How to install and set up LlamaIndex.TS for your project.
 ---

-To install llamaindex, run the following command:
+## Quick Start
+
+Install the core package:

 ```package-install
 npm i llamaindex
 ```

-In most cases, you'll also need an LLM package and the Workflow package to use LlamaIndex. For example, to use the OpenAI LLM with agents, you would install the following:
+In most cases, you'll also need an LLM provider and the Workflow package:

 ```package-install
 npm i @llamaindex/openai @llamaindex/workflow
 ```

-Go to [LLM APIs](/docs/llamaindex/modules/models/llms) to find out how to use other LLMs.
+## Environment Setup

+### API Keys

-## Frameworks
+Most LLM providers require API keys. Set your OpenAI key (or other provider):

-LlamaIndex supports a wide range of frameworks and runtimes. Click on the card below to learn more.
+```bash
+export OPENAI_API_KEY=your-api-key
+```
+
+Or use a `.env` file:
+
+```bash
+echo "OPENAI_API_KEY=your-api-key" > .env
+```
+
+<Callout type="warn">Never commit API keys to your repository.</Callout>
+
+### Loading Environment Variables
+
+For Node.js applications:
+
+```bash
+node --env-file .env your-script.js
+```
+
+For other environments, see the deployment-specific guides below.
+
+## TypeScript Configuration
+
+LlamaIndex.TS is built with TypeScript and provides excellent type safety. Add these settings to your `tsconfig.json`:
+
+```json5
+{
+  "compilerOptions": {
+    // Essential for module resolution
+    "moduleResolution": "bundler", // or "nodenext" | "node16" | "node"
+    
+    // Required for Web Stream API support
+    "lib": ["DOM.AsyncIterable"],
+    
+    // Recommended for better compatibility
+    "target": "es2020",
+    "module": "esnext"
+  }
+}
+```
+
+## Running your first agent
+
+### Set up
+
+If you don't already have a project, you can create a new one in a new folder:
+
+```package-install
+npm init
+npm i -D typescript @types/node
+npm i @llamaindex/openai @llamaindex/workflow llamaindex zod
+```
+
+### Run the agent
+
+Create the file `example.ts`. This code will:
+
+- Create two tools for use by the agent:
+  - A `sumNumbers` tool that adds two numbers
+  - A `divideNumbers` tool that divides numbers
+- Give an example of the data structure we wish to generate
+- Prompt the LLM with instructions and the example, plus a sample transcript
+
+<include cwd>../../examples/agents/agent/openai.ts</include>
+
+To run the code:
+
+```package-install
+npx tsx example.ts
+```
+
+You should expect output something like:
+
+```
+{
+  result: '5 + 5 is 10. Then, 10 divided by 2 is 5.',
+  state: {
+    memory: Memory {
+      messages: [Array],
+      tokenLimit: 30000,
+      shortTermTokenLimitRatio: 0.7,
+      memoryBlocks: [],
+      memoryCursor: 0,
+      adapters: [Object]
+    },
+    scratchpad: [],
+    currentAgentName: 'Agent',
+    agents: [ 'Agent' ],
+    nextAgentName: null
+  }
+}
+Done
+```
+
+## Performance Optimization
+
+### Tokenization Speed
+
+Install `gpt-tokenizer` for 60x faster tokenization (Node.js environments only):
+
+```package-install
+npm i gpt-tokenizer
+```
+
+LlamaIndex will automatically use this when available.
+
+## Deployment Guides
+
+Choose your deployment target:

 <Cards>
-	<Card title={
-		<>
-			<SiNodedotjs className="inline" color="#5FA04E" /> Node.js
-		</>
-	} href="/docs/llamaindex/getting_started/installation/node" />
-	<Card title={
-		<>
-			<SiTypescript className="inline" color="#3178C6" /> TypeScript
-		</>
-	} href="/docs/llamaindex/getting_started/installation/typescript" />
-	<Card title={
-		<>
-			<SiVite className='inline' color='#646CFF' /> Vite
-		</>
-	} href="/docs/llamaindex/getting_started/installation/vite" />
-	<Card
-		title={
-			<>
-				<SiNextdotjs className='inline' /> Next.js (React Server Component)
-			</>
-		}
-		href="/docs/llamaindex/getting_started/installation/next"
-	/>
-	<Card title={
-		<>
-			<SiCloudflareworkers className='inline' color='#F38020' /> Cloudflare Workers
-		</>
-	} href="/docs/llamaindex/getting_started/installation/cloudflare" />
+  <Card 
+    title="Server APIs & Backends" 
+    description="Express, Fastify, Koa, standalone Node.js servers"
+    href="/docs/llamaindex/getting_started/installation/server-apis" 
+  />
+  <Card 
+    title="Serverless Functions" 
+    description="Vercel, Netlify, AWS Lambda, Cloudflare Workers"
+    href="/docs/llamaindex/getting_started/installation/serverless" 
+  />
+  <Card 
+    title="Next.js Applications" 
+    description="API routes, server components, edge runtime"
+    href="/docs/llamaindex/getting_started/installation/nextjs" 
+  />
+  <Card 
+    title="Troubleshooting" 
+    description="Common issues, bundle optimization, compatibility"
+    href="/docs/llamaindex/getting_started/installation/troubleshooting" 
+  />
 </Cards>

-## What's next?
+## LLM/Embedding Providers
+
+Go to [LLM APIs](/docs/llamaindex/modules/models/llms) and [Embedding APIs](/docs/llamaindex/modules/models/embeddings) to find out how to use different LLM and embedding providers beyond OpenAI.
+
+## What's Next?

 <Cards>
-	<Card
-		title="Learn LlamaIndex.TS"
-		description="Learn how to use LlamaIndex.TS by starting with one of our tutorials."
-		href="/docs/llamaindex/tutorials/rag"
-	/>
-	<Card
-		title="Show me code examples"
-		description="Explore code examples using LlamaIndex.TS."
-		href="/docs/llamaindex/getting_started/examples"
-	/>
+  <Card
+    title="Learn LlamaIndex.TS"
+    description="Learn how to use LlamaIndex.TS by starting with one of our tutorials."
+    href="/docs/llamaindex/tutorials/basic_agent"
+  />
+  <Card
+    title="Show me code examples"
+    description="Explore code examples using LlamaIndex.TS."
+    href="/docs/llamaindex/getting_started/examples"
+  />
 </Cards>
@@ -1,4 +1,4 @@
 {
  "title": "Installation",
-  "pages": ["node", "typescript", "next", "vite", "cloudflare"]
+  "pages": ["server-apis", "serverless", "nextjs", "troubleshooting"]
 }
@@ -1,41 +0,0 @@
---
-title: With Next.js
-description: In this guide, you'll learn how to use LlamaIndex with Next.js.
---
-
-Before you start, make sure you have try LlamaIndex.TS in Node.js to make sure you understand the basics.
-
-<Card
-  title="Getting Started with LlamaIndex.TS in Node.js"
-  href="/docs/llamaindex/getting_started/installation/node"
-/>
-
-## Differences between Node.js and Next.js
-
-Next.js is a React framework that has both server side compatibility and client side compatibility.
-This means that you need to be careful when using LlamaIndex.TS in Next.js.
-Don't leak the import data like API keys to the client side.
-
-Also, in Next.js, there is build time and runtime. Some computations can be done at build time like Document embedding could be done at build time for better performance.
-Where as the `llamaindex` package is working with Next.js, some provider packages like `@llamaindex/huggingface` are not working well with Next.js. This is due to the upstream dependencies used by the provider package. 
-
-Make sure to use `withLlamaIndex` to make sure that LlamaIndex.TS works well with Next.js.
-
-```js
-// next.config.mjs / next.config.ts
-import withLlamaIndex from "llamaindex/next";
-
-/** @type {import('next').NextConfig} */
-const nextConfig = {};
-
-export default withLlamaIndex(nextConfig);
-```
-
-If you see any dependency issues, you are welcome to open an issue on the GitHub.
-
-## Edge Runtime
-
-[Vercel Edge Runtime](https://edge-runtime.vercel.app/) is a subset of Node.js APIs. Similar to [Cloudflare Workers](/docs/llamaindex/getting_started/installation/cloudflare#difference-between-nodejs-and-cloudflare-worker),
-it is a serverless platform that runs your code on the edge.
-
-Not all features of Node.js are supported in Vercel Edge Runtime, so does LlamaIndex.TS, we are working on more compatibility with all JavaScript runtimes.
@@ -0,0 +1,405 @@
+---
+title: Next.js Applications
+description: Deploy LlamaIndex.TS in Next.js applications with API routes, server components, and edge runtime.
+---
+
+This guide covers integrating LlamaIndex.TS agents with Next.js applications.
+
+## Essential Configuration
+
+### Next.js Config
+
+Use `withLlamaIndex` to ensure compatibility:
+
+```javascript
+// next.config.mjs
+import withLlamaIndex from "llamaindex/next";
+
+/** @type {import('next').NextConfig} */
+const nextConfig = {
+  // Your existing config
+};
+
+export default withLlamaIndex(nextConfig);
+```
+
+## API Routes
+
+### App Router (Recommended)
+
+```typescript
+// app/api/chat/route.ts
+import { agent } from "@llamaindex/workflow";
+import { tool } from "llamaindex";
+import { openai } from "@llamaindex/openai";
+import { z } from "zod";
+import { NextRequest, NextResponse } from "next/server";
+
+// Initialize agent once (consider using a singleton pattern)
+let myAgent: any = null;
+
+async function initializeAgent() {
+  if (myAgent) return myAgent;
+  
+  try {
+    const greetTool = tool({
+      name: "greet",
+      description: "Greets a user with their name",
+      parameters: z.object({
+        name: z.string(),
+      }),
+      execute: ({ name }) => `Hello, ${name}! How can I help you today?`,
+    });
+
+    myAgent = agent({
+      tools: [greetTool],
+      llm: openai({ model: "gpt-4o-mini" }),
+    });
+    
+    return myAgent;
+  } catch (error) {
+    console.error("Failed to initialize agent:", error);
+    throw error;
+  }
+}
+
+export async function POST(request: NextRequest) {
+  try {
+    const { message } = await request.json();
+    
+    if (!message || typeof message !== 'string') {
+      return NextResponse.json(
+        { error: "Message is required and must be a string" },
+        { status: 400 }
+      );
+    }
+    
+    const agent = await initializeAgent();
+    const result = await agent.run(message);
+    
+    return NextResponse.json({ response: result.data });
+  } catch (error) {
+    console.error("Chat error:", error);
+    return NextResponse.json(
+      { error: "Internal server error" },
+      { status: 500 }
+    );
+  }
+}
+```
+
+### Pages Router (Legacy)
+
+```typescript
+// pages/api/chat.ts
+import { agent } from "@llamaindex/workflow";
+import { tool } from "llamaindex";
+import { openai } from "@llamaindex/openai";
+import { z } from "zod";
+import type { NextApiRequest, NextApiResponse } from "next";
+
+let myAgent: any = null;
+
+async function initializeAgent() {
+  if (myAgent) return myAgent;
+  
+  const timeTool = tool({
+    name: "getCurrentTime",
+    description: "Gets the current time",
+    parameters: z.object({}),
+    execute: () => new Date().toISOString(),
+  });
+
+  myAgent = agent({
+    tools: [timeTool],
+    llm: openai({ model: "gpt-4o-mini" }),
+  });
+  
+  return myAgent;
+}
+
+export default async function handler(
+  req: NextApiRequest,
+  res: NextApiResponse
+) {
+  if (req.method !== "POST") {
+    return res.status(405).json({ error: "Method not allowed" });
+  }
+  
+  try {
+    const { message } = req.body;
+    
+    const agent = await initializeAgent();
+    const result = await agent.run(message);
+    
+    res.json({ response: result.data });
+  } catch (error) {
+    console.error("Chat error:", error);
+    res.status(500).json({ error: "Internal server error" });
+  }
+}
+```
+
+## Server Components
+
+Initialize agents in server components:
+
+```typescript
+// app/chat/page.tsx
+import { agent } from "@llamaindex/workflow";
+import { tool } from "llamaindex";
+import { openai } from "@llamaindex/openai";
+import { z } from "zod";
+
+async function initializeAgent() {
+  const helpTool = tool({
+    name: "getHelp",
+    description: "Provides help information",
+    parameters: z.object({
+      topic: z.string().optional(),
+    }),
+    execute: ({ topic }) => {
+      if (topic) {
+        return `Here's help for ${topic}: This is a helpful resource about ${topic}.`;
+      }
+      return "Available topics: general, troubleshooting, api, deployment";
+    },
+  });
+
+  return agent({
+    tools: [helpTool],
+    llm: openai({ model: "gpt-4o-mini" }),
+  });
+}
+
+export default async function ChatPage() {
+  const chatAgent = await initializeAgent();
+  
+  return (
+    <div>
+      <h1>Chat Interface</h1>
+      <p>Agent initialized and ready to help!</p>
+      {/* Your chat UI components */}
+    </div>
+  );
+}
+```
+
+## Edge Runtime
+
+The Edge Runtime has limited Node.js API access:
+
+```typescript
+// app/api/chat-edge/route.ts
+import { NextRequest, NextResponse } from "next/server";
+
+export const runtime = "edge";
+
+export async function POST(request: NextRequest) {
+  const { setEnvs } = await import("@llamaindex/env");
+  setEnvs(process.env);
+  
+  try {
+    const { message } = await request.json();
+    
+    const { agent } = await import("@llamaindex/workflow");
+    const { tool } = await import("llamaindex");
+    const { openai } = await import("@llamaindex/openai");
+    const { z } = await import("zod");
+
+    const timeTool = tool({
+      name: "time",
+      description: "Gets current time",
+      parameters: z.object({}),
+      execute: () => new Date().toISOString(),
+    });
+
+    const myAgent = agent({
+      tools: [timeTool],
+      llm: openai({ model: "gpt-4o-mini" }),
+    });
+
+    const result = await myAgent.run(message);
+    return NextResponse.json({ response: result.data });
+  } catch (error) {
+    return NextResponse.json({ error: error.message }, { status: 500 });
+  }
+}
+```
+
+## Streaming Responses
+
+Implement streaming for better user experience:
+
+```typescript
+// app/api/chat-stream/route.ts
+import { agent } from "@llamaindex/workflow";
+import { tool } from "llamaindex";
+import { openai } from "@llamaindex/openai";
+import { agentStreamEvent } from "@llamaindex/workflow";
+import { NextRequest } from "next/server";
+import { z } from "zod";
+
+// Initialize agent once (consider using a singleton pattern)
+let myAgent: any = null;
+
+async function initializeAgent() {
+  if (myAgent) return myAgent;
+  
+  try {
+    const greetTool = tool({
+      name: "greet",
+      description: "Greets a user with their name",
+      parameters: z.object({
+        name: z.string(),
+      }),
+      execute: ({ name }) => `Hello, ${name}! How can I help you today?`,
+    });
+
+    myAgent = agent({
+      tools: [greetTool],
+      llm: openai({ model: "gpt-4o-mini" }),
+    });
+    
+    return myAgent;
+  } catch (error) {
+    console.error("Failed to initialize agent:", error);
+    throw error;
+  }
+}
+
+export async function POST(request: NextRequest) {
+  const { message } = await request.json();
+  
+  const stream = new ReadableStream({
+    async start(controller) {
+      try {
+        const agent = await initializeAgent();
+        const events = agent.runStream(message);
+        
+        for await (const event of events) {
+          if (agentStreamEvent.include(event)) {
+            controller.enqueue(new TextEncoder().encode(event.data.delta));
+          }
+        }
+        
+        controller.close();
+      } catch (error) {
+        controller.error(error);
+      }
+    },
+  });
+  
+  return new Response(stream, {
+    headers: {
+      "Content-Type": "text/plain",
+      "Transfer-Encoding": "chunked",
+    },
+  });
+}
+```
+
+## Client-side Integration
+
+### React Hook for API Calls
+
+```typescript
+// hooks/useAgentChat.ts
+import { useState } from "react";
+
+export function useAgentChat() {
+  const [loading, setLoading] = useState(false);
+  const [error, setError] = useState<string | null>(null);
+  const [response, setResponse] = useState<string | null>(null);
+  
+  const chat = async (message: string) => {
+    setLoading(true);
+    setError(null);
+    
+    try {
+      const res = await fetch("/api/chat", {
+        method: "POST",
+        headers: { "Content-Type": "application/json" },
+        body: JSON.stringify({ message }),
+      });
+      
+      if (!res.ok) {
+        throw new Error(`HTTP error! status: ${res.status}`);
+      }
+      
+      const data = await res.json();
+      setResponse(data.response);
+    } catch (err) {
+      setError(err instanceof Error ? err.message : "An error occurred");
+    } finally {
+      setLoading(false);
+    }
+  };
+  
+  return { chat, loading, error, response };
+}
+```
+
+### Chat Component
+
+```typescript
+// components/ChatInterface.tsx
+"use client";
+
+import { useState } from "react";
+import { useAgentChat } from "@/hooks/useAgentChat";
+
+export default function ChatInterface() {
+  const [message, setMessage] = useState("");
+  const { chat, loading, error, response } = useAgentChat();
+  
+  const handleSubmit = async (e: React.FormEvent) => {
+    e.preventDefault();
+    if (!message.trim()) return;
+    
+    await chat(message);
+    setMessage("");
+  };
+  
+  return (
+    <div className="max-w-2xl mx-auto p-4">
+      <form onSubmit={handleSubmit} className="mb-4">
+        <input
+          type="text"
+          value={message}
+          onChange={(e) => setMessage(e.target.value)}
+          placeholder="Send a message..."
+          className="w-full p-2 border rounded"
+          disabled={loading}
+        />
+        <button
+          type="submit"
+          disabled={loading || !message.trim()}
+          className="mt-2 px-4 py-2 bg-blue-500 text-white rounded disabled:opacity-50"
+        >
+          {loading ? "Thinking..." : "Send"}
+        </button>
+      </form>
+      
+      {error && (
+        <div className="p-3 mb-4 bg-red-100 border border-red-400 text-red-700 rounded">
+          Error: {error}
+        </div>
+      )}
+      
+      {response && (
+        <div className="p-3 bg-gray-100 border rounded">
+          <strong>Agent:</strong>
+          <p>{response}</p>
+        </div>
+      )}
+    </div>
+  );
+}
+```
+
+## Next Steps
+
+- Learn about [serverless deployment](/docs/llamaindex/getting_started/installation/serverless)
+- Explore [server APIs](/docs/llamaindex/getting_started/installation/server-apis)
+- Check [troubleshooting guide](/docs/llamaindex/getting_started/installation/troubleshooting) for common issues 
@@ -1,40 +0,0 @@
---
-title: With Node.js/Bun/Deno
-description: In this guide, you'll learn how to use LlamaIndex with Node.js, Bun, and Deno.
---
-
-## Adding environment variables
-
-By default, LlamaIndex uses OpenAI provider, which requires an API key. You can set the `OPENAI_API_KEY` environment variable to authenticate with OpenAI.
-
-```shell
-export OPENAI_API_KEY=your-api-key
-```
-
-Or you can use a `.env` file:
-
-```shell
-echo "OPENAI_API_KEY=your-api-key" > .env
-node --env-file .env your-script.js
-```
-
-<Callout type="warn">Do not commit the api key to git repository.</Callout>
-
-For more information, see the [How to read environment variables from Node.js](https://nodejs.org/en/learn/command-line/how-to-read-environment-variables-from-nodejs).
-
-## Performance Optimization
-
-By the default, we are using `js-tiktoken` for tokenization. You can install `gpt-tokenizer` which is then automatically used by LlamaIndex to get a 60x speedup for tokenization:
-
-```package-install
-npm i gpt-tokenizer
-```
-
-**Note**: This only works for Node.js
-
-## TypeScript support
-
-<Card
-	title="Getting Started with LlamaIndex.TS in TypeScript"
-	href="/docs/llamaindex/getting_started/installation/typescript"
-/>
@@ -0,0 +1,211 @@
+---
+title: Server APIs & Backends
+description: Deploy LlamaIndex.TS in server environments like Express, Fastify, and standalone Node.js applications.
+---
+
+This guide covers adding LlamaIndex.TS agents to traditional server environments where you have full Node.js runtime access.
+
+## Supported Runtimes
+
+LlamaIndex.TS works seamlessly with:
+
+- **Node.js** (v18+)
+- **Bun** (v1.0+)
+- **Deno** (v1.30+)
+
+## Common Server Frameworks
+
+### Express.js
+
+```typescript
+import express from 'express';
+import { agent } from '@llamaindex/workflow';
+import { tool } from 'llamaindex';
+import { openai } from '@llamaindex/openai';
+import { z } from 'zod';
+
+const app = express();
+app.use(express.json());
+
+// Initialize agent once at startup
+let myAgent: any;
+
+async function initializeAgent() {
+  // Create tools for the agent
+  const sumTool = tool({
+    name: "sum",
+    description: "Adds two numbers",
+    parameters: z.object({
+      a: z.number(),
+      b: z.number(),
+    }),
+    execute: ({ a, b }) => a + b,
+  });
+
+  const multiplyTool = tool({
+    name: "multiply",
+    description: "Multiplies two numbers",
+    parameters: z.object({
+      a: z.number(),
+      b: z.number(),
+    }),
+    execute: ({ a, b }) => a * b,
+  });
+
+  // Create the agent
+  myAgent = agent({
+    tools: [sumTool, multiplyTool],
+    llm: openai({ model: "gpt-4o-mini" }),
+  });
+}
+
+app.post('/api/chat', async (req, res) => {
+  try {
+    const { message } = req.body;
+    const result = await myAgent.run(message);
+    res.json({ response: result.data });
+  } catch (error) {
+    res.status(500).json({ error: 'Chat failed' });
+  }
+});
+
+// Initialize and start server
+initializeAgent().then(() => {
+  app.listen(3000, () => {
+    console.log('Server running on port 3000');
+  });
+});
+```
+
+### Fastify
+
+```typescript
+import Fastify from 'fastify';
+import { agent } from '@llamaindex/workflow';
+import { tool } from 'llamaindex';
+import { openai } from '@llamaindex/openai';
+import { z } from 'zod';
+
+const fastify = Fastify();
+let myAgent: any;
+
+async function initializeAgent() {
+  const sumTool = tool({
+    name: "sum",
+    description: "Adds two numbers",
+    parameters: z.object({
+      a: z.number(),
+      b: z.number(),
+    }),
+    execute: ({ a, b }) => a + b,
+  });
+
+  myAgent = agent({
+    tools: [sumTool],
+    llm: openai({ model: "gpt-4o-mini" }),
+  });
+}
+
+fastify.post('/api/chat', async (request, reply) => {
+  try {
+    const { message } = request.body as { message: string };
+    const result = await myAgent.run(message);
+    return { response: result.data };
+  } catch (error) {
+    reply.status(500).send({ error: 'Chat failed' });
+  }
+});
+
+const start = async () => {
+  await initializeAgent();
+  await fastify.listen({ port: 3000 });
+  console.log('Server running on port 3000');
+};
+
+start();
+```
+
+### Hono
+
+```typescript
+import { Hono } from "hono";
+import { agent } from "@llamaindex/workflow";
+import { tool } from "llamaindex";
+import { openai } from "@llamaindex/openai";
+import { z } from "zod";
+
+type Bindings = {
+  OPENAI_API_KEY: string;
+};
+
+const app = new Hono<{ Bindings: Bindings }>();
+
+app.post("/api/chat", async (c) => {
+  const { setEnvs } = await import("@llamaindex/env");
+  setEnvs(c.env);
+  
+  const { message } = await c.req.json();
+  
+  const greetTool = tool({
+    name: "greet",
+    description: "Greets a user",
+    parameters: z.object({
+      name: z.string(),
+    }),
+    execute: ({ name }) => `Hello, ${name}!`,
+  });
+
+  const myAgent = agent({
+    tools: [greetTool],
+    llm: openai({ model: "gpt-4o-mini" }),
+  });
+  
+  try {
+    const result = await myAgent.run(message);
+    return c.json({ response: result.data });
+  } catch (error) {
+    return c.json({ error: error.message }, 500);
+  }
+});
+
+export default app;
+```
+
+## Streaming Responses
+
+For real-time agent responses:
+
+```typescript
+import { agentStreamEvent } from "@llamaindex/workflow";
+
+app.post('/api/chat-stream', async (req, res) => {
+  const { message } = req.body;
+  
+  res.writeHead(200, {
+    'Content-Type': 'text/plain',
+    'Transfer-Encoding': 'chunked',
+  });
+  
+  try {
+    const events = myAgent.runStream(message);
+    
+    for await (const event of events) {
+      if (agentStreamEvent.include(event)) {
+        res.write(event.data.delta);
+      }
+    }
+    
+    res.end();
+  } catch (error) {
+    res.write('Error: ' + error.message);
+    res.end();
+  }
+});
+```
+
+
+## Next Steps
+
+- Learn about [serverless deployment](/docs/llamaindex/getting_started/installation/serverless)
+- Explore [Next.js integration](/docs/llamaindex/getting_started/installation/nextjs)
+- Check [troubleshooting guide](/docs/llamaindex/getting_started/installation/troubleshooting) for common issues 
@@ -0,0 +1,240 @@
+---
+title: Serverless Functions
+description: Deploy LlamaIndex.TS in serverless environments like Vercel, Netlify, AWS Lambda, and Cloudflare Workers.
+---
+
+This guide covers adding LlamaIndex.TS agents to serverless environments where you have execution time and memory constraints.
+
+## Cloudflare Workers
+
+```typescript
+export default {
+  async fetch(request: Request, env: Env): Promise<Response> {
+    const { setEnvs } = await import("@llamaindex/env");
+    setEnvs(env);
+    
+    const { agent } = await import("@llamaindex/workflow");
+    const { openai } = await import("@llamaindex/openai");
+    const { tool } = await import("llamaindex");
+    const { z } = await import("zod");
+
+    const timeTool = tool({
+      name: "getCurrentTime",
+      description: "Gets the current time",
+      parameters: z.object({}),
+      execute: () => new Date().toISOString(),
+    });
+
+    const myAgent = agent({
+      tools: [timeTool],
+      llm: openai({ model: "gpt-4o-mini" }),
+    });
+
+    try {
+      const { message } = await request.json();
+      const result = await myAgent.run(message);
+      
+      return new Response(JSON.stringify({ response: result.data }), {
+        headers: { "Content-Type": "application/json" },
+      });
+    } catch (error) {
+      return new Response(JSON.stringify({ error: error.message }), {
+        status: 500,
+        headers: { "Content-Type": "application/json" },
+      });
+    }
+  },
+};
+```
+
+
+
+## Vercel Functions
+
+### Node.js Runtime
+
+```typescript
+// pages/api/chat.ts or app/api/chat/route.ts
+import { agent } from "@llamaindex/workflow";
+import { tool } from "llamaindex";
+import { openai } from "@llamaindex/openai";
+import { z } from "zod";
+
+export default async function handler(req, res) {
+  if (req.method !== 'POST') {
+    return res.status(405).json({ error: 'Method not allowed' });
+  }
+  
+  const { message } = req.body;
+  
+  const weatherTool = tool({
+    name: "getWeather",
+    description: "Get weather information",
+    parameters: z.object({
+      city: z.string(),
+    }),
+    execute: ({ city }) => `Weather in ${city}: 72°F, sunny`,
+  });
+
+  const myAgent = agent({
+    tools: [weatherTool],
+    llm: openai({ model: "gpt-4o-mini" }),
+  });
+  
+  try {
+    const result = await myAgent.run(message);
+    res.json({ response: result.data });
+  } catch (error) {
+    res.status(500).json({ error: error.message });
+  }
+}
+```
+
+### Edge Runtime
+
+```typescript
+// app/api/chat/route.ts
+import { NextRequest, NextResponse } from "next/server";
+
+export const runtime = "edge";
+
+export async function POST(request: NextRequest) {
+  const { setEnvs } = await import("@llamaindex/env");
+  setEnvs(process.env);
+  
+  const { message } = await request.json();
+  
+  try {
+    // Use simpler tools for edge runtime
+    const { agent } = await import("@llamaindex/workflow");
+    const { tool } = await import("llamaindex");
+    const { openai } = await import("@llamaindex/openai");
+    const { z } = await import("zod");
+
+    const timeTool = tool({
+      name: "time",
+      description: "Gets current time",
+      parameters: z.object({}),
+      execute: () => new Date().toISOString(),
+    });
+
+    const myAgent = agent({
+      tools: [timeTool],
+      llm: openai({ model: "gpt-4o-mini" }),
+    });
+
+    const result = await myAgent.run(message);
+    return NextResponse.json({ response: result.data });
+  } catch (error) {
+    return NextResponse.json({ error: error.message }, { status: 500 });
+  }
+}
+```
+
+## AWS Lambda
+
+```typescript
+import { APIGatewayProxyHandler } from "aws-lambda";
+import { agent } from "@llamaindex/workflow";
+import { tool } from "llamaindex";
+import { openai } from "@llamaindex/openai";
+import { z } from "zod";
+
+export const handler: APIGatewayProxyHandler = async (event, context) => {
+  const { message } = JSON.parse(event.body || "{}");
+  
+  const calculatorTool = tool({
+    name: "calculate",
+    description: "Performs basic math",
+    parameters: z.object({
+      expression: z.string(),
+    }),
+    execute: ({ expression }) => {
+      // Simple calculator implementation
+      try {
+        return `Result: ${eval(expression)}`;
+      } catch {
+        return "Invalid expression";
+      }
+    },
+  });
+
+  const myAgent = agent({
+    tools: [calculatorTool],
+    llm: openai({ model: "gpt-4o-mini" }),
+  });
+  
+  try {
+    const result = await myAgent.run(message);
+    
+    return {
+      statusCode: 200,
+      headers: {
+        "Content-Type": "application/json",
+        "Access-Control-Allow-Origin": "*",
+      },
+      body: JSON.stringify({ response: result.data }),
+    };
+  } catch (error) {
+    return {
+      statusCode: 500,
+      body: JSON.stringify({ error: error.message }),
+    };
+  }
+};
+```
+
+## Netlify Functions
+
+```typescript
+// netlify/functions/chat.ts
+import { Handler } from "@netlify/functions";
+import { agent } from "@llamaindex/workflow";
+import { tool } from "llamaindex";
+import { openai } from "@llamaindex/openai";
+import { z } from "zod";
+
+export const handler: Handler = async (event, context) => {
+  if (event.httpMethod !== "POST") {
+    return { statusCode: 405, body: "Method Not Allowed" };
+  }
+  
+  const { message } = JSON.parse(event.body || "{}");
+  
+  const helpTool = tool({
+    name: "help",
+    description: "Provides help information",
+    parameters: z.object({
+      topic: z.string().optional(),
+    }),
+    execute: ({ topic }) => {
+      return topic ? `Help for ${topic}` : "Available help topics";
+    },
+  });
+
+  const myAgent = agent({
+    tools: [helpTool],
+    llm: openai({ model: "gpt-4o-mini" }),
+  });
+  
+  try {
+    const result = await myAgent.run(message);
+    
+    return {
+      statusCode: 200,
+      body: JSON.stringify({ response: result.data }),
+    };
+  } catch (error) {
+    return {
+      statusCode: 500,
+      body: JSON.stringify({ error: error.message }),
+    };
+  }
+};
+```
+
+## Next Steps
+
+- Learn about [Next.js integration](/docs/llamaindex/getting_started/installation/nextjs)
+- Explore [server deployment](/docs/llamaindex/getting_started/installation/server-apis)
+- Check [troubleshooting guide](/docs/llamaindex/getting_started/installation/troubleshooting) for common issues 
@@ -0,0 +1,501 @@
+---
+title: Troubleshooting
+description: Common issues and solutions when installing and deploying LlamaIndex.TS applications.
+---
+
+This guide addresses common issues you might encounter when installing and deploying LlamaIndex.TS applications across different environments.
+
+## Installation Issues
+
+### Module Resolution Errors
+
+**Problem:** Import errors or module not found errors
+
+**Solution:** Ensure your `tsconfig.json` is properly configured:
+
+```json5
+{
+  "compilerOptions": {
+    "moduleResolution": "bundler", // or "nodenext" | "node16" | "node"
+    "lib": ["DOM.AsyncIterable"],
+    "target": "es2020",
+    "module": "esnext"
+  }
+}
+```
+
+**Alternative solution:** Try different module resolution strategies:
+
+```bash
+# Clear node_modules and reinstall
+rm -rf node_modules package-lock.json
+npm install
+
+# Or try with different package manager
+pnpm install
+# or
+yarn install
+```
+
+### TypeScript Errors
+
+**Problem:** TypeScript compilation errors with LlamaIndex imports
+
+**Solution:** Ensure you have the correct TypeScript configuration:
+
+```json5
+{
+  "compilerOptions": {
+    "strict": true,
+    "skipLibCheck": true, // Skip type checking of node_modules
+    "allowSyntheticDefaultImports": true,
+    "esModuleInterop": true
+  }
+}
+```
+
+### Package Compatibility Issues
+
+**Problem:** Some packages don't work in certain environments
+
+**Common incompatibilities:**
+- `@llamaindex/readers` - May not work in serverless environments
+- `@llamaindex/huggingface` - Limited browser/edge compatibility
+- File system readers - Don't work in browser/edge environments
+
+**Solution:** Use environment-specific alternatives:
+
+```typescript
+// Instead of file system readers in serverless
+// Use remote data sources
+async function loadDocumentsFromAPI() {
+  const response = await fetch('https://api.example.com/documents');
+  const data = await response.json();
+  return data.map(doc => new Document(doc.content));
+}
+```
+
+## Runtime Issues
+
+### Memory Errors
+
+**Problem:** Out of memory errors during index creation or querying
+
+**Solution:** Optimize memory usage:
+
+```typescript
+// Batch process large document sets
+async function batchProcessDocuments(documents: Document[], batchSize = 10) {
+  const results = [];
+  
+  for (let i = 0; i < documents.length; i += batchSize) {
+    const batch = documents.slice(i, i + batchSize);
+    const batchIndex = await VectorStoreIndex.fromDocuments(batch);
+    results.push(batchIndex);
+    
+    // Optional: Add delay between batches
+    await new Promise(resolve => setTimeout(resolve, 100));
+  }
+  
+  return results;
+}
+```
+
+**For serverless environments:**
+
+```typescript
+// Use external vector stores instead of in-memory
+// TODO: Example with Pinecone, Weaviate, etc.
+// const vectorStore = new PineconeVectorStore(/* config */);
+// const index = await VectorStoreIndex.fromVectorStore(vectorStore);
+```
+
+### API Rate Limiting
+
+**Problem:** Rate limiting errors from LLM providers
+
+**Solution:** Implement retry logic with exponential backoff:
+
+```typescript
+async function queryWithRetry(queryEngine: any, question: string, maxRetries = 3) {
+  for (let i = 0; i < maxRetries; i++) {
+    try {
+      return await queryEngine.query(question);
+    } catch (error) {
+      if (error.message.includes('rate limit') && i < maxRetries - 1) {
+        const delay = Math.pow(2, i) * 1000; // Exponential backoff
+        await new Promise(resolve => setTimeout(resolve, delay));
+        continue;
+      }
+      throw error;
+    }
+  }
+}
+```
+
+### Tokenization Performance
+
+**Problem:** Slow tokenization affecting performance
+
+**Solution:** Install faster tokenizer (Node.js only):
+
+```bash
+npm install gpt-tokenizer
+```
+
+LlamaIndex will automatically use this for 60x faster tokenization.
+
+## Bundling Issues
+
+### Bundle Size Too Large
+
+**Problem:** Large bundle sizes affecting performance
+
+**Solution:** Use dynamic imports and code splitting:
+
+```typescript
+// Lazy load LlamaIndex components
+const initializeLlamaIndex = async () => {
+  const { VectorStoreIndex, SimpleDirectoryReader } = await import("llamaindex");
+  return { VectorStoreIndex, SimpleDirectoryReader };
+};
+
+// In your API route
+export async function POST(request: NextRequest) {
+  const { VectorStoreIndex, SimpleDirectoryReader } = await initializeLlamaIndex();
+  // Use the imported modules
+}
+```
+
+### Webpack/Vite Bundling Issues
+
+**Problem:** Bundler compatibility issues
+
+**Solution for Next.js:**
+
+```javascript
+// next.config.mjs
+import withLlamaIndex from "llamaindex/next";
+
+const nextConfig = {
+  webpack: (config, { isServer }) => {
+    // Custom webpack configuration if needed
+    if (!isServer) {
+      config.resolve.fallback = {
+        ...config.resolve.fallback,
+        fs: false,
+        net: false,
+        tls: false,
+      };
+    }
+    return config;
+  },
+};
+
+export default withLlamaIndex(nextConfig);
+```
+
+**Solution for Vite:**
+
+```typescript
+// vite.config.ts
+import { defineConfig } from 'vite';
+
+export default defineConfig({
+  define: {
+    global: 'globalThis',
+  },
+  resolve: {
+    alias: {
+      // Add aliases for problematic modules
+    },
+  },
+  optimizeDeps: {
+    include: ['llamaindex'],
+  },
+});
+```
+
+## Environment-Specific Issues
+
+### Node.js Version Compatibility
+
+**Problem:** Node.js version compatibility issues
+
+**Solution:** Use supported Node.js versions:
+
+```json
+{
+  "engines": {
+    "node": ">=18.0.0"
+  }
+}
+```
+
+**Check your Node.js version:**
+
+```bash
+node --version
+```
+
+### Cloudflare Workers Issues
+
+**Problem:** Module not available in Cloudflare Workers
+
+**Solution:** Use `@llamaindex/env` for environment compatibility:
+
+```typescript
+export default {
+  async fetch(request: Request, env: Env): Promise<Response> {
+    const { setEnvs } = await import("@llamaindex/env");
+    setEnvs(env);
+    
+    // Your LlamaIndex code here
+  },
+};
+```
+
+### Vercel Edge Runtime Issues
+
+**Problem:** Limited Node.js API access in Edge Runtime
+
+**Solution:** Use standard runtime or adapt code:
+
+```typescript
+// Force standard runtime
+export const runtime = "nodejs";
+
+// Or adapt for edge
+export const runtime = "edge";
+
+export async function POST(request: NextRequest) {
+  // Use edge-compatible code only
+  const { setEnvs } = await import("@llamaindex/env");
+  setEnvs(process.env);
+  
+  // Avoid file system operations
+  // Use remote data sources
+}
+```
+
+## Performance Issues
+
+### Slow Query Responses
+
+**Problem:** Slow query performance
+
+**Solution:** Implement caching and optimization:
+
+```typescript
+import { LRUCache } from 'lru-cache';
+
+const queryCache = new LRUCache<string, string>({
+  max: 100,
+  ttl: 1000 * 60 * 10, // 10 minutes
+});
+
+export async function optimizedQuery(question: string, queryEngine: any) {
+  // Check cache first
+  const cached = queryCache.get(question);
+  if (cached) return cached;
+  
+  // Query and cache result
+  const result = await queryEngine.query(question);
+  queryCache.set(question, result);
+  
+  return result;
+}
+```
+
+### Cold Start Issues
+
+**Problem:** Slow cold starts in serverless environments
+
+**Solution:** Pre-warm your functions:
+
+```typescript
+// Pre-initialize outside handler
+let cachedQueryEngine: any = null;
+
+export async function handler(event: any) {
+  if (!cachedQueryEngine) {
+    cachedQueryEngine = await initializeQueryEngine();
+  }
+  
+  // Use cached engine
+  return await cachedQueryEngine.query(question);
+}
+```
+
+## Environment Variable Issues
+
+### Missing API Keys
+
+**Problem:** API key not found or invalid
+
+**Solution:** Verify environment variable setup:
+
+```typescript
+// Check if API key is available
+if (!process.env.OPENAI_API_KEY) {
+  throw new Error('OPENAI_API_KEY environment variable is required');
+}
+
+// For debugging (remove in production)
+console.log('API Key present:', !!process.env.OPENAI_API_KEY);
+```
+
+### Environment Variable Loading
+
+**Problem:** Environment variables not loading correctly
+
+**Solution:** Use proper loading mechanisms:
+
+```typescript
+// For Node.js
+import 'dotenv/config';
+
+// For Next.js - use .env.local
+// Variables are automatically loaded
+
+// For Cloudflare Workers
+export default {
+  async fetch(request: Request, env: Env): Promise<Response> {
+    // Use env parameter, not process.env
+    const apiKey = env.OPENAI_API_KEY;
+    // ...
+  },
+};
+```
+
+## Common Error Messages
+
+### "Cannot find module 'llamaindex'"
+
+**Cause:** Package not installed or module resolution issue
+
+**Solution:**
+```bash
+npm install llamaindex
+```
+
+### "Module not found: Can't resolve 'fs'"
+
+**Cause:** File system modules used in browser/edge environment
+
+**Solution:**
+```typescript
+// Use dynamic imports with fallbacks
+const loadDocuments = async () => {
+  if (typeof window !== 'undefined') {
+    // Browser environment - use alternative
+    return await loadDocumentsFromAPI();
+  } else {
+    // Node.js environment - use file system
+    const { SimpleDirectoryReader } = await import('llamaindex');
+    return await new SimpleDirectoryReader('data').loadData();
+  }
+};
+```
+
+### "ReferenceError: global is not defined"
+
+**Cause:** Global polyfill missing in browser environments
+
+**Solution:**
+```typescript
+// Add to your app entry point
+if (typeof global === 'undefined') {
+  global = globalThis;
+}
+```
+
+### "Cannot read properties of undefined (reading 'query')"
+
+**Cause:** Query engine not properly initialized
+
+**Solution:**
+```typescript
+// Always check initialization
+if (!queryEngine) {
+  throw new Error('Query engine not initialized');
+}
+
+// Or use optional chaining
+const response = await queryEngine?.query(question);
+```
+
+## Debugging Tips
+
+### Enable Debug Logging
+
+```typescript
+// Enable debug logging
+process.env.DEBUG = "llamaindex:*";
+
+// Or specific modules
+process.env.DEBUG = "llamaindex:vector-store";
+```
+
+### Check Package Versions
+
+```bash
+npm list llamaindex
+npm list @llamaindex/openai
+```
+
+### Test in Isolation
+
+```typescript
+// Create minimal test case
+import { VectorStoreIndex } from 'llamaindex';
+
+async function testBasic() {
+  try {
+    console.log('Testing basic import...');
+    const index = new VectorStoreIndex();
+    console.log('Success!');
+  } catch (error) {
+    console.error('Error:', error);
+  }
+}
+
+testBasic();
+```
+
+## Getting Help
+
+### Before Asking for Help
+
+1. **Check this troubleshooting guide**
+2. **Search existing GitHub issues**
+3. **Try minimal reproduction**
+4. **Check your environment configuration**
+
+### When Reporting Issues
+
+Include:
+- Node.js version (`node --version`)
+- Package versions (`npm list llamaindex`)
+- Environment (Node.js, Cloudflare Workers, Vercel, etc.)
+- Minimal code reproduction
+- Full error message and stack trace
+
+### Useful Resources
+
+- [GitHub Issues](https://github.com/run-llama/LlamaIndexTS/issues)
+- [Discord Community](https://discord.gg/dGcwcsnxhU)
+- [Documentation](https://docs.llamaindex.ai/)
+
+## Next Steps
+
+If you're still experiencing issues:
+
+1. **Check specific deployment guides:**
+   - [Server APIs](/docs/llamaindex/getting_started/installation/server-apis)
+   - [Serverless Functions](/docs/llamaindex/getting_started/installation/serverless)
+   - [Next.js Applications](/docs/llamaindex/getting_started/installation/nextjs)
+
+2. **Open an issue** on GitHub with a minimal reproduction
+
+3. **Join our Discord** for community support 
@@ -1,99 +0,0 @@
---
-title: With TypeScript
-description: In this guide, you'll learn how to use LlamaIndex with TypeScript
---
-
-LlamaIndex.TS is written in TypeScript and designed to be used in TypeScript projects.
-
-We put a lot of work on strong typing to make sure you have a great typing experience with code completion such as:
-
-```ts twoslash
-import { PromptTemplate } from 'llamaindex'
-const promptTemplate = new PromptTemplate({
-  template: `Context information from multiple sources is below.
---------------------
-{context}
---------------------
-Given the information from multiple sources and not prior knowledge.
-Answer the query in the style of a Shakespeare play"
-Query: {query}
-Answer:`,
-	templateVars: ["context", "query"],
-});
-// @noErrors
-promptTemplate.format({
-	c
-//^|
-})
-```
-
-## Enable TypeScript
-
-Make sure to set [moduleResolution](https://www.typescriptlang.org/docs/handbook/modules/theory.html#module-resolution) in your `tsconfig.json` file:
-
-```json5
-{
-  compilerOptions: {
-    // ⬇️ add this line to your tsconfig.json
-    moduleResolution: "bundler", // or "nodenext" | "node16" | "node"
-  },
-}
-```
-
-We recommend using `bundler` or `nodenext`, but due to popularity of `node`, we still added support for it.
-
-## Enable AsyncIterable for `Web Stream` API
-
-Some modules uses `Web Stream` API like `ReadableStream` and `WritableStream`, you need to enable `DOM.AsyncIterable` in your `tsconfig.json`.
-
-```json5
-{
-  compilerOptions: {
-    // ⬇️ add this lib to your tsconfig.json
-    lib: ["DOM.AsyncIterable"],
-  },
-}
-```
-
-```typescript
-import { tool } from 'llamaindex'
-import { agent } from "@llamaindex/workflow";
-import { openai } from "@llamaindex/openai";
-
-Settings.llm = openai({
-  model: "gpt-4o-mini",
-});
-
-const addTool = tool({
-  name: "add", 
-  description: "Adds two numbers",
-  parameters: z.object({x: z.number(), y: z.number()}),
-  execute: ({ x, y }) => x + y,
-});
-
-const myAgent = agent({
-  tools: [addTool],
-});
-
-// Chat with the agent
-const context = myAgent.run("Hello, how are you?");
-
-for await (const event of context) {
-  if (event instanceof AgentStream) {
-    for (const chunk of event.data.delta) {
-      process.stdout.write(chunk); // stream response
-    }
-  } else {
-    console.log(event); // other events
-  }
-}
-
-```
-
-## Run TypeScript Script in Node.js
-
-We recommend to use [tsx](https://www.npmjs.com/package/tsx) to run TypeScript script in Node.js.
-
-```shell
-node --import tsx ./my-script.ts
-```
@@ -1,23 +0,0 @@
---
-title: With Vite
-description: In this guide, you'll learn how to use LlamaIndex with Vite
---
-
-Before you start, make sure you have try LlamaIndex.TS in Node.js to make sure you understand the basics.
-
-<Card
-  title="Getting Started with LlamaIndex.TS in Node.js"
-  href="/docs/llamaindex/getting_started/installation/node"
-/>
-
-Also, make sure you have a basic understanding of [Vite](https://vitejs.dev/).
-
-## Why mention Vite?
-
-Vite.js is widely used in building many web applications, like React.js, even for some native app like [Electron](https://www.electronjs.org/).
-
-However, it's not a ready-to-use solution for a Node.js-like application using Vite, as Vite is designed for web applications(run in browser).
-
-There's some plugin/framework based on Vite, like [Waku.gg](https://github.com/dai-shi/waku), or [Electron Vite](https://electron-vite.org/)
-
-For now, we have no clear solution for bundling LlamaIndex.TS with Vite, if you have any idea/solution, please let us know.
@@ -1,21 +1,118 @@
 ---
-title: What is LlamaIndex.TS
-description: LlamaIndex is the leading data framework for building LLM applications
+title: Welcome to LlamaIndex.TS
+description: LlamaIndex.TS is the leading framework for utilizing context engineering to build LLM applications in JavaScript and TypeScript.
 ---

-LlamaIndex is a framework for building context-augmented generative AI applications with LLMs including agents and workflows.
+LlamaIndex.TS is a **framework for utilizing context engineering to build generative AI applications** with large language models. From rapid-prototyping RAG chatbots to deploying multi-agent workflows in production, LlamaIndex gives you everything you need — all in idiomatic TypeScript.

-The TypeScript implementation is designed for JavaScript server side applications using <SiNodedotjs className="inline" color="#5FA04E" /> Node.js, <SiDeno className="inline" color="#70FFAF" /> Deno, <SiBun className="inline" /> Bun, <SiCloudflareworkers className="inline" color="#F38020" /> Cloudflare Workers, and more.
+Built for modern JavaScript runtimes like <SiNodedotjs className="inline" color="#5FA04E" /> **Node.js**, <SiDeno className="inline" color="#70FFAF" /> **Deno**, <SiBun className="inline" /> **Bun**, <SiCloudflareworkers className="inline" color="#F38020" /> **Cloudflare Workers**, and more.

-LlamaIndex.TS provides tools for beginners, advanced users, and everyone in between.
+<div className="grid grid-cols-1 gap-4 sm:grid-cols-2 lg:grid-cols-3 my-6">
+  <a href="#introduction" className="block rounded-lg border border-gray-600/40 p-4 hover:border-gray-400 hover:bg-gray-700/20 no-underline">
+    <h3 className="mb-1 text-lg font-semibold underline">Introduction</h3>
+    <p className="text-sm text-gray-400 no-underline">Context engineering, agents &amp; workflows — what do they mean?</p>
+  </a>

-Try it out with a starter example using StackBlitz:
+  <a href="#use-cases" className="block rounded-lg border border-gray-600/40 p-4 hover:border-gray-400 hover:bg-gray-700/20 no-underline">
+    <h3 className="mb-1 text-lg font-semibold underline">Use cases</h3>
+    <p className="text-sm text-gray-400 no-underline">See what you can build with LlamaIndex.TS.</p>
+  </a>
+
+  <a href="#getting-started" className="block rounded-lg border border-gray-600/40 p-4 hover:border-gray-400 hover:bg-gray-700/20 no-underline">
+    <h3 className="mb-1 text-lg font-semibold underline">Getting started</h3>
+    <p className="text-sm text-gray-400 no-underline">Your first app in 5 lines of code.</p>
+  </a>
+
+  <a href="https://docs.cloud.llamaindex.ai/" className="block rounded-lg border border-gray-600/40 p-4 hover:border-gray-400 hover:bg-gray-700/20 no-underline" target="_blank" rel="noopener noreferrer">
+    <h3 className="mb-1 text-lg font-semibold underline">LlamaCloud</h3>
+    <p className="text-sm text-gray-400 no-underline">Managed parsing, extraction &amp; retrieval pipelines.</p>
+  </a>
+
+  <a href="#community" className="block rounded-lg border border-gray-600/40 p-4 hover:border-gray-400 hover:bg-gray-700/20 no-underline">
+    <h3 className="mb-1 text-lg font-semibold underline">Community</h3>
+    <p className="text-sm text-gray-400 no-underline">Join thousands of builders on Discord, Twitter, and more.</p>
+  </a>
+
+  <a href="#related-projects" className="block rounded-lg border border-gray-600/40 p-4 hover:border-gray-400 hover:bg-gray-700/20 no-underline">
+    <h3 className="mb-1 text-lg font-semibold underline">Related projects</h3>
+    <p className="text-sm text-gray-400 no-underline">Connectors, demos &amp; starter kits.</p>
+  </a>
+</div>
+
+## Introduction
+
+### What are agents?
+
+[Agents](/docs/llamaindex/tutorials/agents/1_setup) are LLM-powered assistants that can reason, use external tools, and take actions to accomplish tasks such as research, data extraction, and automation. 
+LlamaIndex.TS provides foundational building blocks for creating and orchestrating these agents.
+
+### What are workflows?
+
+[Workflows](/docs/llamaindex/tutorials/workflows) are multi-step, event-driven processes that combine agents, data connectors, and other tools to solve complex problems. 
+With LlamaIndex.TS you can chain together retrieval, generation, and tool-calling steps and then deploy the entire pipeline as a microservice.
+
+### What is context engineering?
+
+LLMs come pre-trained on vast public corpora, but not on **your** private or domain-specific data. 
+Context engineering bridges that gap by injecting the right pieces of your data into the LLM prompt at the right time. 
+The most popular example is [Retrieval-Augmented Generation (RAG)](/docs/llamaindex/getting_started/concepts), but the same idea powers agent memory, evaluation, extraction, summarisation, and more.
+
+LlamaIndex.TS gives you:
+
+- **Data connectors** to ingest from APIs, files, SQL, and dozens more sources.
+- **Indexes & retrievers** to store and retrieve your data for LLM consumption.
+- **Agents and Engines** to query and use chat+reasoning interfaces over your data.
+- **Workflows** for fine-grained orchestration of your data and LLM-powered agents.
+- **Observability** integrations so you can iterate with confidence.
+
+You can learn more about these concepts in our [concepts guide](/docs/llamaindex/getting_started/concepts).
+
+## Use cases
+
+Popular scenarios include:
+
+- [LLM-Powered Agents](/docs/llamaindex/tutorials/agents/1_setup)
+- [Indexing and Retrieval](/docs/llamaindex/tutorials/rag)
+- [Extracting Structured Data](/docs/llamaindex/tutorials/structured_data_extraction)
+- [Custom Orchestration with Workflows](/docs/llamaindex/tutorials/workflows)
+
+## Getting started
+
+The fastest way to get started is in StackBlitz below — no local setup required:

 <iframe
  className="w-full h-[440px]"
  aria-label="LlamaIndex.TS Starter"
-  aria-description="This is a starter example for LlamaIndex.TS, it shows the basic usage of the library."
+  aria-description="Interactive starter for LlamaIndex.TS"
  src="https://stackblitz.com/github/run-llama/LlamaIndexTS/tree/main/examples?embed=1&file=starter.ts"
 />

-You'll need an OpenAI API key to run this example. You can retrieve it from [OpenAI](https://platform.openai.com/api-keys).
+Want to learn more? We have several tutorials to get you started:
+
+- [Installation + Runtime Guide](/docs/llamaindex/getting_started/installation)
+- [Create your first agent](/docs/llamaindex/tutorials/agents/1_setup)
+- [Learn how to index data and chat with it](/docs/llamaindex/tutorials/rag)
+- [Learn how to write your own workflows and agents](/docs/llamaindex/tutorials/workflows)
+
+---
+
+## LlamaCloud
+
+Need an end-to-end managed pipeline? Check out **[LlamaCloud](https://cloud.llamaindex.ai/)**: best-in-class document parsing (LlamaParse), extraction (LlamaExtract), and indexing services with generous free tiers.
+
+---
+
+## Community
+
+- [Twitter](https://twitter.com/llama_index)
+- [Discord](https://discord.gg/dGcwcsnxhU)
+- [LinkedIn](https://www.linkedin.com/company/llamaindex/)
+
+We 💜 contributors! View our [contributing guide](https://github.com/run-llama/LlamaIndexTS/blob/main/CONTRIBUTING.md) to get started.
+
+## Related projects
+
+- [Python framework GitHub](https://github.com/run-llama/llama_index)
+- [Python docs](https://docs.llamaindex.ai/)
+- [create-llama](https://www.npmjs.com/package/create-llama) — scaffold a new project in seconds!
+- [UI Components](https://ui.llamaindex.ai/) — build chat applications with our Next.js components.
@@ -34,6 +34,7 @@ const jokeAgent = agent({
 // Run the workflow
 const result = await jokeAgent.run("Tell me something funny");
 console.log(result.data.result); // Baby Llama is called cria
+console.log(result.data.message); // { role: 'assistant', content: 'Baby Llama is called cria' }
 ```

 ### Event Streaming
@@ -0,0 +1,164 @@
+---
+title: Low-Level LLM Execution
+---
+
+Sometimes your need more control over LLM interactions than what high-level agents provide. The `llm.exec` method makes it simple for you to make a single LLM call with tools but hides the complexity of executing the tools and generating the tool messages.
+
+## When to Use `llm.exec`
+
+Use `llm.exec` when you need to:
+- Build custom agent logic in [workflow](/docs/llamaindex/modules/agents/workflows) steps
+- Have precise control over message handling and tool execution
+
+## Basic Usage
+
+The `llm.exec` method takes messages and tools as parameter and executes one LLM call.
+The LLM might either request to call one or more of the tools or generate an assistant message as result.
+For each tool call that is requested, `llm.exec` executes it and generates the two tool call messages (call and result). If no tool call is requested, just the assistant message is returned. 
+
+```ts
+import { openai } from "@llamaindex/openai";
+import { ChatMessage, tool } from "llamaindex";
+import z from "zod";
+
+const llm = openai({ model: "gpt-4.1-mini" });
+const messages = [
+  {
+    content: "What's the weather like in San Francisco?",
+    role: "user",
+  } as ChatMessage,
+];
+
+const { newMessages, toolCalls } = await llm.exec({
+  messages,
+  tools: [
+    tool({
+      name: "get_weather",
+      description: "Get the current weather for a location",
+      parameters: z.object({
+        address: z.string().describe("The address"),
+      }),
+      execute: ({ address }) => {
+        return `It's sunny in ${address}!`;
+      },
+    }),
+  ],
+});
+
+// Add the new messages (including tool calls and responses) to your conversation
+messages.push(...newMessages);
+```
+
+> `newMessages` is an array as each tool call generates two messages: a tool call message and the tool call result message.
+
+## Agent Loop Pattern
+
+A common pattern is to use `llm.exec` in a loop until the LLM stops making tool calls:
+
+```ts
+import { openai } from "@llamaindex/openai";
+import { ChatMessage, tool } from "llamaindex";
+import z from "zod";
+
+async function runAgentLoop() {
+  const llm = openai({ model: "gpt-4.1-mini" });
+  const messages = [
+    {
+      content: "What's the weather like in San Francisco?",
+      role: "user",
+    } as ChatMessage,
+  ];
+
+  let exit = false;
+  do {
+    const { newMessages, toolCalls } = await llm.exec({
+      messages,
+      tools: [
+        tool({
+          name: "get_weather",
+          description: "Get the current weather for a location",
+          parameters: z.object({
+            address: z.string().describe("The address"),
+          }),
+          execute: ({ address }) => {
+            return `It's sunny in ${address}!`;
+          },
+        }),
+      ],
+    });
+    
+    console.log(newMessages);
+    messages.push(...newMessages);
+    
+    // Exit when no more tool calls are made
+    exit = toolCalls.length === 0;
+  } while (!exit);
+}
+```
+
+## Streaming Support
+
+For real-time responses, use the `stream` option to get the assistant's response as streamed tokens:
+
+```ts
+import { openai } from "@llamaindex/openai";
+import { tool } from "llamaindex";
+import z from "zod";
+
+async function streamingAgentLoop() {
+  const llm = openai({ model: "gpt-4o-mini" });
+  const messages = [
+    {
+      content: "What's the weather like in San Francisco?",
+      role: "user",
+    } as ChatMessage,
+  ];
+
+  let exit = false;
+  do {
+    const { stream, newMessages, toolCalls } = await llm.exec({
+      messages,
+      tools: [
+        tool({
+          name: "get_weather",
+          description: "Get the current weather for a location",
+          parameters: z.object({
+            address: z.string().describe("The address"),
+          }),
+          execute: ({ address }) => {
+            return `It's sunny in ${address}!`;
+          },
+        }),
+      ],
+      stream: true,
+    });
+    
+    // Stream the response token by token
+    for await (const chunk of stream) {
+      process.stdout.write(chunk.delta);
+    }
+    
+    messages.push(...newMessages());
+    
+    exit = toolCalls.length === 0;
+  } while (!exit);
+}
+```
+
+> `newMessages` is a function when streaming. The reason is that the result only is available after streaming. Calling it before, will throw an error.
+
+## Return Values
+
+`llm.exec` returns an object with:
+
+- **`newMessages`**: Array of new chat messages including the LLM response and any tool call messages (call or result). This is a function return the array when streaming.
+- **`toolCalls`**: Array of tool calls made by the LLM
+- **`stream`**: Async iterable for streaming responses (only when `stream: true`)
+
+## Best Practices
+
+For using `llm.exec` in an agent loop, take care to:
+
+1. **Maintain message history**: Always add `newMessages` to your conversation history
+2. **Set exit conditions**: Implement proper logic to avoid infinite loops
+
@@ -1,4 +1,10 @@
 {
  "title": "Agents",
-  "pages": ["tool", "agent_workflow", "workflows", "natural_language_workflow"]
+  "pages": [
+    "tool",
+    "agent_workflow",
+    "workflows",
+    "low-level",
+    "natural_language_workflow"
+  ]
 }
@@ -106,34 +106,40 @@ const memory = createMemory({

 Long-term memory is represented as `Memory Block` objects. These objects contain information that are from previous user sessions or from the beginning of the current conversation. When memory is retrieved (by calling `getLLM`), the short-term and long-term memories are merged together within the given `tokenLimit`. 

-Currently, there are two predefined memory blocks:
+Currently, there are three predefined memory blocks:

 - `staticBlock`: A memory block that stores a static piece of information.
 - `factExtractionBlock`: A memory block that extracts facts from the chat history.
+- `vectorBlock`: A memory block that stores and retrieves chat messages from a vector database using semantic similarity search. Messages are stored individually and retrieved based on their relevance to recent conversation context. Here we've passed in the `vectorStore` to use to store and retrieve the chat messages.

 This sounds a bit complicated, but it's actually quite simple. Let's look at an example:

 ```ts
-import { createMemory, factExtractionBlock, staticBlock } from "llamaindex";
+import { createMemory, factExtractionBlock, staticBlock, vectorBlock } from "llamaindex";
+import { QdrantVectorStore } from "@llamaindex/qdrant";
+import { OpenAIEmbedding } from "@llamaindex/openai";

 const memoryBlocks= [
  staticBlock({
-    id: "core_info",
    content: "My name is Logan, and I live in Saskatoon. I work at LlamaIndex.",
  }),
  factExtractionBlock({
-    id: "user-extracted_info",
    priority: 1,
    llm: llm,
    maxFacts: 50,
  }),
+  vectorBlock({
+    vectorStore: new QdrantVectorStore({ url: "http://localhost:6333" }),
+    priority: 2,
+  }),
 ];
 ```

-Here, we've setup two memory blocks:
+Here, we've setup three memory blocks:

- `core_info`: A static memory block that stores some core information about the user. This information will always be inserted into the memory. The type used is `MessageContent` to support multi-modal content.
- `extracted_info`: An extracted memory block that will extract information from the chat history. Here we've passed in the `llm` to use to extract facts from the chat history, and set the `maxFacts` to 50. If the number of extracted facts exceeds this limit, the `maxFacts` will be automatically summarized and reduced to leave room for new information.
+- `staticBlock`: A static memory block that stores some core information about the user. This information will always be inserted into the memory. The type used is `MessageContent` to support multi-modal content.
+- `factExtractionBlock`: An extracted memory block that will extract information from the chat history. Here we've passed in the `llm` to use to extract facts from the chat history, and set the `maxFacts` to 50. If the number of extracted facts exceeds this limit, the `maxFacts` will be automatically summarized and reduced to leave room for new information.
+- `vectorBlock`: A vector memory block that will store in a vector database and retrieve them from there. Messages are stored individually and retrieved based on their relevance to recent conversation context. Here we've passed in the `vectorStore` to use to store and retrieve the chat messages.

 You'll also notice that we've set the `priority` for the `factExtractionBlock` block. This is used to determine the handling when the memory blocks content (i.e. long-term memory) + short-term memory exceeds the token limit on the `Memory` object.

@@ -158,6 +164,46 @@ When memory is retrieved (using `getLLM`), the short-term and long-term memories

 The amount of short-term memory included is specified by the `shortTermTokenLimitRatio`. If it's set to `0.7`, 70% of the `tokenLimit` is used for short-term memory (not including the static memory block).

+
+#### VectorBlock Configuration Options
+
+The `vectorBlock` offers several configuration options to customize its behavior:
+
+```ts
+vectorBlock({
+  vectorStore: new QdrantVectorStore({ url: "http://localhost:6333" }),
+  priority: 2,
+  retrievalContextWindow: 5, // Number of recent messages to use for context when retrieving
+  formatTemplate: new PromptTemplate({ template: "Context: {{ context }}" }), // Custom formatting template
+  nodePostprocessors: [/* custom postprocessors */], // Apply processing to retrieved nodes
+  queryOptions: {
+    similarityTopK: 3, // Number of top similar results to return (default: 2)
+    mode: VectorStoreQueryMode.DEFAULT, // Query mode for the vector store
+    sessionFilterKey: "session_id", // Metadata key for session filtering (default: "session_id")
+    // Custom filters can be added here - session filter is automatically included
+    filters: {
+      filters: [
+        { key: "custom_field", value: "custom_value", operator: "==" }
+      ],
+      condition: "and"
+    }
+  }
+})
+```
+
+**Key Configuration Options:**
+
+- **`retrievalContextWindow`**: Number of recent messages to consider when creating the retrieval query (default: 5). A larger window provides more context but may be less precise.
+- **`formatTemplate`**: Template for formatting retrieved information before adding to memory. Defaults to a simple context template.
+- **`nodePostprocessors`**: Array of postprocessors to apply to retrieved nodes, useful for filtering or transforming results.
+- **`queryOptions.similarityTopK`**: Number of most similar messages to retrieve from the vector store (default: 2).
+- **`queryOptions.sessionFilterKey`**: Metadata key used to isolate memory between different sessions (default: "session_id").
+- **`queryOptions.filters`**: Additional metadata filters for retrieval. The session filter is automatically added to ensure memory isolation.
+
+**Session Isolation:**
+
+The vectorBlock automatically adds a session filter using the block's ID to ensure that memories from different sessions don't interfere with each other. This filter uses the `sessionFilterKey` (default: "session_id") and can be customized if needed.
+
 ## Persistence with Snapshots

 Save and restore memory state:
@@ -5,13 +5,13 @@ title: Bedrock
 ## Installation

 ```package-install
-npm i llamaindex @llamaindex/community
+npm i llamaindex @llamaindex/aws
 ```

 ## Usage

 ```ts
-import { BEDROCK_MODELS, Bedrock } from "@llamaindex/community";
+import { BEDROCK_MODELS, Bedrock } from "@llamaindex/aws";

 Settings.llm = new Bedrock({
  model: BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_HAIKU,
@@ -23,9 +23,19 @@ Settings.llm = new Bedrock({
 });
 ```

-Currently only supports Anthropic and Meta models:
+Supported models are listed below (accessible by BEDROCK_MODELS).

 ```ts
+AMAZON_TITAN_TG1_LARGE = "amazon.titan-tg1-large";
+AMAZON_TITAN_TEXT_EXPRESS_V1 = "amazon.titan-text-express-v1";
+AI21_J2_GRANDE_INSTRUCT = "ai21.j2-grande-instruct";
+AI21_J2_JUMBO_INSTRUCT = "ai21.j2-jumbo-instruct";
+AI21_J2_MID = "ai21.j2-mid";
+AI21_J2_MID_V1 = "ai21.j2-mid-v1";
+AI21_J2_ULTRA = "ai21.j2-ultra";
+AI21_J2_ULTRA_V1 = "ai21.j2-ultra-v1";
+COHERE_COMMAND_TEXT_V14 = "cohere.command-text-v14";
+
 ANTHROPIC_CLAUDE_INSTANT_1 = "anthropic.claude-instant-v1";
 ANTHROPIC_CLAUDE_2 = "anthropic.claude-v2";
 ANTHROPIC_CLAUDE_2_1 = "anthropic.claude-v2:1";
@@ -33,7 +43,12 @@ ANTHROPIC_CLAUDE_3_SONNET = "anthropic.claude-3-sonnet-20240229-v1:0";
 ANTHROPIC_CLAUDE_3_HAIKU = "anthropic.claude-3-haiku-20240307-v1:0";
 ANTHROPIC_CLAUDE_3_OPUS = "anthropic.claude-3-opus-20240229-v1:0"; // available on us-west-2
 ANTHROPIC_CLAUDE_3_5_SONNET = "anthropic.claude-3-5-sonnet-20240620-v1:0";
+ANTHROPIC_CLAUDE_3_5_SONNET_V2 = "anthropic.claude-3-5-sonnet-20241022-v2:0";
 ANTHROPIC_CLAUDE_3_5_HAIKU = "anthropic.claude-3-5-haiku-20241022-v1:0";
+ANTHROPIC_CLAUDE_3_7_SONNET = "anthropic.claude-3-7-sonnet-20250219-v1:0";
+ANTHROPIC_CLAUDE_4_SONNET = "anthropic.claude-sonnet-4-20250514-v1:0";
+ANTHROPIC_CLAUDE_4_OPUS = "anthropic.claude-opus-4-20250514-v1:0";
+
 META_LLAMA2_13B_CHAT = "meta.llama2-13b-chat-v1";
 META_LLAMA2_70B_CHAT = "meta.llama2-70b-chat-v1";
 META_LLAMA3_8B_INSTRUCT = "meta.llama3-8b-instruct-v1:0";
@@ -45,41 +60,66 @@ META_LLAMA3_2_1B_INSTRUCT = "meta.llama3-2-1b-instruct-v1:0"; // only available
 META_LLAMA3_2_3B_INSTRUCT = "meta.llama3-2-3b-instruct-v1:0"; // only available via inference endpoints (see below)
 META_LLAMA3_2_11B_INSTRUCT = "meta.llama3-2-11b-instruct-v1:0"; // only available via inference endpoints (see below), multimodal and function call supported
 META_LLAMA3_2_90B_INSTRUCT = "meta.llama3-2-90b-instruct-v1:0"; // only available via inference endpoints (see below), multimodal and function call supported
+META_LLAMA3_3_70B_INSTRUCT = "meta.llama3-3-70b-instruct-v1:0";
+
+MISTRAL_7B_INSTRUCT = "mistral.mistral-7b-instruct-v0:2";
+MISTRAL_MIXTRAL_7B_INSTRUCT = "mistral.mixtral-8x7b-instruct-v0:1";
+MISTRAL_MIXTRAL_LARGE_2402 = "mistral.mistral-large-2402-v1:0";
+
 AMAZON_NOVA_PREMIER_1 = "amazon.nova-premier-v1:0";
 AMAZON_NOVA_PRO_1 = "amazon.nova-pro-v1:0";
 AMAZON_NOVA_LITE_1 = "amazon.nova-lite-v1:0";
 AMAZON_NOVA_MICRO_1 = "amazon.nova-micro-v1:0";
 ```

-You can also use Bedrock's Inference endpoints by using the model names:
+You can also use Bedrock's Inference endpoints by using the model names (accessible by INFERENCE_BEDROCK_MODELS).
+Note that the region must be set correctly.

 ```ts
-// US
+//US
 US_ANTHROPIC_CLAUDE_3_HAIKU = "us.anthropic.claude-3-haiku-20240307-v1:0";
+US_ANTHROPIC_CLAUDE_3_5_HAIKU = "us.anthropic.claude-3-5-haiku-20241022-v1:0";
 US_ANTHROPIC_CLAUDE_3_OPUS = "us.anthropic.claude-3-opus-20240229-v1:0";
 US_ANTHROPIC_CLAUDE_3_SONNET = "us.anthropic.claude-3-sonnet-20240229-v1:0";
 US_ANTHROPIC_CLAUDE_3_5_SONNET = "us.anthropic.claude-3-5-sonnet-20240620-v1:0";
-US_ANTHROPIC_CLAUDE_3_5_SONNET_V2 =
-  "us.anthropic.claude-3-5-sonnet-20241022-v2:0";
+US_ANTHROPIC_CLAUDE_3_5_SONNET_V2 = "us.anthropic.claude-3-5-sonnet-20241022-v2:0";
+US_ANTHROPIC_CLAUDE_3_7_SONNET = "us.anthropic.claude-3-7-sonnet-20250219-v1:0";
+US_ANTHROPIC_CLAUDE_4_SONNET = "us.anthropic.claude-sonnet-4-20250514-v1:0";
+US_ANTHROPIC_CLAUDE_4_OPUS = "us.anthropic.claude-opus-4-20250514-v1:0";
 US_META_LLAMA_3_2_1B_INSTRUCT = "us.meta.llama3-2-1b-instruct-v1:0";
 US_META_LLAMA_3_2_3B_INSTRUCT = "us.meta.llama3-2-3b-instruct-v1:0";
 US_META_LLAMA_3_2_11B_INSTRUCT = "us.meta.llama3-2-11b-instruct-v1:0";
 US_META_LLAMA_3_2_90B_INSTRUCT = "us.meta.llama3-2-90b-instruct-v1:0";
-US_AMAZON_NOVA_PRO_1 = "us.amazon.nova-premier-v1:0";
+US_META_LLAMA_3_3_70B_INSTRUCT = "us.meta.llama3-3-70b-instruct-v1:0";
+US_AMAZON_NOVA_PREMIER_1 = "us.amazon.nova-premier-v1:0";
 US_AMAZON_NOVA_PRO_1 = "us.amazon.nova-pro-v1:0";
 US_AMAZON_NOVA_LITE_1 = "us.amazon.nova-lite-v1:0";
 US_AMAZON_NOVA_MICRO_1 = "us.amazon.nova-micro-v1:0";

-// EU
+//EU
 EU_ANTHROPIC_CLAUDE_3_HAIKU = "eu.anthropic.claude-3-haiku-20240307-v1:0";
+EU_ANTHROPIC_CLAUDE_3_5_HAIKU = "eu.anthropic.claude-3-5-haiku-20240307-v1:0";
 EU_ANTHROPIC_CLAUDE_3_SONNET = "eu.anthropic.claude-3-sonnet-20240229-v1:0";
 EU_ANTHROPIC_CLAUDE_3_5_SONNET = "eu.anthropic.claude-3-5-sonnet-20240620-v1:0";
+EU_ANTHROPIC_CLAUDE_3_7_SONNET = "eu.anthropic.claude-3-7-sonnet-20250219-v1:0";
+EU_ANTHROPIC_CLAUDE_4_SONNET = "eu.anthropic.claude-sonnet-4-20250514-v1:0";
+EU_ANTHROPIC_CLAUDE_4_OPUS = "eu.anthropic.claude-opus-4-20250514-v1:0";
 EU_META_LLAMA_3_2_1B_INSTRUCT = "eu.meta.llama3-2-1b-instruct-v1:0";
 EU_META_LLAMA_3_2_3B_INSTRUCT = "eu.meta.llama3-2-3b-instruct-v1:0";
-EU_AMAZON_NOVA_PRO_1 = "eu.amazon.nova-premier-v1:0";
+EU_AMAZON_NOVA_PREMIER_1 = "eu.amazon.nova-premier-v1:0";
 EU_AMAZON_NOVA_PRO_1 = "eu.amazon.nova-pro-v1:0";
 EU_AMAZON_NOVA_LITE_1 = "eu.amazon.nova-lite-v1:0";
 EU_AMAZON_NOVA_MICRO_1 = "eu.amazon.nova-micro-v1:0";
+
+//APAC
+APAC_ANTHROPIC_CLAUDE_3_5_SONNET = "apac.anthropic.claude-3-5-sonnet-20240620-v1:0";
+APAC_ANTHROPIC_CLAUDE_3_5_SONNET_V2 = "apac.anthropic.claude-3-5-sonnet-20241022-v2:0";
+APAC_ANTHROPIC_CLAUDE_3_7_SONNET = "apac.anthropic.claude-3-7-sonnet-20250219-v1:0";
+APAC_ANTHROPIC_CLAUDE_3_HAIKU = "apac.anthropic.claude-3-haiku-20240307-v1:0";
+APAC_ANTHROPIC_CLAUDE_3_SONNET = "apac.anthropic.claude-3-sonnet-20240229-v1:0";
+APAC_AMAZON_NOVA_PRO_1 = "apac.amazon.nova-pro-v1:0";
+APAC_AMAZON_NOVA_LITE_1 = "apac.amazon.nova-lite-v1:0";
+APAC_AMAZON_NOVA_MICRO_1 = "apac.amazon.nova-micro-v1:0";
 ```

 Sonnet, Haiku and Opus are multimodal, image_url only supports base64 data url format, e.g. `data:image/jpeg;base64,SGVsbG8sIFdvcmxkIQ==`
@@ -87,10 +127,11 @@ Sonnet, Haiku and Opus are multimodal, image_url only supports base64 data url f
 ## Full Example

 ```ts
-import { BEDROCK_MODELS, Bedrock } from "llamaindex";
+import { INFERENCE_BEDROCK_MODELS, Bedrock } from "@llamaindex/aws";

 Settings.llm = new Bedrock({
-  model: BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_HAIKU,
+  model: INFERENCE_BEDROCK_MODELS.US_ANTHROPIC_CLAUDE_3_SONNET,
+  region: "us-east-1",
 });

 async function main() {
@@ -119,7 +160,7 @@ async function main() {
 ## Agent Example

 ```ts
-import { BEDROCK_MODELS, Bedrock } from "@llamaindex/community";
+import { BEDROCK_MODELS, Bedrock } from "@llamaindex/aws";
 import { tool } from "llamaindex";
 import { agent } from "@llamaindex/workflow";
 import { z } from "zod";
@@ -38,10 +38,13 @@ You should expect output something like:
 {
  result: '5 + 5 is 10. Then, 10 divided by 2 is 5.',
  state: {
-    memory: ChatMemoryBuffer {
-      chatStore: SimpleChatStore {},
-      chatStoreKey: 'chat_history',
-      tokenLimit: 750000
+    memory: Memory {
+      messages: [Array],
+      tokenLimit: 30000,
+      shortTermTokenLimitRatio: 0.7,
+      memoryBlocks: [],
+      memoryCursor: 0,
+      adapters: [Object]
    },
    scratchpad: [],
    currentAgentName: 'Agent',
@@ -1,5 +1,47 @@
 # @llamaindex/cloudflare-worker-agent-test

+## 0.0.184
+
+### Patch Changes
+
+- llamaindex@0.11.23
+
+## 0.0.183
+
+### Patch Changes
+
+- llamaindex@0.11.22
+
+## 0.0.182
+
+### Patch Changes
+
+- llamaindex@0.11.21
+
+## 0.0.181
+
+### Patch Changes
+
+- llamaindex@0.11.20
+
+## 0.0.180
+
+### Patch Changes
+
+- llamaindex@0.11.19
+
+## 0.0.179
+
+### Patch Changes
+
+- llamaindex@0.11.18
+
+## 0.0.178
+
+### Patch Changes
+
+- llamaindex@0.11.17
+
 ## 0.0.177

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/cloudflare-worker-agent-test",
-  "version": "0.0.177",
+  "version": "0.0.184",
  "type": "module",
  "private": true,
  "scripts": {
@@ -1,5 +1,44 @@
 # @llamaindex/llama-parse-browser-test

+## 0.0.82
+
+### Patch Changes
+
+- @llamaindex/cloud@4.0.27
+
+## 0.0.81
+
+### Patch Changes
+
+- @llamaindex/cloud@4.0.26
+
+## 0.0.80
+
+### Patch Changes
+
+- Updated dependencies [2967d57]
+  - @llamaindex/cloud@4.0.25
+
+## 0.0.79
+
+### Patch Changes
+
+- @llamaindex/cloud@4.0.24
+
+## 0.0.78
+
+### Patch Changes
+
+- Updated dependencies [a1b1598]
+  - @llamaindex/cloud@4.0.23
+
+## 0.0.77
+
+### Patch Changes
+
+- Updated dependencies [d2be868]
+  - @llamaindex/cloud@4.0.22
+
 ## 0.0.76

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/llama-parse-browser-test",
  "private": true,
-  "version": "0.0.76",
+  "version": "0.0.82",
  "type": "module",
  "scripts": {
    "dev": "vite",
@@ -1,5 +1,47 @@
 # @llamaindex/next-agent-test

+## 0.1.184
+
+### Patch Changes
+
+- llamaindex@0.11.23
+
+## 0.1.183
+
+### Patch Changes
+
+- llamaindex@0.11.22
+
+## 0.1.182
+
+### Patch Changes
+
+- llamaindex@0.11.21
+
+## 0.1.181
+
+### Patch Changes
+
+- llamaindex@0.11.20
+
+## 0.1.180
+
+### Patch Changes
+
+- llamaindex@0.11.19
+
+## 0.1.179
+
+### Patch Changes
+
+- llamaindex@0.11.18
+
+## 0.1.178
+
+### Patch Changes
+
+- llamaindex@0.11.17
+
 ## 0.1.177

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/next-agent-test",
-  "version": "0.1.177",
+  "version": "0.1.184",
  "private": true,
  "scripts": {
    "dev": "next dev",
@@ -1,5 +1,47 @@
 # test-edge-runtime

+## 0.1.183
+
+### Patch Changes
+
+- llamaindex@0.11.23
+
+## 0.1.182
+
+### Patch Changes
+
+- llamaindex@0.11.22
+
+## 0.1.181
+
+### Patch Changes
+
+- llamaindex@0.11.21
+
+## 0.1.180
+
+### Patch Changes
+
+- llamaindex@0.11.20
+
+## 0.1.179
+
+### Patch Changes
+
+- llamaindex@0.11.19
+
+## 0.1.178
+
+### Patch Changes
+
+- llamaindex@0.11.18
+
+## 0.1.177
+
+### Patch Changes
+
+- llamaindex@0.11.17
+
 ## 0.1.176

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/nextjs-edge-runtime-test",
-  "version": "0.1.176",
+  "version": "0.1.183",
  "private": true,
  "scripts": {
    "dev": "next dev",
@@ -1,5 +1,61 @@
 # @llamaindex/next-node-runtime

+## 0.1.53
+
+### Patch Changes
+
+- llamaindex@0.11.23
+- @llamaindex/huggingface@0.1.23
+- @llamaindex/readers@3.1.17
+
+## 0.1.52
+
+### Patch Changes
+
+- llamaindex@0.11.22
+
+## 0.1.51
+
+### Patch Changes
+
+- llamaindex@0.11.21
+- @llamaindex/huggingface@0.1.22
+- @llamaindex/readers@3.1.16
+
+## 0.1.50
+
+### Patch Changes
+
+- llamaindex@0.11.20
+- @llamaindex/huggingface@0.1.21
+- @llamaindex/readers@3.1.15
+
+## 0.1.49
+
+### Patch Changes
+
+- @llamaindex/huggingface@0.1.20
+
+## 0.1.48
+
+### Patch Changes
+
+- llamaindex@0.11.19
+- @llamaindex/huggingface@0.1.19
+- @llamaindex/readers@3.1.14
+
+## 0.1.47
+
+### Patch Changes
+
+- llamaindex@0.11.18
+
+## 0.1.46
+
+### Patch Changes
+
+- llamaindex@0.11.17
+
 ## 0.1.45

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/next-node-runtime-test",
-  "version": "0.1.45",
+  "version": "0.1.53",
  "private": true,
  "scripts": {
    "dev": "next dev",
@@ -1,5 +1,47 @@
 # vite-import-llamaindex

+## 0.0.50
+
+### Patch Changes
+
+- llamaindex@0.11.23
+
+## 0.0.49
+
+### Patch Changes
+
+- llamaindex@0.11.22
+
+## 0.0.48
+
+### Patch Changes
+
+- llamaindex@0.11.21
+
+## 0.0.47
+
+### Patch Changes
+
+- llamaindex@0.11.20
+
+## 0.0.46
+
+### Patch Changes
+
+- llamaindex@0.11.19
+
+## 0.0.45
+
+### Patch Changes
+
+- llamaindex@0.11.18
+
+## 0.0.44
+
+### Patch Changes
+
+- llamaindex@0.11.17
+
 ## 0.0.43

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "vite-import-llamaindex",
  "private": true,
-  "version": "0.0.43",
+  "version": "0.0.50",
  "type": "module",
  "scripts": {
    "build": "vite build",
@@ -1,5 +1,47 @@
 # @llamaindex/waku-query-engine-test

+## 0.0.184
+
+### Patch Changes
+
+- llamaindex@0.11.23
+
+## 0.0.183
+
+### Patch Changes
+
+- llamaindex@0.11.22
+
+## 0.0.182
+
+### Patch Changes
+
+- llamaindex@0.11.21
+
+## 0.0.181
+
+### Patch Changes
+
+- llamaindex@0.11.20
+
+## 0.0.180
+
+### Patch Changes
+
+- llamaindex@0.11.19
+
+## 0.0.179
+
+### Patch Changes
+
+- llamaindex@0.11.18
+
+## 0.0.178
+
+### Patch Changes
+
+- llamaindex@0.11.17
+
 ## 0.0.177

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/waku-query-engine-test",
-  "version": "0.0.177",
+  "version": "0.0.184",
  "type": "module",
  "private": true,
  "scripts": {
@@ -23,7 +23,7 @@ await test("pinecone", async (t) => {
  });

  const vectorStore = new PineconeVectorStore({
-    embeddingModel: openaiEmbedding,
+    embedModel: openaiEmbedding,
  });

  t.after(async () => {
@@ -1,5 +1,240 @@
 # examples

+## 0.3.34
+
+### Patch Changes
+
+- Updated dependencies [f29799e]
+- Updated dependencies [7224c06]
+  - @llamaindex/workflow@1.1.19
+  - @llamaindex/core@0.6.18
+  - llamaindex@0.11.23
+  - @llamaindex/cloud@4.0.27
+  - @llamaindex/node-parser@2.0.18
+  - @llamaindex/anthropic@0.3.20
+  - @llamaindex/assemblyai@0.1.17
+  - @llamaindex/clip@0.0.69
+  - @llamaindex/cohere@0.0.32
+  - @llamaindex/deepinfra@0.0.69
+  - @llamaindex/discord@0.1.17
+  - @llamaindex/google@0.3.17
+  - @llamaindex/huggingface@0.1.23
+  - @llamaindex/jinaai@0.0.29
+  - @llamaindex/mistral@0.1.18
+  - @llamaindex/mixedbread@0.0.32
+  - @llamaindex/notion@0.1.17
+  - @llamaindex/ollama@0.1.18
+  - @llamaindex/openai@0.4.13
+  - @llamaindex/perplexity@0.0.26
+  - @llamaindex/portkey-ai@0.0.60
+  - @llamaindex/replicate@0.0.60
+  - @llamaindex/bm25-retriever@0.0.7
+  - @llamaindex/astra@0.0.32
+  - @llamaindex/azure@0.1.30
+  - @llamaindex/chroma@0.0.32
+  - @llamaindex/elastic-search@0.1.18
+  - @llamaindex/firestore@1.0.25
+  - @llamaindex/milvus@0.1.27
+  - @llamaindex/mongodb@0.0.33
+  - @llamaindex/pinecone@0.1.18
+  - @llamaindex/postgres@0.0.61
+  - @llamaindex/qdrant@0.1.28
+  - @llamaindex/supabase@0.1.19
+  - @llamaindex/upstash@0.0.32
+  - @llamaindex/weaviate@0.0.33
+  - @llamaindex/vercel@0.1.18
+  - @llamaindex/voyage-ai@1.0.24
+  - @llamaindex/readers@3.1.17
+  - @llamaindex/tools@0.1.8
+  - @llamaindex/deepseek@0.0.30
+  - @llamaindex/fireworks@0.0.29
+  - @llamaindex/groq@0.0.85
+  - @llamaindex/together@0.0.29
+  - @llamaindex/vllm@0.0.55
+  - @llamaindex/xai@0.0.16
+
+## 0.3.33
+
+### Patch Changes
+
+- Updated dependencies [38da40b]
+  - @llamaindex/core@0.6.17
+  - @llamaindex/cloud@4.0.26
+  - llamaindex@0.11.21
+  - @llamaindex/node-parser@2.0.17
+  - @llamaindex/anthropic@0.3.19
+  - @llamaindex/assemblyai@0.1.16
+  - @llamaindex/clip@0.0.68
+  - @llamaindex/cohere@0.0.31
+  - @llamaindex/deepinfra@0.0.68
+  - @llamaindex/discord@0.1.16
+  - @llamaindex/google@0.3.16
+  - @llamaindex/huggingface@0.1.22
+  - @llamaindex/jinaai@0.0.28
+  - @llamaindex/mistral@0.1.17
+  - @llamaindex/mixedbread@0.0.31
+  - @llamaindex/notion@0.1.16
+  - @llamaindex/ollama@0.1.17
+  - @llamaindex/openai@0.4.12
+  - @llamaindex/perplexity@0.0.25
+  - @llamaindex/portkey-ai@0.0.59
+  - @llamaindex/replicate@0.0.59
+  - @llamaindex/bm25-retriever@0.0.6
+  - @llamaindex/astra@0.0.31
+  - @llamaindex/azure@0.1.29
+  - @llamaindex/chroma@0.0.31
+  - @llamaindex/elastic-search@0.1.17
+  - @llamaindex/firestore@1.0.24
+  - @llamaindex/milvus@0.1.26
+  - @llamaindex/mongodb@0.0.32
+  - @llamaindex/pinecone@0.1.17
+  - @llamaindex/postgres@0.0.60
+  - @llamaindex/qdrant@0.1.27
+  - @llamaindex/supabase@0.1.18
+  - @llamaindex/upstash@0.0.31
+  - @llamaindex/weaviate@0.0.32
+  - @llamaindex/vercel@0.1.17
+  - @llamaindex/voyage-ai@1.0.23
+  - @llamaindex/readers@3.1.16
+  - @llamaindex/tools@0.1.7
+  - @llamaindex/workflow@1.1.17
+  - @llamaindex/deepseek@0.0.29
+  - @llamaindex/fireworks@0.0.28
+  - @llamaindex/groq@0.0.84
+  - @llamaindex/together@0.0.28
+  - @llamaindex/vllm@0.0.54
+  - @llamaindex/xai@0.0.15
+
+## 0.3.32
+
+### Patch Changes
+
+- Updated dependencies [650eeb1]
+- Updated dependencies [a8ec08c]
+- Updated dependencies [2967d57]
+  - @llamaindex/google@0.3.15
+  - @llamaindex/core@0.6.16
+  - @llamaindex/workflow@1.1.16
+  - @llamaindex/cloud@4.0.25
+  - llamaindex@0.11.20
+  - @llamaindex/node-parser@2.0.16
+  - @llamaindex/anthropic@0.3.18
+  - @llamaindex/assemblyai@0.1.15
+  - @llamaindex/clip@0.0.67
+  - @llamaindex/cohere@0.0.30
+  - @llamaindex/deepinfra@0.0.67
+  - @llamaindex/discord@0.1.15
+  - @llamaindex/huggingface@0.1.21
+  - @llamaindex/jinaai@0.0.27
+  - @llamaindex/mistral@0.1.16
+  - @llamaindex/mixedbread@0.0.30
+  - @llamaindex/notion@0.1.15
+  - @llamaindex/ollama@0.1.16
+  - @llamaindex/openai@0.4.11
+  - @llamaindex/perplexity@0.0.24
+  - @llamaindex/portkey-ai@0.0.58
+  - @llamaindex/replicate@0.0.58
+  - @llamaindex/bm25-retriever@0.0.5
+  - @llamaindex/astra@0.0.30
+  - @llamaindex/azure@0.1.28
+  - @llamaindex/chroma@0.0.30
+  - @llamaindex/elastic-search@0.1.16
+  - @llamaindex/firestore@1.0.23
+  - @llamaindex/milvus@0.1.25
+  - @llamaindex/mongodb@0.0.31
+  - @llamaindex/pinecone@0.1.16
+  - @llamaindex/postgres@0.0.59
+  - @llamaindex/qdrant@0.1.26
+  - @llamaindex/supabase@0.1.17
+  - @llamaindex/upstash@0.0.30
+  - @llamaindex/weaviate@0.0.31
+  - @llamaindex/vercel@0.1.16
+  - @llamaindex/voyage-ai@1.0.22
+  - @llamaindex/readers@3.1.15
+  - @llamaindex/tools@0.1.6
+  - @llamaindex/deepseek@0.0.28
+  - @llamaindex/fireworks@0.0.27
+  - @llamaindex/groq@0.0.83
+  - @llamaindex/together@0.0.27
+  - @llamaindex/vllm@0.0.53
+  - @llamaindex/xai@0.0.14
+
+## 0.3.31
+
+### Patch Changes
+
+- Updated dependencies [d8f4f6a]
+- Updated dependencies [856dd8c]
+  - @llamaindex/supabase@0.1.16
+  - @llamaindex/openai@0.4.10
+  - @llamaindex/clip@0.0.66
+  - @llamaindex/deepinfra@0.0.66
+  - @llamaindex/deepseek@0.0.27
+  - @llamaindex/fireworks@0.0.26
+  - @llamaindex/groq@0.0.82
+  - @llamaindex/huggingface@0.1.20
+  - @llamaindex/jinaai@0.0.26
+  - @llamaindex/perplexity@0.0.23
+  - @llamaindex/azure@0.1.27
+  - @llamaindex/together@0.0.26
+  - @llamaindex/vllm@0.0.52
+  - @llamaindex/xai@0.0.13
+
+## 0.3.30
+
+### Patch Changes
+
+- Updated dependencies [7ad3411]
+- Updated dependencies [5da5b3c]
+- Updated dependencies [a1fdb07]
+- Updated dependencies [ddc0eaf]
+  - @llamaindex/core@0.6.15
+  - @llamaindex/tools@0.1.5
+  - @llamaindex/workflow@1.1.15
+  - @llamaindex/openai@0.4.9
+  - @llamaindex/anthropic@0.3.17
+  - @llamaindex/cloud@4.0.24
+  - llamaindex@0.11.19
+  - @llamaindex/node-parser@2.0.15
+  - @llamaindex/assemblyai@0.1.14
+  - @llamaindex/clip@0.0.65
+  - @llamaindex/cohere@0.0.29
+  - @llamaindex/deepinfra@0.0.65
+  - @llamaindex/discord@0.1.14
+  - @llamaindex/google@0.3.14
+  - @llamaindex/huggingface@0.1.19
+  - @llamaindex/jinaai@0.0.25
+  - @llamaindex/mistral@0.1.15
+  - @llamaindex/mixedbread@0.0.29
+  - @llamaindex/notion@0.1.14
+  - @llamaindex/ollama@0.1.15
+  - @llamaindex/perplexity@0.0.22
+  - @llamaindex/portkey-ai@0.0.57
+  - @llamaindex/replicate@0.0.57
+  - @llamaindex/bm25-retriever@0.0.4
+  - @llamaindex/astra@0.0.29
+  - @llamaindex/azure@0.1.26
+  - @llamaindex/chroma@0.0.29
+  - @llamaindex/elastic-search@0.1.15
+  - @llamaindex/firestore@1.0.22
+  - @llamaindex/milvus@0.1.24
+  - @llamaindex/mongodb@0.0.30
+  - @llamaindex/pinecone@0.1.15
+  - @llamaindex/postgres@0.0.58
+  - @llamaindex/qdrant@0.1.25
+  - @llamaindex/supabase@0.1.15
+  - @llamaindex/upstash@0.0.29
+  - @llamaindex/weaviate@0.0.30
+  - @llamaindex/vercel@0.1.15
+  - @llamaindex/voyage-ai@1.0.21
+  - @llamaindex/readers@3.1.14
+  - @llamaindex/deepseek@0.0.26
+  - @llamaindex/fireworks@0.0.25
+  - @llamaindex/groq@0.0.81
+  - @llamaindex/together@0.0.25
+  - @llamaindex/vllm@0.0.51
+  - @llamaindex/xai@0.0.12
+
 ## 0.3.29

 ### Patch Changes
@@ -3,7 +3,7 @@
 */
 import { openai } from "@llamaindex/openai";
 import { agent } from "@llamaindex/workflow";
-import { getWeatherTool } from "../../deprecated/agents/utils/tools";
+import { getWeatherTool } from "../tools/tools";

 async function main() {
  const weatherAgent = agent({
@@ -24,6 +24,7 @@ async function main() {
    state: result.data.state,
  });
  console.log(`${JSON.stringify(caResult, null, 2)}`);
+  console.log("assistant message:", result.data.message);
 }

 main().catch((error) => {
@@ -1,6 +1,6 @@
 import { ollama } from "@llamaindex/ollama";
 import { agent } from "@llamaindex/workflow";
-import { getWeatherTool } from "../../deprecated/agents/utils/tools";
+import { getWeatherTool } from "../tools/tools";

 async function main() {
  const myAgent = agent({
@@ -0,0 +1,150 @@
+/**
+ * Example: Vector Memory Block
+ *
+ * This example demonstrates how to use the VectorMemoryBlock to store and retrieve
+ * conversation history using vector similarity search. The vector memory block
+ * stores messages in a vector store and can retrieve relevant context based on
+ * semantic similarity to recent messages.
+ */
+
+import { OpenAI, OpenAIEmbedding } from "@llamaindex/openai";
+import { QdrantVectorStore } from "@llamaindex/qdrant";
+import { createMemory, vectorBlock } from "llamaindex";
+
+// Set up the LLM and embedding model
+const llm = new OpenAI({ model: "gpt-4.1-mini" });
+const embedModel = new OpenAIEmbedding({ model: "text-embedding-3-small" });
+
+// Simulate a conversation with some context
+// This conversation has 8 messages, which is more than the token limit of 100 tokens (set below)
+// The last 4 messages are kept in to short term memory block (as their tokens are in the limit)
+// Whereas the first 5 messages are added to long term memory block (in here we will use the vector memory block with Qdrant)
+const CONVERSATION_TURNS = [
+  //// This is the first 5 messages that are added to long term memory block (vector memory block)
+  {
+    role: "user",
+    content: "Hi, I'm Sarah and I work as a data scientist at Google.",
+  },
+  {
+    role: "assistant",
+    content:
+      "Hello Sarah! It's great to meet you. Data science at Google must be exciting!",
+  },
+  {
+    role: "user",
+    content:
+      "Yes, I specialize in machine learning and natural language processing.",
+  },
+  {
+    role: "assistant",
+    content: "That's impressive! ML and NLP are fascinating fields.",
+  },
+  {
+    role: "user",
+    content:
+      "I have a PhD in Computer Science from Stanford, and I love hiking on weekends.",
+  },
+
+  //// This is the last 4 messages that are added to short term memory block
+  {
+    role: "assistant",
+    content:
+      "Wow, Stanford PhD! And hiking is a great way to unwind from tech work.",
+  },
+  {
+    role: "user",
+    content: "I also have two cats named Whiskers and Mittens.",
+  },
+  {
+    role: "assistant",
+    content:
+      "Cats make wonderful companions! Whiskers and Mittens are cute names.",
+  },
+  {
+    role: "user",
+    content: "Summary information about Sarah and her cats",
+  },
+];
+
+async function main() {
+  console.log("=== Vector Memory Block Example ===\n");
+
+  /**
+   * Create a vector store. You can quickly get a local instance of Qdrant running with Docker:
+   * ```bash
+   * docker pull qdrant/qdrant
+   * docker run -p 6333:6333 qdrant/qdrant
+   * ```
+   *
+   * Go to http://localhost:6333/dashboard#/collections to see your data
+   */
+  const vectorStore = new QdrantVectorStore({
+    url: "http://localhost:6333",
+    embedModel,
+  });
+
+  // Create a vector memory block using the factory function
+  const vectorMemoryBlock = vectorBlock({
+    vectorStore,
+    priority: 5,
+  });
+
+  // Create a memory store with the vector memory block
+  const memory = createMemory([], {
+    llm,
+    memoryBlocks: [vectorMemoryBlock],
+    tokenLimit: 100,
+    shortTermTokenLimitRatio: 0.7,
+  });
+
+  // Store the conversation history in the vector memory
+  console.log(`Adding ${CONVERSATION_TURNS.length} messages to the memory...`);
+  for (const message of CONVERSATION_TURNS) {
+    await memory.add(message);
+  }
+
+  // Retrieve relevant context for the current user request
+  console.log("Retrieving relevant context...");
+  const chatHistory = await memory.getLLM();
+
+  // You will see there's 1 generated context message from vector memory block, and 4 messages from short term memory block
+  console.log("Chat memory:", chatHistory);
+
+  // Now simulate the assistant responding with context
+  console.log("\nAssistant response with context:");
+  const response = await llm.chat({
+    messages: chatHistory,
+  });
+  console.log(response.message.content);
+
+  // Try adding more messages to the memory
+  const newMessages = [
+    {
+      role: "user",
+      content: "Write a long paragraph about weather in Tokyo",
+    },
+    {
+      role: "assistant",
+      content:
+        "The weather in Tokyo is sunny and warm. The temperature is around 20 degrees Celsius. The weather is very nice and the people are friendly.",
+    },
+    {
+      role: "user",
+      content: "What is the weather in Tokyo?",
+    },
+  ];
+  // Add the new messages to the memory
+  for (const message of newMessages) {
+    await memory.add(message);
+  }
+
+  // Try retrieving the new messages
+  const newChatHistory = await memory.getLLM();
+  // You can see now that new chat history will contain the nodes (separated by `\n`) in the
+  // context message that is generated by the vector memory block
+  // The number of retrieved nodes is set by `similarityTopK` in `queryOptions` of `vectorBlock`
+  // (default `similarityTopK` is 2)
+  console.log("New chat history:", newChatHistory);
+}
+
+main().catch(console.error);
@@ -1,7 +1,7 @@
-import { OpenAI } from "@llamaindex/openai";
+import { openai } from "@llamaindex/openai";

 async function main() {
-  const llm = new OpenAI({ model: "gpt-4-turbo" });
+  const llm = openai({ model: "gpt-4.1-mini" });
  const args: Parameters<typeof llm.chat>[0] = {
    additionalChatOptions: {
      tool_choice: "auto",
@@ -0,0 +1,46 @@
+import { openai } from "@llamaindex/openai";
+import { tool } from "llamaindex";
+import z from "zod";
+
+import { ChatMessage } from "llamaindex";
+
+async function main() {
+  const llm = openai({ model: "gpt-4.1-mini" });
+  const messages = [
+    {
+      content: `What's the weather like in San Francisco?`,
+      role: "user",
+    } as ChatMessage,
+  ];
+
+  let exit = false;
+  do {
+    const { stream, newMessages, toolCalls } = await llm.exec({
+      messages,
+      tools: [
+        tool({
+          name: "get_weather",
+          description: "Get the current weather for a location",
+          parameters: z.object({
+            address: z.string().describe("The address"),
+          }),
+          execute: ({ address }) => {
+            return `It's sunny in ${address}!`;
+          },
+        }),
+      ],
+      stream: true,
+    });
+    for await (const chunk of stream) {
+      process.stdout.write(chunk.delta);
+    }
+    messages.push(...newMessages());
+    // exit condition to stop the agent loop
+    // here we can also check for specific tool calls or limit the number of llm.exec calls
+    exit = toolCalls.length === 0;
+  } while (!exit);
+}
+
+(async function () {
+  await main();
+})();
@@ -0,0 +1,43 @@
+import { openai } from "@llamaindex/openai";
+import { ChatMessage, tool } from "llamaindex";
+import z from "zod";
+
+async function main() {
+  const llm = openai({ model: "gpt-4.1-mini" });
+  const messages = [
+    {
+      content: `What's the weather like in San Francisco?`,
+      role: "user",
+    } as ChatMessage,
+  ];
+
+  let exit = false;
+  do {
+    const { newMessages, toolCalls } = await llm.exec({
+      messages,
+      tools: [
+        tool({
+          name: "get_weather",
+          description: "Get the current weather for a location",
+          parameters: z.object({
+            address: z.string().describe("The address"),
+          }),
+          execute: ({ address }) => {
+            return `It's sunny in ${address}!`;
+          },
+        }),
+      ],
+    });
+    console.log(newMessages);
+    messages.push(...newMessages);
+    // exit condition to stop the agent loop
+    // here we can also check for specific tool calls or limit the number of llm.exec calls
+    exit = toolCalls.length === 0;
+  } while (!exit);
+}
+
+(async function () {
+  console.log("Starting...");
+  await main();
+  console.log("Done");
+})();
@@ -4,7 +4,7 @@ import {
  getCurrentIDTool,
  getUserInfoTool,
  getWeatherTool,
-} from "./utils/tools";
+} from "../../agents/tools/tools";

 async function main() {
  // Create an OpenAIAgent with the function tools
@@ -3,7 +3,7 @@ import {
  getCurrentIDTool,
  getUserInfoTool,
  getWeatherTool,
-} from "./utils/tools";
+} from "../../agents/tools/tools";

 async function main() {
  // Create an OpenAIAgent with the function tools
@@ -30,6 +30,12 @@ async function main() {
  );
  // and print out the text part
  console.log(textPart?.text);
+
+  const imageId = response.message.options?.image_id;
+  if (imageId) {
+    console.log("Image ID for multi-turn generation:", imageId);
+    console.log("Use this image_id in subsequent requests to modify the image");
+  }
 }

 main().catch(console.error);
@@ -0,0 +1,89 @@
+import { openaiResponses } from "@llamaindex/openai";
+import fs from "fs";
+import { MessageContentDetail } from "llamaindex";
+
+async function main() {
+  const llm = openaiResponses({
+    model: "gpt-4.1-mini",
+    builtInTools: [{ type: "image_generation" }],
+  });
+
+  // First turn: Generate initial image
+  console.log("=== First Turn: Generate initial image ===");
+  const firstResponse = await llm.chat({
+    messages: [
+      {
+        role: "user",
+        content:
+          "Generate an image of a cute tiny llama wearing a hat playing with a cat on a meadow",
+      },
+    ],
+  });
+
+  const firstContent = firstResponse.message.content as MessageContentDetail[];
+  const firstImagePart = firstContent.find((part) => part.type === "image");
+  const firstTextPart = firstContent.find((part) => part.type === "text");
+
+  // Save the first image
+  if (firstImagePart?.data) {
+    fs.writeFileSync(
+      "llama-initial.png",
+      Buffer.from(firstImagePart.data as string, "base64"),
+    );
+    console.log("First image saved as 'llama-initial.png'");
+  }
+
+  if (firstTextPart?.text) {
+    console.log("First response:", firstTextPart.text);
+  }
+
+  // Get the image_id from the response options for multi-turn
+  const imageId = firstResponse.message.options?.image_id;
+  console.log("Image ID for multi-turn:", imageId);
+
+  if (imageId) {
+    // Second turn: Modify the image using the image_id
+    console.log("\n=== Second Turn: Modify the image ===");
+    const secondResponse = await llm.chat({
+      messages: [
+        {
+          role: "user",
+          content:
+            "Generate an image of a cute tiny llama wearing a hat playing with a cat on a meadow",
+        },
+        {
+          role: "assistant",
+          content: firstContent,
+          options: { image_id: imageId },
+        },
+        {
+          role: "user",
+          content:
+            "Now add a rainbow in the background and make the llama's hat blue",
+        },
+      ],
+    });
+
+    const secondContent = secondResponse.message
+      .content as MessageContentDetail[];
+    const secondImagePart = secondContent.find((part) => part.type === "image");
+    const secondTextPart = secondContent.find((part) => part.type === "text");
+
+    // Save the modified image
+    if (secondImagePart?.data) {
+      fs.writeFileSync(
+        "llama-modified.png",
+        Buffer.from(secondImagePart.data as string, "base64"),
+      );
+      console.log("Modified image saved as 'llama-modified.png'");
+    }
+
+    if (secondTextPart?.text) {
+      console.log("Second response:", secondTextPart.text);
+    }
+  } else {
+    console.log("No image_id received, cannot perform multi-turn generation");
+  }
+}
+
+main().catch(console.error);
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/examples",
-  "version": "0.3.29",
+  "version": "0.3.34",
  "private": true,
  "scripts": {
    "lint": "eslint .",
@@ -11,52 +11,52 @@
    "@azure/cosmos": "^4.1.1",
    "@azure/identity": "^4.4.1",
    "@azure/search-documents": "^12.1.0",
-    "@llamaindex/anthropic": "^0.3.16",
-    "@llamaindex/assemblyai": "^0.1.13",
-    "@llamaindex/astra": "^0.0.28",
-    "@llamaindex/azure": "^0.1.25",
-    "@llamaindex/bm25-retriever": "^0.0.3",
-    "@llamaindex/chroma": "^0.0.28",
-    "@llamaindex/clip": "^0.0.64",
-    "@llamaindex/cloud": "^4.0.19",
-    "@llamaindex/cohere": "^0.0.28",
-    "@llamaindex/core": "^0.6.14",
-    "@llamaindex/deepinfra": "^0.0.64",
-    "@llamaindex/deepseek": "^0.0.25",
-    "@llamaindex/discord": "^0.1.13",
-    "@llamaindex/elastic-search": "^0.1.14",
+    "@llamaindex/anthropic": "^0.3.20",
+    "@llamaindex/assemblyai": "^0.1.17",
+    "@llamaindex/astra": "^0.0.32",
+    "@llamaindex/azure": "^0.1.30",
+    "@llamaindex/bm25-retriever": "^0.0.7",
+    "@llamaindex/chroma": "^0.0.32",
+    "@llamaindex/clip": "^0.0.69",
+    "@llamaindex/cloud": "^4.0.27",
+    "@llamaindex/cohere": "^0.0.32",
+    "@llamaindex/core": "^0.6.18",
+    "@llamaindex/deepinfra": "^0.0.69",
+    "@llamaindex/deepseek": "^0.0.30",
+    "@llamaindex/discord": "^0.1.17",
+    "@llamaindex/elastic-search": "^0.1.18",
    "@llamaindex/env": "^0.1.30",
-    "@llamaindex/firestore": "^1.0.21",
-    "@llamaindex/fireworks": "^0.0.24",
-    "@llamaindex/google": "^0.3.13",
-    "@llamaindex/groq": "^0.0.80",
-    "@llamaindex/huggingface": "^0.1.18",
-    "@llamaindex/jinaai": "^0.0.24",
-    "@llamaindex/milvus": "^0.1.23",
-    "@llamaindex/mistral": "^0.1.14",
-    "@llamaindex/mixedbread": "^0.0.28",
-    "@llamaindex/mongodb": "^0.0.29",
-    "@llamaindex/node-parser": "^2.0.14",
-    "@llamaindex/notion": "^0.1.13",
-    "@llamaindex/ollama": "^0.1.14",
-    "@llamaindex/openai": "^0.4.8",
-    "@llamaindex/perplexity": "^0.0.21",
-    "@llamaindex/pinecone": "^0.1.14",
-    "@llamaindex/portkey-ai": "^0.0.56",
-    "@llamaindex/postgres": "^0.0.57",
-    "@llamaindex/qdrant": "^0.1.24",
-    "@llamaindex/readers": "^3.1.13",
-    "@llamaindex/replicate": "^0.0.56",
-    "@llamaindex/supabase": "^0.1.14",
-    "@llamaindex/together": "^0.0.24",
-    "@llamaindex/tools": "^0.1.4",
-    "@llamaindex/upstash": "^0.0.28",
-    "@llamaindex/vercel": "^0.1.14",
-    "@llamaindex/vllm": "^0.0.50",
-    "@llamaindex/voyage-ai": "^1.0.20",
-    "@llamaindex/weaviate": "^0.0.29",
-    "@llamaindex/workflow": "^1.1.14",
-    "@llamaindex/xai": "^0.0.11",
+    "@llamaindex/firestore": "^1.0.25",
+    "@llamaindex/fireworks": "^0.0.29",
+    "@llamaindex/google": "^0.3.17",
+    "@llamaindex/groq": "^0.0.85",
+    "@llamaindex/huggingface": "^0.1.23",
+    "@llamaindex/jinaai": "^0.0.29",
+    "@llamaindex/milvus": "^0.1.27",
+    "@llamaindex/mistral": "^0.1.18",
+    "@llamaindex/mixedbread": "^0.0.32",
+    "@llamaindex/mongodb": "^0.0.33",
+    "@llamaindex/node-parser": "^2.0.18",
+    "@llamaindex/notion": "^0.1.17",
+    "@llamaindex/ollama": "^0.1.18",
+    "@llamaindex/openai": "^0.4.13",
+    "@llamaindex/perplexity": "^0.0.26",
+    "@llamaindex/pinecone": "^0.1.18",
+    "@llamaindex/portkey-ai": "^0.0.60",
+    "@llamaindex/postgres": "^0.0.61",
+    "@llamaindex/qdrant": "^0.1.28",
+    "@llamaindex/readers": "^3.1.17",
+    "@llamaindex/replicate": "^0.0.60",
+    "@llamaindex/supabase": "^0.1.19",
+    "@llamaindex/together": "^0.0.29",
+    "@llamaindex/tools": "^0.1.8",
+    "@llamaindex/upstash": "^0.0.32",
+    "@llamaindex/vercel": "^0.1.18",
+    "@llamaindex/vllm": "^0.0.55",
+    "@llamaindex/voyage-ai": "^1.0.24",
+    "@llamaindex/weaviate": "^0.0.33",
+    "@llamaindex/workflow": "^1.1.19",
+    "@llamaindex/xai": "^0.0.16",
    "@notionhq/client": "^4.0.0",
    "@pinecone-database/pinecone": "^4.0.0",
    "@vercel/postgres": "^0.10.0",
@@ -65,7 +65,7 @@
    "commander": "^12.1.0",
    "dotenv": "^17.2.0",
    "js-tiktoken": "^1.0.14",
-    "llamaindex": "^0.11.14",
+    "llamaindex": "^0.11.23",
    "mongodb": "6.7.0",
    "postgres": "^3.4.4",
    "wikipedia": "^2.1.2",
@@ -15,7 +15,7 @@ async function main() {
  const vectorStore = new QdrantVectorStore({
    url: process.env.QDRANT_URL,
    apiKey: process.env.QDRANT_API_KEY,
-    embeddingModel: embedding,
+    embedModel: embedding,
    collectionName: "gemini_test",
  });
  const storageContext = await storageContextFromDefaults({ vectorStore });
@@ -16,7 +16,7 @@ async function main() {
  const vectorStore = new QdrantVectorStore({
    url: process.env.QDRANT_URL,
    apiKey: process.env.QDRANT_API_KEY,
-    embeddingModel: embedding,
+    embedModel: embedding,
    collectionName: "jina_test",
  });
  const storageContext = await storageContextFromDefaults({ vectorStore });
@@ -43,6 +43,11 @@
    "vitest": "^3.1.1"
  },
  "packageManager": "pnpm@10.8.1",
+  "pnpm": {
+    "overrides": {
+      "@notionhq/client": "4.0.0"
+    }
+  },
  "lint-staged": {
    "*.{js,jsx,ts,tsx}": [
      "eslint --fix",
@@ -1,5 +1,47 @@
 # @llamaindex/autotool

+## 8.0.23
+
+### Patch Changes
+
+- llamaindex@0.11.23
+
+## 8.0.22
+
+### Patch Changes
+
+- llamaindex@0.11.22
+
+## 8.0.21
+
+### Patch Changes
+
+- llamaindex@0.11.21
+
+## 8.0.20
+
+### Patch Changes
+
+- llamaindex@0.11.20
+
+## 8.0.19
+
+### Patch Changes
+
+- llamaindex@0.11.19
+
+## 8.0.18
+
+### Patch Changes
+
+- llamaindex@0.11.18
+
+## 8.0.17
+
+### Patch Changes
+
+- llamaindex@0.11.17
+
 ## 8.0.16

 ### Patch Changes
@@ -1,5 +1,54 @@
 # @llamaindex/autotool-01-node-example

+## 0.0.131
+
+### Patch Changes
+
+- llamaindex@0.11.23
+- @llamaindex/autotool@8.0.23
+
+## 0.0.130
+
+### Patch Changes
+
+- llamaindex@0.11.22
+- @llamaindex/autotool@8.0.22
+
+## 0.0.129
+
+### Patch Changes
+
+- llamaindex@0.11.21
+- @llamaindex/autotool@8.0.21
+
+## 0.0.128
+
+### Patch Changes
+
+- llamaindex@0.11.20
+- @llamaindex/autotool@8.0.20
+
+## 0.0.127
+
+### Patch Changes
+
+- llamaindex@0.11.19
+- @llamaindex/autotool@8.0.19
+
+## 0.0.126
+
+### Patch Changes
+
+- llamaindex@0.11.18
+- @llamaindex/autotool@8.0.18
+
+## 0.0.125
+
+### Patch Changes
+
+- llamaindex@0.11.17
+- @llamaindex/autotool@8.0.17
+
 ## 0.0.124

 ### Patch Changes
@@ -13,5 +13,5 @@
  "scripts": {
    "start": "node --import tsx --import @llamaindex/autotool/node ./src/index.ts"
  },
-  "version": "0.0.124"
+  "version": "0.0.131"
 }
@@ -6,7 +6,7 @@
    "url": "git+https://github.com/run-llama/LlamaIndexTS.git",
    "directory": "packages/autotool"
  },
-  "version": "8.0.16",
+  "version": "8.0.23",
  "description": "auto transpile your JS function to LLM Agent compatible",
  "files": [
    "dist",
@@ -1,5 +1,48 @@
 # @llamaindex/cloud

+## 4.0.27
+
+### Patch Changes
+
+- Updated dependencies [f29799e]
+- Updated dependencies [7224c06]
+  - @llamaindex/core@0.6.18
+
+## 4.0.26
+
+### Patch Changes
+
+- Updated dependencies [38da40b]
+  - @llamaindex/core@0.6.17
+
+## 4.0.25
+
+### Patch Changes
+
+- 2967d57: Default to \_public agent url id
+- Updated dependencies [a8ec08c]
+  - @llamaindex/core@0.6.16
+
+## 4.0.24
+
+### Patch Changes
+
+- Updated dependencies [7ad3411]
+- Updated dependencies [5da5b3c]
+  - @llamaindex/core@0.6.15
+
+## 4.0.23
+
+### Patch Changes
+
+- a1b1598: fix: add generic types into agent data responses
+
+## 4.0.22
+
+### Patch Changes
+
+- d2be868: Bug fixes for new beta agent-data cloud API
+
 ## 4.0.21

 ### Patch Changes
@@ -1,6 +1,6 @@
 {
  "name": "@llamaindex/cloud",
-  "version": "4.0.21",
+  "version": "4.0.27",
  "type": "module",
  "license": "MIT",
  "scripts": {
@@ -13,20 +13,20 @@
    "./api",
    "./reader",
    "./parse",
-    "./agent"
+    "./beta/agent"
  ],
  "exports": {
    "./openapi.json": "./openapi.json",
-    "./agent": {
+    "./beta/agent": {
      "require": {
-        "types": "./agent/dist/index.d.cts",
-        "default": "./agent/dist/index.cjs"
+        "types": "./beta/agent/dist/index.d.cts",
+        "default": "./beta/agent/dist/index.cjs"
      },
      "import": {
-        "types": "./agent/dist/index.d.ts",
-        "default": "./agent/dist/index.js"
+        "types": "./beta/agent/dist/index.d.ts",
+        "default": "./beta/agent/dist/index.js"
      },
-      "default": "./agent/dist/index.js"
+      "default": "./beta/agent/dist/index.js"
    },
    "./api": {
      "require": {
@@ -1,136 +0,0 @@
-import { createClient, createConfig } from "@hey-api/client-fetch";
-import { getEnv } from "@llamaindex/env";
-import {
-  createAgentDataApiV1BetaAgentDataPost,
-  deleteAgentDataApiV1BetaAgentDataItemIdDelete,
-  getAgentDataApiV1BetaAgentDataItemIdGet,
-  searchAgentDataApiV1BetaAgentDataSearchPost,
-  updateAgentDataApiV1BetaAgentDataItemIdPut,
-  type AgentData,
-  type PaginatedResponseAgentData,
-  type SearchRequest,
-} from "../client";
-
-type AgentClientOptions = {
-  apiKey?: string;
-  baseUrl?: string;
-  collection: string;
-  agentUrlId: string;
-};
-
-/**
- * Async client for agent data operations
- */
-export class AgentClient {
-  private client: ReturnType<typeof createClient>;
-  private baseUrl: string;
-  private headers: Record<string, string>;
-  private collection: string;
-  private agentUrlId: string;
-
-  constructor(options: AgentClientOptions) {
-    this.collection = options.collection;
-    this.agentUrlId = options.agentUrlId;
-    const apiKey = options?.apiKey || getEnv("LLAMA_CLOUD_API_KEY");
-    this.baseUrl = options?.baseUrl || "https://api.cloud.llamaindex.ai/";
-
-    this.headers = {
-      "X-SDK-Name": "llamaindex-ts",
-      ...(apiKey && { Authorization: `Bearer ${apiKey}` }),
-    };
-
-    this.client = createClient(
-      createConfig({
-        baseUrl: this.baseUrl,
-        headers: this.headers,
-      }),
-    );
-  }
-
-  /**
-   * Create new agent data
-   */
-  async createItem<T>(data: T): Promise<AgentData> {
-    const response = await createAgentDataApiV1BetaAgentDataPost({
-      throwOnError: true,
-      body: {
-        collection: this.collection,
-        agent_slug: this.agentUrlId,
-        data: data as Record<string, unknown>,
-      },
-      client: this.client,
-    });
-
-    return response.data;
-  }
-
-  /**
-   * Get agent data by ID
-   */
-  async getItem(id: string): Promise<AgentData | null> {
-    try {
-      const response = await getAgentDataApiV1BetaAgentDataItemIdGet({
-        throwOnError: true,
-        path: { item_id: id },
-        client: this.client,
-      });
-
-      return response.data;
-    } catch (error) {
-      if (
-        error instanceof Error &&
-        "response" in error &&
-        (error as { response?: { status?: number } }).response?.status === 404
-      ) {
-        return null;
-      }
-      throw error;
-    }
-  }
-
-  /**
-   * Update agent data
-   */
-  async updateItem<T>(id: string, data: T): Promise<AgentData> {
-    const response = await updateAgentDataApiV1BetaAgentDataItemIdPut({
-      throwOnError: true,
-      path: { item_id: id },
-      body: {
-        data: data as Record<string, unknown>,
-      },
-      client: this.client,
-    });
-
-    return response.data;
-  }
-
-  /**
-   * Delete agent data
-   */
-  async delete(id: string): Promise<void> {
-    await deleteAgentDataApiV1BetaAgentDataItemIdDelete({
-      throwOnError: true,
-      path: { item_id: id },
-      client: this.client,
-    });
-  }
-
-  /**
-   * List agent data
-   */
-  async list(options: SearchRequest): Promise<PaginatedResponseAgentData> {
-    const response = await searchAgentDataApiV1BetaAgentDataSearchPost({
-      throwOnError: true,
-      body: {
-        ...options,
-      },
-      client: this.client,
-    });
-
-    return response.data;
-  }
-}
-
-export function createAgentClient(options: AgentClientOptions): AgentClient {
-  return new AgentClient(options);
-}
@@ -1 +0,0 @@
-export { AgentClient, createAgentClient } from "./client";
@@ -0,0 +1,329 @@
+import { createClient, createConfig } from "@hey-api/client-fetch";
+import { getEnv } from "@llamaindex/env";
+import {
+  aggregateAgentDataApiV1BetaAgentDataAggregatePost,
+  createAgentDataApiV1BetaAgentDataPost,
+  deleteAgentDataApiV1BetaAgentDataItemIdDelete,
+  getAgentDataApiV1BetaAgentDataItemIdGet,
+  searchAgentDataApiV1BetaAgentDataSearchPost,
+  updateAgentDataApiV1BetaAgentDataItemIdPut,
+  type AgentData,
+  type AggregateGroup,
+} from "../../client";
+import type {
+  AggregateAgentDataOptions,
+  SearchAgentDataOptions,
+  TypedAgentData,
+  TypedAgentDataItems,
+  TypedAggregateGroup,
+  TypedAggregateGroupItems,
+} from "./types";
+
+/**
+ * Async client for agent data operations
+ */
+export class AgentClient<T = unknown> {
+  private client: ReturnType<typeof createClient>;
+  private baseUrl: string;
+  private headers: Record<string, string>;
+  private collection: string;
+  private agentUrlId: string;
+
+  constructor({
+    apiKey = getEnv("LLAMA_CLOUD_API_KEY"),
+    baseUrl = "https://api.cloud.llamaindex.ai/",
+    collection = "default",
+    agentUrlId = "_public",
+  }: {
+    apiKey?: string;
+    baseUrl?: string;
+    collection?: string;
+    agentUrlId?: string;
+  }) {
+    this.baseUrl = baseUrl;
+
+    this.headers = {
+      "X-SDK-Name": "llamaindex-ts",
+      ...(apiKey && { Authorization: `Bearer ${apiKey}` }),
+    };
+
+    this.client = createClient(
+      createConfig({
+        baseUrl: this.baseUrl,
+        headers: this.headers,
+      }),
+    );
+
+    this.collection = collection;
+    this.agentUrlId = agentUrlId;
+  }
+
+  /**
+   * Create new agent data
+   */
+  async createItem(data: T): Promise<TypedAgentData<T>> {
+    const response = await createAgentDataApiV1BetaAgentDataPost({
+      throwOnError: true,
+      body: {
+        agent_slug: this.agentUrlId,
+        collection: this.collection,
+        data: data as Record<string, unknown>,
+      },
+      client: this.client,
+    });
+
+    return this.transformResponse(response.data);
+  }
+
+  /**
+   * Get agent data by ID
+   */
+  async getItem(id: string): Promise<TypedAgentData<T> | null> {
+    try {
+      const response = await getAgentDataApiV1BetaAgentDataItemIdGet({
+        throwOnError: true,
+        path: { item_id: id },
+        client: this.client,
+      });
+
+      return this.transformResponse(response.data);
+    } catch (error) {
+      if (
+        error instanceof Error &&
+        "response" in error &&
+        (error as { response?: { status?: number } }).response?.status === 404
+      ) {
+        return null;
+      }
+      throw error;
+    }
+  }
+
+  /**
+   * Update agent data
+   */
+  async updateItem(id: string, data: T): Promise<TypedAgentData<T>> {
+    const response = await updateAgentDataApiV1BetaAgentDataItemIdPut({
+      throwOnError: true,
+      path: { item_id: id },
+      body: {
+        data: data as Record<string, unknown>,
+      },
+      client: this.client,
+    });
+
+    return this.transformResponse(response.data);
+  }
+
+  /**
+   * Delete agent data
+   */
+  async deleteItem(id: string): Promise<void> {
+    await deleteAgentDataApiV1BetaAgentDataItemIdDelete({
+      throwOnError: true,
+      path: { item_id: id },
+      client: this.client,
+    });
+  }
+
+  /**
+   * Search agent data
+   */
+  async search(
+    options: SearchAgentDataOptions,
+  ): Promise<TypedAgentDataItems<T>> {
+    const response = await searchAgentDataApiV1BetaAgentDataSearchPost({
+      throwOnError: true,
+      body: {
+        agent_slug: this.agentUrlId,
+        ...(this.collection !== undefined && {
+          collection: this.collection,
+        }),
+        ...(options.filter !== undefined && { filter: options.filter }),
+        ...(options.orderBy !== undefined && { order_by: options.orderBy }),
+        ...(options.pageSize !== undefined && { page_size: options.pageSize }),
+        ...(options.offset !== undefined && { offset: options.offset }),
+        ...(options.includeTotal !== undefined && {
+          include_total: options.includeTotal,
+        }),
+      },
+      client: this.client,
+    });
+
+    const result: TypedAgentDataItems<T> = {
+      items: response.data.items.map((item: AgentData) =>
+        this.transformResponse(item),
+      ),
+    };
+
+    if (
+      response.data.total_size !== null &&
+      response.data.total_size !== undefined
+    ) {
+      result.totalSize = response.data.total_size;
+    }
+
+    if (
+      response.data.next_page_token !== null &&
+      response.data.next_page_token !== undefined
+    ) {
+      result.nextPageToken = response.data.next_page_token;
+    }
+
+    return result;
+  }
+
+  /**
+   * Aggregate agent data into groups
+   */
+  async aggregate(
+    options: AggregateAgentDataOptions,
+  ): Promise<TypedAggregateGroupItems<T>> {
+    const response = await aggregateAgentDataApiV1BetaAgentDataAggregatePost({
+      throwOnError: true,
+      body: {
+        agent_slug: this.agentUrlId,
+        ...(this.collection !== undefined && {
+          collection: this.collection,
+        }),
+        ...(options.filter !== undefined && { filter: options.filter }),
+        ...(options.groupBy !== undefined && { group_by: options.groupBy }),
+        ...(options.count !== undefined && { count: options.count }),
+        ...(options.first !== undefined && { first: options.first }),
+        ...(options.orderBy !== undefined && { order_by: options.orderBy }),
+        ...(options.offset !== undefined && { offset: options.offset }),
+        ...(options.pageSize !== undefined && { page_size: options.pageSize }),
+      },
+      client: this.client,
+    });
+
+    const result: TypedAggregateGroupItems<T> = {
+      items: response.data.items.map((item) =>
+        this.transformAggregateResponse(item),
+      ),
+    };
+
+    if (
+      response.data.total_size !== null &&
+      response.data.total_size !== undefined
+    ) {
+      result.totalSize = response.data.total_size;
+    }
+
+    if (
+      response.data.next_page_token !== null &&
+      response.data.next_page_token !== undefined
+    ) {
+      result.nextPageToken = response.data.next_page_token;
+    }
+
+    return result;
+  }
+
+  /**
+   * Transform API response to typed data
+   */
+  private transformResponse(data: AgentData): TypedAgentData<T> {
+    const result: TypedAgentData<T> = {
+      id: data.id!,
+      agentUrlId: data.agent_slug,
+      data: data.data as T,
+      createdAt: new Date(data.created_at!),
+      updatedAt: new Date(data.updated_at!),
+    };
+
+    if (data.collection !== undefined) {
+      result.collection = data.collection;
+    }
+
+    return result;
+  }
+
+  /**
+   * Transform API aggregate response to typed data
+   */
+  private transformAggregateResponse(
+    data: AggregateGroup,
+  ): TypedAggregateGroup<T> {
+    const result: TypedAggregateGroup<T> = {
+      groupKey: data.group_key,
+    };
+
+    if (data.count !== null && data.count !== undefined) {
+      result.count = data.count;
+    }
+
+    if (data.first_item !== null && data.first_item !== undefined) {
+      result.firstItem = data.first_item as T;
+    }
+
+    return result;
+  }
+}
+
+export interface AgentDataClientOptions<T = unknown> {
+  /** API key for the client */
+  apiKey?: string;
+  /** Base URL for the client */
+  /** Base URL of the llama cloud api */
+  baseUrl?: string;
+  /** If running in an agent runtime, optionally provide the window url to infer the agent url id */
+  windowUrl?: string;
+  /** Agent URL ID for the client, if not provided, it will be inferred from the window url, or fall back to "default" */
+  agentUrlId?: string;
+  /** Collection name for the client, defaults to "default" */
+  collection?: string;
+}
+/**
+ * Create a new AsyncAgentDataClient instance. Does it's best to infer an agent url id from environment.
+ * Pass in the window url and/or env to infer the agent url id from them.
+ * @param options - The options for the client
+ * @returns A new AgentClient instance
+ */
+export function createAgentDataClient<T = unknown>({
+  apiKey,
+  baseUrl,
+  windowUrl,
+  env,
+  agentUrlId,
+  collection = "default",
+}: {
+  apiKey?: string;
+  baseUrl?: string;
+  windowUrl?: string;
+  env?: Record<string, string>;
+  agentUrlId?: string;
+  collection?: string;
+} = {}): AgentClient<T> {
+  if (env && !agentUrlId) {
+    agentUrlId =
+      env.LLAMA_DEPLOY_DEPLOYMENT_NAME ||
+      env.NEXT_PUBLIC_LLAMA_DEPLOY_DEPLOYMENT_NAME ||
+      env.VITE_LLAMA_DEPLOY_DEPLOYMENT_NAME;
+  }
+  if (windowUrl && !agentUrlId) {
+    try {
+      const url = new URL(windowUrl);
+      const path = url.pathname;
+      const isLocalhost = // local agents should default to _public, otherwise a full deployment is required
+        url.hostname.includes("localhost") ||
+        url.hostname.includes("127.0.0.1");
+      if (path.startsWith("/deployments/") && !isLocalhost) {
+        // /deployments/<agent-url-id>/ui/ -> ["", "deployments", "<agent-url-id>", "ui"]
+        agentUrlId = path.split("/")[2];
+      }
+    } catch (error) {
+      console.warn(
+        "Failed to infer agent url id from window url, falling back to default",
+        error,
+      );
+    }
+  }
+
+  return new AgentClient({
+    ...(apiKey && { apiKey }),
+    ...(baseUrl && { baseUrl }),
+    ...(agentUrlId && { agentUrlId }),
+    collection,
+  });
+}
@@ -0,0 +1,16 @@
+export { AgentClient, createAgentDataClient } from "./client";
+
+export type {
+  AggregateAgentDataOptions,
+  ComparisonOperator,
+  ExtractedData,
+  FilterOperation,
+  SearchAgentDataOptions,
+  StatusType,
+  TypedAgentData,
+  TypedAgentDataItems,
+  TypedAggregateGroup,
+  TypedAggregateGroupItems,
+} from "./types";
+
+export { StatusType as StatusTypeEnum } from "./types";
@@ -0,0 +1,138 @@
+import type { FilterOperation as RawFilterOperation } from "../../client/types.gen";
+/**
+ * Status types for agent data processing
+ */
+export const StatusType = {
+  ERROR: "error",
+  ACCEPTED: "accepted",
+  REJECTED: "rejected",
+  PENDING_REVIEW: "pending_review",
+} as const;
+
+export type StatusType = (typeof StatusType)[keyof typeof StatusType];
+
+export const ComparisonOperator = {
+  GT: "gt",
+  GTE: "gte",
+  LT: "lt",
+  LTE: "lte",
+  EQ: "eq",
+  INCLUDES: "includes",
+} as const;
+
+export type ComparisonOperator =
+  (typeof ComparisonOperator)[keyof typeof ComparisonOperator];
+
+/**
+ * Filter operation for searching/filtering agent data
+ */
+export type FilterOperation = RawFilterOperation;
+
+/**
+ * Base extracted data interface
+ */
+export interface ExtractedData<T = unknown> {
+  /** The original data that was extracted from the document. For tracking changes. Should not be updated. */
+  original_data: T;
+  /** The latest state of the data. Will differ if data has been updated. */
+  data?: T;
+  /** The status of the extracted data. Prefer to use the StatusType values, but any string is allowed. */
+  status: StatusType | string;
+  /** Confidence scores, if any, for each primitive field in the original_data data. */
+  confidence?: Record<string, unknown>;
+  /** The ID of the file that was used to extract the data. */
+  file_id?: string;
+  /** The name of the file that was used to extract the data. */
+  file_name?: string;
+  /** The hash of the file that was used to extract the data. */
+  file_hash?: string;
+  /** Additional metadata about the extracted data, such as errors, tokens, etc. */
+  metadata?: Record<string, unknown>;
+}
+
+/**
+ * TypedAgentData interface for typed agent data
+ */
+export interface TypedAgentData<T = unknown> {
+  /** The unique ID of the agent data record. */
+  id: string;
+  /** The ID of the agent that created the data. */
+  agentUrlId: string;
+  /** The collection of the agent data. */
+  collection?: string;
+  /** The data of the agent data. Usually an ExtractedData&lt;SomeOtherType&gt; */
+  data: T;
+  /** The date and time the data was created. */
+  createdAt: Date;
+  /** The date and time the data was last updated. */
+  updatedAt: Date;
+}
+
+/**
+ * Paginated response of typed agent data items
+ */
+export interface TypedAgentDataItems<T = unknown> {
+  items: TypedAgentData<T>[];
+  totalSize?: number;
+  nextPageToken?: string;
+}
+
+/**
+ * Options for listing agent data
+ */
+export interface SearchAgentDataOptions {
+  /** Filter options for the list. */
+  filter?: Record<string, FilterOperation>;
+  /** Order by options for the list. */
+  orderBy?: string;
+  /** Page size for the list. */
+  pageSize?: number;
+  /** Offset for the list. */
+  offset?: number;
+  /**
+   * Whether to include the total number of items in the response.
+   * Should use only for first request to build total pagination, and not subsequent requests.
+   */
+  includeTotal?: boolean;
+}
+
+/**
+ * Options for aggregating agent data
+ */
+export interface AggregateAgentDataOptions {
+  /** Filter options for the aggregation. */
+  filter?: Record<string, FilterOperation>;
+  /** Fields to group by. */
+  groupBy?: string[];
+  /** Whether to count the number of items in each group. */
+  count?: boolean;
+  /** Whether to return the first item in each group. */
+  first?: boolean;
+  /** Order by options for the aggregation. */
+  orderBy?: string;
+  /** Offset for the aggregation. */
+  offset?: number;
+  /** Page size for the aggregation. */
+  pageSize?: number;
+}
+
+/**
+ * Single aggregation group result
+ */
+export interface TypedAggregateGroup<T = unknown> {
+  /** The group key values */
+  groupKey: Record<string, unknown>;
+  /** Count of items in the group */
+  count?: number;
+  /** First item in the group */
+  firstItem?: T;
+}
+
+/**
+ * Paginated response of aggregated agent data
+ */
+export interface TypedAggregateGroupItems<T = unknown> {
+  items: TypedAggregateGroup<T>[];
+  totalSize?: number;
+  nextPageToken?: string;
+}
@@ -0,0 +1,770 @@
+# @llamaindex/community
+
+## 0.0.100
+
+### Patch Changes
+
+- Updated dependencies [f29799e]
+- Updated dependencies [7224c06]
+  - @llamaindex/core@0.6.18
+
+## 0.0.99
+
+### Patch Changes
+
+- c65a2dc: Deprecate community package and link to AWS package
+
+## 0.0.98
+
+### Patch Changes
+
+- Updated dependencies [9b2e25a]
+  - @llamaindex/core@0.6.4
+  - @llamaindex/env@0.1.30
+
+## 0.0.97
+
+### Patch Changes
+
+- Updated dependencies [3ee8c83]
+  - @llamaindex/core@0.6.3
+
+## 0.0.96
+
+### Patch Changes
+
+- e9bf442: fix: update the tool call schema for nova
+
+## 0.0.95
+
+### Patch Changes
+
+- 411dcea: Add Nova Premier to AWS Nova models. Add EU endpoints
+
+## 0.0.94
+
+### Patch Changes
+
+- Updated dependencies [9c63f3f]
+  - @llamaindex/core@0.6.2
+
+## 0.0.93
+
+### Patch Changes
+
+- Updated dependencies [1b6f368]
+- Updated dependencies [eaf326e]
+  - @llamaindex/core@0.6.1
+
+## 0.0.92
+
+### Patch Changes
+
+- 1325178: fix: stringify all tool results for anthropic on bedrock
+
+## 0.0.91
+
+### Patch Changes
+
+- 5189b44: fix: add retry handling logic to parser reader and fix lint issues
+- 3fd4cc3: feat: use google's new gen ai library to support multimodal output
+- Updated dependencies [21bebfc]
+- Updated dependencies [93bc0ff]
+- Updated dependencies [91a18e7]
+- Updated dependencies [5189b44]
+  - @llamaindex/core@0.6.0
+
+## 0.0.90
+
+### Patch Changes
+
+- Updated dependencies [40ee761]
+  - @llamaindex/core@0.5.8
+
+## 0.0.89
+
+### Patch Changes
+
+- Updated dependencies [4bac71d]
+  - @llamaindex/core@0.5.7
+
+## 0.0.88
+
+### Patch Changes
+
+- e28c29d: Added Llama 3.3 70B Instruct support
+- Updated dependencies [beb922b]
+  - @llamaindex/env@0.1.29
+  - @llamaindex/core@0.5.6
+
+## 0.0.87
+
+### Patch Changes
+
+- Updated dependencies [5668970]
+  - @llamaindex/core@0.5.5
+
+## 0.0.86
+
+### Patch Changes
+
+- Updated dependencies [ad3c7f1]
+  - @llamaindex/core@0.5.4
+
+## 0.0.85
+
+### Patch Changes
+
+- 1914b52: Added Claude 3.7 Sonnet support
+- Updated dependencies [cb021e7]
+  - @llamaindex/core@0.5.3
+
+## 0.0.84
+
+### Patch Changes
+
+- Updated dependencies [d952e68]
+  - @llamaindex/core@0.5.2
+
+## 0.0.83
+
+### Patch Changes
+
+- Updated dependencies [cc50c9c]
+  - @llamaindex/env@0.1.28
+  - @llamaindex/core@0.5.1
+
+## 0.0.82
+
+### Patch Changes
+
+- Updated dependencies [6a4a737]
+- Updated dependencies [d924c63]
+  - @llamaindex/core@0.5.0
+
+## 0.0.81
+
+### Patch Changes
+
+- 1c908fd: Revert previous release (not working with CJS)
+- Updated dependencies [1c908fd]
+  - @llamaindex/core@0.4.23
+  - @llamaindex/env@0.1.27
+
+## 0.0.80
+
+### Patch Changes
+
+- cb608b5: fix: bundle output incorrect
+- Updated dependencies [cb608b5]
+  - @llamaindex/core@0.4.22
+  - @llamaindex/env@0.1.26
+
+## 0.0.79
+
+### Patch Changes
+
+- Updated dependencies [9456616]
+- Updated dependencies [1931bbc]
+  - @llamaindex/core@0.4.21
+
+## 0.0.78
+
+### Patch Changes
+
+- Updated dependencies [d211b7a]
+  - @llamaindex/core@0.4.20
+
+## 0.0.77
+
+### Patch Changes
+
+- 24caf93: fix: added inference profile mapping for nova models"
+- Updated dependencies [a9b5b99]
+  - @llamaindex/core@0.4.19
+
+## 0.0.76
+
+### Patch Changes
+
+- c1850ee: feat: Amazon Nova support via Bedrock
+- Updated dependencies [b504303]
+- Updated dependencies [e0f6cc3]
+  - @llamaindex/env@0.1.25
+  - @llamaindex/core@0.4.18
+
+## 0.0.75
+
+### Patch Changes
+
+- Updated dependencies [3d1808b]
+  - @llamaindex/core@0.4.17
+
+## 0.0.74
+
+### Patch Changes
+
+- 8be4589: chore: bump version
+- Updated dependencies [8be4589]
+  - @llamaindex/core@0.4.16
+  - @llamaindex/env@0.1.24
+
+## 0.0.73
+
+### Patch Changes
+
+- Updated dependencies [d2b2722]
+  - @llamaindex/env@0.1.23
+  - @llamaindex/core@0.4.15
+
+## 0.0.72
+
+### Patch Changes
+
+- Updated dependencies [969365c]
+  - @llamaindex/env@0.1.22
+  - @llamaindex/core@0.4.14
+
+## 0.0.71
+
+### Patch Changes
+
+- 90d265c: chore: bump version
+- Updated dependencies [90d265c]
+  - @llamaindex/core@0.4.13
+  - @llamaindex/env@0.1.21
+
+## 0.0.70
+
+### Patch Changes
+
+- Updated dependencies [ef4f63d]
+  - @llamaindex/core@0.4.12
+
+## 0.0.69
+
+### Patch Changes
+
+- Updated dependencies [6d22fa2]
+  - @llamaindex/core@0.4.11
+
+## 0.0.68
+
+### Patch Changes
+
+- Updated dependencies [a7b0ac3]
+- Updated dependencies [c69605f]
+  - @llamaindex/core@0.4.10
+
+## 0.0.67
+
+### Patch Changes
+
+- Updated dependencies [7ae6eaa]
+  - @llamaindex/core@0.4.9
+
+## 0.0.66
+
+### Patch Changes
+
+- Updated dependencies [f865c98]
+  - @llamaindex/core@0.4.8
+
+## 0.0.65
+
+### Patch Changes
+
+- Updated dependencies [d89ebe0]
+- Updated dependencies [fd8c882]
+  - @llamaindex/core@0.4.7
+
+## 0.0.64
+
+### Patch Changes
+
+- Updated dependencies [4fc001c]
+  - @llamaindex/env@0.1.20
+  - @llamaindex/core@0.4.6
+
+## 0.0.63
+
+### Patch Changes
+
+- Updated dependencies [ad85bd0]
+  - @llamaindex/core@0.4.5
+  - @llamaindex/env@0.1.19
+
+## 0.0.62
+
+### Patch Changes
+
+- Updated dependencies [a8d3fa6]
+  - @llamaindex/env@0.1.18
+  - @llamaindex/core@0.4.4
+
+## 0.0.61
+
+### Patch Changes
+
+- 487782c: Add missing inference endpoints for Haiku 3.5
+- Updated dependencies [95a5cc6]
+  - @llamaindex/core@0.4.3
+
+## 0.0.60
+
+### Patch Changes
+
+- Updated dependencies [14cc9eb]
+  - @llamaindex/env@0.1.17
+  - @llamaindex/core@0.4.2
+
+## 0.0.59
+
+### Patch Changes
+
+- 47a7c3e: feat: added support for Haiku 3.5 via Bedrock
+
+## 0.0.58
+
+### Patch Changes
+
+- Updated dependencies [9c73f0a]
+  - @llamaindex/core@0.4.1
+
+## 0.0.57
+
+### Patch Changes
+
+- Updated dependencies [359fd33]
+- Updated dependencies [efb7e1b]
+- Updated dependencies [98ba1e7]
+- Updated dependencies [620c63c]
+  - @llamaindex/core@0.4.0
+
+## 0.0.56
+
+### Patch Changes
+
+- Updated dependencies [60b185f]
+  - @llamaindex/core@0.3.7
+
+## 0.0.55
+
+### Patch Changes
+
+- Updated dependencies [691c5bc]
+  - @llamaindex/core@0.3.6
+
+## 0.0.54
+
+### Patch Changes
+
+- Updated dependencies [fa60fc6]
+  - @llamaindex/env@0.1.16
+  - @llamaindex/core@0.3.5
+
+## 0.0.53
+
+### Patch Changes
+
+- Updated dependencies [e2a0876]
+  - @llamaindex/core@0.3.4
+
+## 0.0.52
+
+### Patch Changes
+
+- a5a75f6: feat: added sonnet 3.5 v2
+
+## 0.0.51
+
+### Patch Changes
+
+- Updated dependencies [0493f67]
+  - @llamaindex/core@0.3.3
+
+## 0.0.50
+
+### Patch Changes
+
+- Updated dependencies [4ba2cfe]
+  - @llamaindex/env@0.1.15
+  - @llamaindex/core@0.3.2
+
+## 0.0.49
+
+### Patch Changes
+
+- a75af83: refactor: move some llm and embedding to single package
+- Updated dependencies [ae49ff4]
+- Updated dependencies [a75af83]
+  - @llamaindex/env@0.1.14
+  - @llamaindex/core@0.3.1
+
+## 0.0.48
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [96fc69c]
+  - @llamaindex/core@0.3.0
+
+## 0.0.47
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+  - @llamaindex/core@0.2.12
+
+## 0.0.46
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - @llamaindex/core@0.2.11
+
+## 0.0.45
+
+### Patch Changes
+
+- Updated dependencies [3489e7d]
+- Updated dependencies [468bda5]
+  - @llamaindex/core@0.2.10
+
+## 0.0.44
+
+### Patch Changes
+
+- Updated dependencies [b17d439]
+  - @llamaindex/core@0.2.9
+
+## 0.0.43
+
+### Patch Changes
+
+- 2774e80: feat: added meta3.2 support via Bedrock including vision, tool call and inference region support
+
+## 0.0.42
+
+### Patch Changes
+
+- df441e2: fix: consoleLogger is missing from `@llamaindex/env`
+- Updated dependencies [df441e2]
+  - @llamaindex/core@0.2.8
+  - @llamaindex/env@0.1.13
+
+## 0.0.41
+
+### Patch Changes
+
+- Updated dependencies [6cce3b1]
+  - @llamaindex/core@0.2.7
+
+## 0.0.40
+
+### Patch Changes
+
+- 50e6b57: feat: add Amazon Bedrock Retriever
+- Updated dependencies [8b7fdba]
+  - @llamaindex/core@0.2.6
+
+## 0.0.39
+
+### Patch Changes
+
+- Updated dependencies [d902cc3]
+  - @llamaindex/core@0.2.5
+
+## 0.0.38
+
+### Patch Changes
+
+- Updated dependencies [b48bcc3]
+  - @llamaindex/core@0.2.4
+  - @llamaindex/env@0.1.12
+
+## 0.0.37
+
+### Patch Changes
+
+- Updated dependencies [2cd1383]
+  - @llamaindex/core@0.2.3
+
+## 0.0.36
+
+### Patch Changes
+
+- Updated dependencies [749b43a]
+  - @llamaindex/core@0.2.2
+
+## 0.0.35
+
+### Patch Changes
+
+- Updated dependencies [ac07e3c]
+- Updated dependencies [70ccb4a]
+- Updated dependencies [1a6137b]
+- Updated dependencies [ac07e3c]
+  - @llamaindex/core@0.2.1
+  - @llamaindex/env@0.1.11
+
+## 0.0.34
+
+### Patch Changes
+
+- Updated dependencies [11feef8]
+  - @llamaindex/core@0.2.0
+
+## 0.0.33
+
+### Patch Changes
+
+- Updated dependencies [711c814]
+  - @llamaindex/core@0.1.12
+
+## 0.0.32
+
+### Patch Changes
+
+- Updated dependencies [4648da6]
+  - @llamaindex/env@0.1.10
+  - @llamaindex/core@0.1.11
+
+## 0.0.31
+
+### Patch Changes
+
+- Updated dependencies [0148354]
+  - @llamaindex/core@0.1.10
+
+## 0.0.30
+
+### Patch Changes
+
+- Updated dependencies [e27e7dd]
+  - @llamaindex/core@0.1.9
+
+## 0.0.29
+
+### Patch Changes
+
+- 58abc57: fix: align version
+- Updated dependencies [58abc57]
+  - @llamaindex/core@0.1.8
+  - @llamaindex/env@0.1.9
+
+## 0.0.28
+
+### Patch Changes
+
+- Updated dependencies [04b2f8e]
+  - @llamaindex/core@0.1.7
+
+## 0.0.27
+
+### Patch Changes
+
+- Updated dependencies [0452af9]
+  - @llamaindex/core@0.1.6
+
+## 0.0.26
+
+### Patch Changes
+
+- 224d507: fix: prevent tool calling getting mixed with conversation
+- 376d29a: feat: added tool calling and agent support for llama3.1 504B
+
+## 0.0.25
+
+### Patch Changes
+
+- Updated dependencies [91d02a4]
+  - @llamaindex/core@0.1.5
+
+## 0.0.24
+
+### Patch Changes
+
+- 3d9a802: feat: added llama 3.1
+- Updated dependencies [15962b3]
+  - @llamaindex/core@0.1.4
+
+## 0.0.23
+
+### Patch Changes
+
+- Updated dependencies [6cf6ae6]
+  - @llamaindex/core@0.1.3
+
+## 0.0.22
+
+### Patch Changes
+
+- Updated dependencies [b974eea]
+  - @llamaindex/core@0.1.2
+
+## 0.0.21
+
+### Patch Changes
+
+- Updated dependencies [b3681bf]
+  - @llamaindex/core@0.1.1
+
+## 0.0.20
+
+### Patch Changes
+
+- 56746c2: fix: llama3 patched to handle empty content (can happen with system) and added max tokens export
+
+## 0.0.19
+
+### Patch Changes
+
+- 16ef5dd: refactor: depends on core pacakge instead of llamaindex
+- Updated dependencies [16ef5dd]
+- Updated dependencies [16ef5dd]
+  - @llamaindex/core@0.1.0
+
+## 0.0.18
+
+### Patch Changes
+
+- llamaindex@0.4.14
+
+## 0.0.17
+
+### Patch Changes
+
+- Updated dependencies [e8f8bea]
+- Updated dependencies [304484b]
+  - llamaindex@0.4.13
+
+## 0.0.16
+
+### Patch Changes
+
+- f326ab8: chore: bump version
+- Updated dependencies [f326ab8]
+  - llamaindex@0.4.12
+
+## 0.0.15
+
+### Patch Changes
+
+- Updated dependencies [8bf5b4a]
+  - llamaindex@0.4.11
+
+## 0.0.14
+
+### Patch Changes
+
+- Updated dependencies [7dce3d2]
+  - llamaindex@0.4.10
+
+## 0.0.13
+
+### Patch Changes
+
+- Updated dependencies [3a96a48]
+  - llamaindex@0.4.9
+
+## 0.0.12
+
+### Patch Changes
+
+- Updated dependencies [83ebdfb]
+  - llamaindex@0.4.8
+
+## 0.0.11
+
+### Patch Changes
+
+- Updated dependencies [41fe871]
+- Updated dependencies [321c39d]
+- Updated dependencies [f7f1af0]
+  - llamaindex@0.4.7
+
+## 0.0.10
+
+### Patch Changes
+
+- Updated dependencies [1feb23b]
+- Updated dependencies [08c55ec]
+  - llamaindex@0.4.6
+
+## 0.0.9
+
+### Patch Changes
+
+- Updated dependencies [6c3e5d0]
+  - llamaindex@0.4.5
+
+## 0.0.8
+
+### Patch Changes
+
+- Updated dependencies [42eb73a]
+  - llamaindex@0.4.4
+
+## 0.0.7
+
+### Patch Changes
+
+- Updated dependencies [2ef62a9]
+  - llamaindex@0.4.3
+
+## 0.0.6
+
+### Patch Changes
+
+- a87a4d1: feat: added tool support calling for Bedrock's Calude and general llm support for agents
+- Updated dependencies [a87a4d1]
+- Updated dependencies [0730140]
+  - llamaindex@0.4.2
+
+## 0.0.5
+
+### Patch Changes
+
+- ed467a9: Add model ids for Anthropic Claude 3.5 Sonnet model on Anthropic and Bedrock
+- Updated dependencies [3c47910]
+- Updated dependencies [ed467a9]
+- Updated dependencies [cba5406]
+  - llamaindex@0.4.1
+
+## 0.0.4
+
+### Patch Changes
+
+- b1a4a74: docs: updated Bedrock Opus region and added a basic README
+- Updated dependencies [436bc41]
+- Updated dependencies [a44e54f]
+- Updated dependencies [a51ed8d]
+- Updated dependencies [d3b635b]
+  - llamaindex@0.4.0
+
+## 0.0.3
+
+### Patch Changes
+
+- Updated dependencies [6bc5bdd]
+- Updated dependencies [bf25ff6]
+- Updated dependencies [e6d6576]
+  - llamaindex@0.3.17
+
+## 0.0.2
+
+### Patch Changes
+
+- 8832669: Community bedrock support added
+- Updated dependencies [11ae926]
+- Updated dependencies [631f000]
+- Updated dependencies [1378ec4]
+- Updated dependencies [6b1ded4]
+- Updated dependencies [4d4bd85]
+- Updated dependencies [24a9d1e]
+- Updated dependencies [45952de]
+- Updated dependencies [54230f0]
+- Updated dependencies [a29d835]
+- Updated dependencies [73819bf]
+  - llamaindex@0.3.16
@@ -0,0 +1,17 @@
+# @llamaindex/community
+
+AWS package for LlamaIndexTS, deprecated, use [@llamaindex/aws](https://www.npmjs.com/package/@llamaindex/aws) instead.
+
+## Current Features:
+
+- Bedrock support for Amazon Nova models Pro, Lite and Micro
+- Bedrock support for the Anthropic Claude Models [usage](https://ts.llamaindex.ai/docs/llamaindex/modules/llms/bedrock) including the latest Sonnet 3.5 v2 and Haiku 3.5
+- Bedrock support for the Meta LLama 2, 3, 3.1 and 3.2 Models [usage](https://ts.llamaindex.ai/docs/llamaindex/modules/llms/bedrock)
+- Meta LLama3.1 405b and Llama3.2 tool call support
+- Meta 3.2 11B and 90B vision support
+- Bedrock support for querying Knowledge Base
+- Bedrock: [Supported Regions and models for cross-region inference](https://docs.aws.amazon.com/bedrock/latest/userguide/cross-region-inference-support.html)
+
+## LICENSE
+
+MIT
@@ -0,0 +1,53 @@
+{
+  "name": "@llamaindex/community",
+  "description": "Community package for LlamaIndexTS",
+  "version": "0.0.100",
+  "type": "module",
+  "types": "dist/type/index.d.ts",
+  "main": "dist/cjs/index.js",
+  "exports": {
+    ".": {
+      "import": {
+        "types": "./dist/type/index.d.ts",
+        "default": "./dist/index.js"
+      },
+      "require": {
+        "types": "./dist/type/index.d.ts",
+        "default": "./dist/index.cjs"
+      }
+    },
+    "./llm/bedrock": {
+      "import": {
+        "types": "./dist/type/llm/bedrock.d.ts",
+        "default": "./dist/llm/bedrock/index.js"
+      },
+      "require": {
+        "types": "./dist/type/llm/bedrock.d.ts",
+        "default": "./dist/llm/bedrock/index.cjs"
+      }
+    }
+  },
+  "files": [
+    "dist",
+    "CHANGELOG.md",
+    "!**/*.tsbuildinfo"
+  ],
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/run-llama/LlamaIndexTS.git",
+    "directory": "packages/community"
+  },
+  "scripts": {
+    "build": "bunchee",
+    "dev": "bunchee --watch"
+  },
+  "devDependencies": {
+    "@types/node": "^22.9.0"
+  },
+  "dependencies": {
+    "@aws-sdk/client-bedrock-agent-runtime": "^3.706.0",
+    "@aws-sdk/client-bedrock-runtime": "^3.706.0",
+    "@llamaindex/core": "workspace:*",
+    "@llamaindex/env": "workspace:*"
+  }
+}
@@ -0,0 +1,8 @@
+export {
+  BEDROCK_MODELS,
+  BEDROCK_MODEL_MAX_TOKENS,
+  Bedrock,
+  INFERENCE_BEDROCK_MODELS,
+  INFERENCE_TO_BEDROCK_MAP,
+} from "./llm/bedrock/index.js";
+export { AmazonKnowledgeBaseRetriever } from "./retrievers/bedrock.js";
@@ -0,0 +1,134 @@
+import type {
+  ContentBlockDelta,
+  ConverseOutput,
+  ConverseRequest,
+  ConverseResponse,
+  ConverseStreamOutput,
+  InvokeModelCommandInput,
+  InvokeModelWithResponseStreamCommandInput,
+  ResponseStream,
+} from "@aws-sdk/client-bedrock-runtime";
+import type {
+  BaseTool,
+  ChatMessage,
+  LLMMetadata,
+  ToolCall,
+  ToolCallLLMMessageOptions,
+} from "@llamaindex/core/llms";
+import { toUtf8 } from "../utils";
+
+import { Provider, type BedrockChatStreamResponse } from "../provider";
+import {
+  mapBaseToolsToAmazonTools,
+  mapChatMessagesToAmazonMessages,
+} from "./utils";
+
+export class AmazonProvider extends Provider<ConverseStreamOutput> {
+  // eslint-disable-next-line @typescript-eslint/no-explicit-any
+  getResultFromResponse(response: Record<string, any>): ConverseResponse {
+    return JSON.parse(toUtf8(response.body));
+  }
+
+  getToolsFromResponse<ToolContent>(response: ConverseOutput): ToolContent[] {
+    return (
+      response.message?.content
+        ?.filter((item) => item.toolUse)
+        .map(
+          (item) =>
+            ({
+              id: item.toolUse!.toolUseId,
+              name: item.toolUse!.name,
+              input: item.toolUse!.input
+                ? JSON.parse(item.toolUse!.input as string)
+                : "",
+            }) as ToolContent,
+        ) ?? []
+    );
+  }
+
+  getTextFromResponse(response: ConverseResponse): string {
+    const result = this.getResultFromResponse(response);
+    const content = result.output?.message?.content ?? [];
+    return content.map((item) => item.text).join(" ");
+  }
+
+  getTextFromStreamResponse(response: ResponseStream): string {
+    const event: ConverseStreamOutput | undefined =
+      this.getStreamingEventResponse(response);
+    if (!event || !event.contentBlockDelta) return "";
+    const delta: ContentBlockDelta | undefined = event.contentBlockDelta.delta;
+    return delta?.text || "";
+  }
+
+  async *reduceStream(
+    stream: AsyncIterable<ResponseStream>,
+  ): BedrockChatStreamResponse {
+    let toolId: string | undefined = undefined;
+    let toolName: string | undefined = undefined;
+    for await (const response of stream) {
+      const event = this.getStreamingEventResponse(response);
+      const delta = this.getTextFromStreamResponse(response);
+
+      let options: undefined | ToolCallLLMMessageOptions = undefined;
+      if (event?.contentBlockStart && event.contentBlockStart.start?.toolUse) {
+        toolId = event.contentBlockStart.start?.toolUse.toolUseId;
+        toolName = event.contentBlockStart.start?.toolUse.name;
+        continue;
+      }
+      if (
+        toolId &&
+        toolName &&
+        event?.contentBlockDelta?.delta?.toolUse?.input
+      ) {
+        options = {
+          toolCall: [
+            {
+              id: toolId,
+              name: toolName,
+              input: JSON.parse(event?.contentBlockDelta?.delta?.toolUse.input),
+            } as ToolCall,
+          ],
+        };
+        toolId = undefined;
+        toolName = undefined;
+      }
+
+      if (!delta && !options) continue;
+
+      yield {
+        delta: options ? "" : delta,
+        options,
+        raw: response,
+      };
+    }
+  }
+
+  getRequestBody<T extends ChatMessage>(
+    metadata: LLMMetadata,
+    messages: T[],
+    tools: BaseTool[] = [],
+    options: Omit<ConverseRequest, "modelId" | "messages" | "inferenceConfig">,
+  ): InvokeModelCommandInput | InvokeModelWithResponseStreamCommandInput {
+    const request: Omit<ConverseRequest, "modelId"> = {
+      ...options,
+      messages: mapChatMessagesToAmazonMessages(messages),
+      inferenceConfig: {
+        maxTokens: metadata.maxTokens,
+        temperature: metadata.temperature,
+        topP: metadata.topP,
+      },
+    };
+    if (tools.length) {
+      request.toolConfig = {
+        tools: mapBaseToolsToAmazonTools(tools),
+      };
+    }
+
+    return {
+      modelId: metadata.model,
+      contentType: "application/json",
+      accept: "application/json",
+      body: JSON.stringify(request),
+    };
+  }
+}
@@ -0,0 +1,5 @@
+import type { ConverseRequest, Message } from "@aws-sdk/client-bedrock-runtime";
+
+export type AmazonMessages = ConverseRequest["messages"];
+
+export type AmazonMessage = Message;
@@ -0,0 +1,141 @@
+import type {
+  ImageBlock,
+  ImageFormat,
+  Message,
+  Tool,
+} from "@aws-sdk/client-bedrock-runtime";
+import type {
+  BaseTool,
+  ChatMessage,
+  MessageContentDetail,
+  ToolCallLLMMessageOptions,
+} from "@llamaindex/core/llms";
+import { extractDataUrlComponents } from "../utils";
+
+import type { JSONObject } from "@llamaindex/core/global";
+import { mapMessageContentToMessageContentDetails } from "../../utils";
+import type { AmazonMessage, AmazonMessages } from "./types";
+
+const ACCEPTED_IMAGE_MIME_TYPES = [
+  "image/jpeg",
+  "image/png",
+  "image/webp",
+  "image/gif",
+] as const;
+
+const ACCEPTED_IMAGE_MIME_TYPE_FORMAT_MAP: Record<
+  (typeof ACCEPTED_IMAGE_MIME_TYPES)[number],
+  ImageFormat
+> = {
+  "image/jpeg": "jpeg",
+  "image/png": "png",
+  "image/webp": "webp",
+  "image/gif": "gif",
+};
+
+export const mapImageContent = (imageUrl: string): ImageBlock => {
+  if (!imageUrl.startsWith("data:"))
+    throw new Error(
+      "For Amazon please only use base64 data url, e.g.: data:image/jpeg;base64,SGVsbG8sIFdvcmxkIQ==",
+    );
+  const { mimeType, base64: data } = extractDataUrlComponents(imageUrl);
+  if (
+    !ACCEPTED_IMAGE_MIME_TYPES.includes(
+      mimeType as keyof typeof ACCEPTED_IMAGE_MIME_TYPE_FORMAT_MAP,
+    )
+  )
+    throw new Error(
+      `Amazon only accepts the following mimeTypes: ${ACCEPTED_IMAGE_MIME_TYPES.join("\n")}`,
+    );
+
+  return {
+    format:
+      ACCEPTED_IMAGE_MIME_TYPE_FORMAT_MAP[
+        mimeType as keyof typeof ACCEPTED_IMAGE_MIME_TYPE_FORMAT_MAP
+      ],
+
+    // @ts-expect-error: there's a mistake in the "@aws-sdk/client-bedrock-runtime" compared to the actual api
+    source: { bytes: data },
+  };
+};
+
+export const mapMessageContentDetailToAmazonContent = <
+  T extends MessageContentDetail,
+>(
+  detail: T,
+): Message["content"] => {
+  let content: Message["content"] = [];
+
+  if (detail.type === "text") {
+    content = [{ text: detail.text }];
+  } else if (detail.type === "image_url") {
+    content = [{ image: mapImageContent(detail.image_url.url) }];
+  } else {
+    throw new Error("Unsupported content detail type");
+  }
+  return content;
+};
+
+export const mapChatMessagesToAmazonMessages = <
+  T extends ChatMessage<ToolCallLLMMessageOptions>,
+>(
+  messages: T[],
+): AmazonMessages => {
+  return messages.flatMap((msg: T): AmazonMessage[] => {
+    return mapMessageContentToMessageContentDetails(msg.content).map(
+      (detail: MessageContentDetail): AmazonMessage => {
+        if (msg.options && "toolCall" in msg.options) {
+          return {
+            role: "assistant",
+            content: msg.options.toolCall.map((call) => ({
+              toolUse: {
+                toolUseId: call.id,
+                name: call.name,
+                input: call.input as JSONObject,
+              },
+            })),
+          };
+        }
+        if (msg.options && "toolResult" in msg.options) {
+          return {
+            role: "user",
+            content: [
+              {
+                toolResult: {
+                  toolUseId: msg.options.toolResult.id,
+                  content: [
+                    {
+                      text: msg.options.toolResult.result,
+                    },
+                  ],
+                },
+              },
+            ],
+          };
+        }
+
+        return {
+          role: msg.role === "assistant" ? "assistant" : "user",
+          content: mapMessageContentDetailToAmazonContent(detail),
+        };
+      },
+    );
+  });
+};
+
+export const mapBaseToolsToAmazonTools = (tools?: BaseTool[]): Tool[] => {
+  if (!tools) return [];
+  return tools.map((tool: BaseTool) => {
+    const {
+      metadata: { parameters, ...options },
+    } = tool;
+    return {
+      toolSpec: {
+        ...options,
+        inputSchema: {
+          json: parameters,
+        },
+      },
+    } as Tool;
+  });
+};
@@ -0,0 +1,156 @@
+import {
+  type InvokeModelCommandInput,
+  type InvokeModelWithResponseStreamCommandInput,
+  ResponseStream,
+} from "@aws-sdk/client-bedrock-runtime";
+import type {
+  BaseTool,
+  ChatMessage,
+  LLMMetadata,
+  PartialToolCall,
+  ToolCall,
+  ToolCallLLMMessageOptions,
+} from "@llamaindex/core/llms";
+import { type BedrockChatStreamResponse, Provider } from "../provider";
+import { toUtf8 } from "../utils";
+import type {
+  AnthropicAdditionalChatOptions,
+  AnthropicNoneStreamingResponse,
+  AnthropicStreamEvent,
+  AnthropicTextContent,
+  ToolBlock,
+} from "./types";
+
+import {
+  mapBaseToolsToAnthropicTools,
+  mapChatMessagesToAnthropicMessages,
+} from "./utils";
+
+export class AnthropicProvider extends Provider<AnthropicStreamEvent> {
+  getResultFromResponse(
+    // eslint-disable-next-line @typescript-eslint/no-explicit-any
+    response: Record<string, any>,
+  ): AnthropicNoneStreamingResponse {
+    return JSON.parse(toUtf8(response.body));
+  }
+
+  getToolsFromResponse<AnthropicToolContent>(
+    // eslint-disable-next-line @typescript-eslint/no-explicit-any
+    response: Record<string, any>,
+  ): AnthropicToolContent[] {
+    const result = this.getResultFromResponse(response);
+    return result.content
+      .filter((item) => item.type === "tool_use")
+      .map((item) => item as AnthropicToolContent);
+  }
+
+  // eslint-disable-next-line @typescript-eslint/no-explicit-any
+  getTextFromResponse(response: Record<string, any>): string {
+    const result = this.getResultFromResponse(response);
+    return result.content
+      .filter((item) => item.type === "text")
+      .map((item) => (item as AnthropicTextContent).text)
+      .join(" ");
+  }
+
+  // eslint-disable-next-line @typescript-eslint/no-explicit-any
+  getTextFromStreamResponse(response: Record<string, any>): string {
+    const event = this.getStreamingEventResponse(response);
+    if (event?.type === "content_block_delta") {
+      if (event.delta.type === "text_delta") return event.delta.text;
+      if (event.delta.type === "input_json_delta")
+        return event.delta.partial_json;
+    }
+    return "";
+  }
+
+  async *reduceStream(
+    stream: AsyncIterable<ResponseStream>,
+  ): BedrockChatStreamResponse {
+    let collecting = [];
+    let tool: ToolBlock | undefined = undefined;
+    // #TODO this should be broken down into a separate consumer
+    for await (const response of stream) {
+      const delta = this.getTextFromStreamResponse(response);
+      const event = this.getStreamingEventResponse(response);
+      if (
+        event?.type === "content_block_start" &&
+        event.content_block.type === "tool_use"
+      ) {
+        tool = event.content_block;
+        continue;
+      }
+
+      if (
+        event?.type === "content_block_delta" &&
+        event.delta.type === "input_json_delta"
+      ) {
+        collecting.push(event.delta.partial_json);
+      }
+
+      let options: undefined | ToolCallLLMMessageOptions = undefined;
+      if (tool && collecting.length) {
+        const input = collecting.filter((item) => item).join("");
+        // We have all we need to parse the tool_use json
+        if (event?.type === "content_block_stop") {
+          options = {
+            toolCall: [
+              {
+                id: tool.id,
+                name: tool.name,
+                input: JSON.parse(input),
+              } as ToolCall,
+            ],
+          };
+          // reset the collection/tool
+          collecting = [];
+          tool = undefined;
+        } else {
+          options = {
+            toolCall: [
+              {
+                id: tool.id,
+                name: tool.name,
+                input,
+              } as PartialToolCall,
+            ],
+          };
+        }
+      }
+      if (!delta && !options) continue;
+
+      yield {
+        delta: options ? "" : delta,
+        options,
+        raw: response,
+      };
+    }
+  }
+
+  getRequestBody<T extends ChatMessage<ToolCallLLMMessageOptions>>(
+    metadata: LLMMetadata,
+    messages: T[],
+    tools?: BaseTool[],
+    options?: AnthropicAdditionalChatOptions,
+  ): InvokeModelCommandInput | InvokeModelWithResponseStreamCommandInput {
+    const extra: Record<string, unknown> = {};
+    if (options?.toolChoice) {
+      extra["tool_choice"] = options?.toolChoice;
+    }
+    const mapped = mapChatMessagesToAnthropicMessages(messages);
+    return {
+      modelId: metadata.model,
+      contentType: "application/json",
+      accept: "application/json",
+      body: JSON.stringify({
+        anthropic_version: "bedrock-2023-05-31",
+        messages: mapped,
+        tools: mapBaseToolsToAnthropicTools(tools),
+        max_tokens: metadata.maxTokens,
+        temperature: metadata.temperature,
+        top_p: metadata.topP,
+        ...extra,
+      }),
+    };
+  }
+}
@@ -0,0 +1,161 @@
+import type { ToolMetadata } from "@llamaindex/core/llms";
+import type { InvocationMetrics } from "../types";
+
+export type ToolChoice =
+  | { type: "any" }
+  | { type: "auto" }
+  | { type: "tool"; name: string };
+
+export interface ThinkingConfigDisabled {
+  type: "disabled";
+}
+
+export interface ThinkingConfigEnabled {
+  budget_tokens: number;
+  type: "enabled";
+}
+
+export type AnthropicAdditionalChatOptions = {
+  toolChoice: ToolChoice;
+  thinking?: ThinkingConfigDisabled | ThinkingConfigEnabled;
+};
+
+type Usage = {
+  input_tokens: number;
+  output_tokens: number;
+};
+
+type Message = {
+  id: string;
+  type: string;
+  role: string;
+  content: string[];
+  model: string;
+  stop_reason: string | null;
+  stop_sequence: string | null;
+  usage: Usage;
+};
+
+export type ToolBlock = {
+  id: string;
+  input: unknown;
+  name: string;
+  type: "tool_use";
+};
+
+export type TextBlock = {
+  type: "text";
+  text: string;
+};
+
+type ContentBlockStart = {
+  type: "content_block_start";
+  index: number;
+  content_block: ToolBlock | TextBlock;
+};
+
+type Delta =
+  | {
+      type: "text_delta";
+      text: string;
+    }
+  | {
+      type: "input_json_delta";
+      partial_json: string;
+    };
+
+type ContentBlockDelta = {
+  type: "content_block_delta";
+  index: number;
+  delta: Delta;
+};
+
+type ContentBlockStop = {
+  type: "content_block_stop";
+  index: number;
+};
+
+type MessageDelta = {
+  type: "message_delta";
+  delta: {
+    stop_reason: string;
+    stop_sequence: string | null;
+  };
+  usage: Usage;
+};
+
+export type MessageStop = {
+  type: "message_stop";
+  "amazon-bedrock-invocationMetrics": InvocationMetrics;
+};
+
+export type AnthropicStreamEvent =
+  | { type: "message_start"; message: Message }
+  | ContentBlockStart
+  | ContentBlockDelta
+  | ContentBlockStop
+  | MessageDelta
+  | MessageStop;
+
+export type AnthropicContent =
+  | AnthropicTextContent
+  | AnthropicImageContent
+  | AnthropicToolContent
+  | AnthropicToolResultContent;
+
+export type AnthropicTextContent = {
+  type: "text";
+  text: string;
+};
+
+export type AnthropicToolContent = {
+  type: "tool_use";
+  id: string;
+  name: string;
+  input: Record<string, unknown>;
+};
+
+export type AnthropicToolResultContent = {
+  type: "tool_result";
+  tool_use_id: string;
+  content: string;
+};
+
+export type AnthropicMediaTypes =
+  | "image/jpeg"
+  | "image/png"
+  | "image/webp"
+  | "image/gif";
+
+export type AnthropicImageSource = {
+  type: "base64";
+  media_type: AnthropicMediaTypes;
+  data: string; // base64 encoded image bytes
+};
+
+export type AnthropicImageContent = {
+  type: "image";
+  source: AnthropicImageSource;
+};
+
+export type AnthropicMessage = {
+  role: "user" | "assistant";
+  content: AnthropicContent[];
+};
+
+export type AnthropicNoneStreamingResponse = {
+  id: string;
+  type: "message";
+  role: "assistant";
+  content: AnthropicContent[];
+  model: string;
+  stop_reason: "end_turn" | "max_tokens" | "stop_sequence";
+  stop_sequence?: string;
+  usage: { input_tokens: number; output_tokens: number };
+};
+
+export type AnthropicTool = {
+  name: string;
+  description: string;
+  input_schema: ToolMetadata["parameters"];
+};
@@ -0,0 +1,166 @@
+import type { JSONObject } from "@llamaindex/core/global";
+import type {
+  BaseTool,
+  ChatMessage,
+  MessageContent,
+  MessageContentDetail,
+  ToolCallLLMMessageOptions,
+} from "@llamaindex/core/llms";
+import { mapMessageContentToMessageContentDetails } from "../../utils";
+import { extractDataUrlComponents } from "../utils";
+import type {
+  AnthropicContent,
+  AnthropicImageContent,
+  AnthropicMediaTypes,
+  AnthropicMessage,
+  AnthropicTextContent,
+  AnthropicTool,
+} from "./types.js";
+
+const ACCEPTED_IMAGE_MIME_TYPES = [
+  "image/jpeg",
+  "image/png",
+  "image/webp",
+  "image/gif",
+];
+
+export const mergeNeighboringSameRoleMessages = (
+  messages: AnthropicMessage[],
+): AnthropicMessage[] => {
+  return messages.reduce(
+    (result: AnthropicMessage[], current: AnthropicMessage, index: number) => {
+      if (index > 0 && messages[index - 1]!.role === current.role) {
+        result[result.length - 1]!.content = [
+          ...result[result.length - 1]!.content,
+          ...current.content,
+        ];
+      } else {
+        result.push(current);
+      }
+      return result;
+    },
+    [],
+  );
+};
+
+export const mapMessageContentDetailToAnthropicContent = <
+  T extends MessageContentDetail,
+>(
+  detail: T,
+): AnthropicContent => {
+  let content: AnthropicContent;
+
+  if (detail.type === "text") {
+    content = mapTextContent(detail.text);
+  } else if (detail.type === "image_url") {
+    content = mapImageContent(detail.image_url.url);
+  } else {
+    throw new Error("Unsupported content detail type");
+  }
+  return content;
+};
+
+export const mapMessageContentToAnthropicContent = <T extends MessageContent>(
+  content: T,
+): AnthropicContent[] => {
+  return mapMessageContentToMessageContentDetails(content).map(
+    mapMessageContentDetailToAnthropicContent,
+  );
+};
+
+export const mapBaseToolsToAnthropicTools = (
+  tools?: BaseTool[],
+): AnthropicTool[] => {
+  if (!tools) return [];
+  return tools.map((tool: BaseTool) => {
+    const {
+      metadata: { parameters, ...options },
+    } = tool;
+    return {
+      ...options,
+      input_schema: parameters,
+    };
+  });
+};
+
+export const mapChatMessagesToAnthropicMessages = <
+  T extends ChatMessage<ToolCallLLMMessageOptions>,
+>(
+  messages: T[],
+): AnthropicMessage[] => {
+  const mapped = messages
+    .flatMap((msg: T): AnthropicMessage[] => {
+      if (msg.options && "toolCall" in msg.options) {
+        return [
+          {
+            role: "assistant",
+            content: msg.options.toolCall.map((call) => ({
+              type: "tool_use",
+              id: call.id,
+              name: call.name,
+              input: call.input as JSONObject,
+            })),
+          },
+        ];
+      }
+      if (msg.options && "toolResult" in msg.options) {
+        return [
+          {
+            role: "user",
+            content: [
+              {
+                type: "tool_result",
+                tool_use_id: msg.options.toolResult.id,
+                content: JSON.stringify(msg.options.toolResult.result),
+              },
+            ],
+          },
+        ];
+      }
+      return mapMessageContentToMessageContentDetails(msg.content).map(
+        (detail: MessageContentDetail): AnthropicMessage => {
+          const content = mapMessageContentDetailToAnthropicContent(detail);
+
+          return {
+            role: msg.role === "assistant" ? "assistant" : "user",
+            content: [content],
+          };
+        },
+      );
+    })
+    .filter((message: AnthropicMessage) => {
+      const content = message.content[0]!;
+      if (content.type === "text" && !content.text) return false;
+      if (content.type === "image" && !content.source.data) return false;
+      if (content.type === "image" && message.role === "assistant")
+        return false;
+      return true;
+    });
+
+  return mergeNeighboringSameRoleMessages(mapped);
+};
+
+export const mapTextContent = (text: string): AnthropicTextContent => {
+  return { type: "text", text };
+};
+
+export const mapImageContent = (imageUrl: string): AnthropicImageContent => {
+  if (!imageUrl.startsWith("data:"))
+    throw new Error(
+      "For Anthropic please only use base64 data url, e.g.: data:image/jpeg;base64,SGVsbG8sIFdvcmxkIQ==",
+    );
+  const { mimeType, base64: data } = extractDataUrlComponents(imageUrl);
+  if (!ACCEPTED_IMAGE_MIME_TYPES.includes(mimeType))
+    throw new Error(
+      `Anthropic only accepts the following mimeTypes: ${ACCEPTED_IMAGE_MIME_TYPES.join("\n")}`,
+    );
+
+  return {
+    type: "image",
+    source: {
+      type: "base64",
+      media_type: mimeType as AnthropicMediaTypes,
+      data,
+    },
+  };
+};
@@ -0,0 +1,513 @@
+import {
+  BedrockRuntimeClient,
+  type BedrockRuntimeClientConfig,
+  InvokeModelCommand,
+  InvokeModelWithResponseStreamCommand,
+} from "@aws-sdk/client-bedrock-runtime";
+import {
+  type ChatMessage,
+  type ChatResponse,
+  type CompletionResponse,
+  type LLMChatParamsNonStreaming,
+  type LLMChatParamsStreaming,
+  type LLMCompletionParamsNonStreaming,
+  type LLMCompletionParamsStreaming,
+  type LLMMetadata,
+  ToolCallLLM,
+  type ToolCallLLMMessageOptions,
+} from "@llamaindex/core/llms";
+import { streamConverter } from "@llamaindex/core/utils";
+import {
+  type BedrockAdditionalChatOptions,
+  type BedrockChatStreamResponse,
+  Provider,
+} from "./provider";
+
+import { wrapLLMEvent } from "@llamaindex/core/decorator";
+import { mapMessageContentToMessageContentDetails } from "../utils";
+import { AmazonProvider } from "./amazon/provider";
+import { AnthropicProvider } from "./anthropic/provider";
+import { MetaProvider } from "./meta/provider";
+
+// Other providers should go here
+export const PROVIDERS: { [key: string]: Provider } = {
+  anthropic: new AnthropicProvider(),
+  meta: new MetaProvider(),
+  amazon: new AmazonProvider(),
+};
+
+export type BedrockChatParamsStreaming = LLMChatParamsStreaming<
+  BedrockAdditionalChatOptions,
+  ToolCallLLMMessageOptions
+>;
+
+export type BedrockChatParamsNonStreaming = LLMChatParamsNonStreaming<
+  BedrockAdditionalChatOptions,
+  ToolCallLLMMessageOptions
+>;
+
+export type BedrockChatNonStreamResponse =
+  ChatResponse<ToolCallLLMMessageOptions>;
+
+export const BEDROCK_MODELS = {
+  AMAZON_TITAN_TG1_LARGE: "amazon.titan-tg1-large",
+  AMAZON_TITAN_TEXT_EXPRESS_V1: "amazon.titan-text-express-v1",
+  AI21_J2_GRANDE_INSTRUCT: "ai21.j2-grande-instruct",
+  AI21_J2_JUMBO_INSTRUCT: "ai21.j2-jumbo-instruct",
+  AI21_J2_MID: "ai21.j2-mid",
+  AI21_J2_MID_V1: "ai21.j2-mid-v1",
+  AI21_J2_ULTRA: "ai21.j2-ultra",
+  AI21_J2_ULTRA_V1: "ai21.j2-ultra-v1",
+  COHERE_COMMAND_TEXT_V14: "cohere.command-text-v14",
+  ANTHROPIC_CLAUDE_INSTANT_1: "anthropic.claude-instant-v1",
+  ANTHROPIC_CLAUDE_1: "anthropic.claude-v1", // EOF: No longer supported
+  ANTHROPIC_CLAUDE_2: "anthropic.claude-v2",
+  ANTHROPIC_CLAUDE_2_1: "anthropic.claude-v2:1",
+  ANTHROPIC_CLAUDE_3_SONNET: "anthropic.claude-3-sonnet-20240229-v1:0",
+  ANTHROPIC_CLAUDE_3_HAIKU: "anthropic.claude-3-haiku-20240307-v1:0",
+  ANTHROPIC_CLAUDE_3_OPUS: "anthropic.claude-3-opus-20240229-v1:0",
+  ANTHROPIC_CLAUDE_3_5_SONNET: "anthropic.claude-3-5-sonnet-20240620-v1:0",
+  ANTHROPIC_CLAUDE_3_5_SONNET_V2: "anthropic.claude-3-5-sonnet-20241022-v2:0",
+  ANTHROPIC_CLAUDE_3_5_HAIKU: "anthropic.claude-3-5-haiku-20241022-v1:0",
+  ANTHROPIC_CLAUDE_3_7_SONNET: "anthropic.claude-3-7-sonnet-20250219-v1:0",
+  META_LLAMA2_13B_CHAT: "meta.llama2-13b-chat-v1",
+  META_LLAMA2_70B_CHAT: "meta.llama2-70b-chat-v1",
+  META_LLAMA3_8B_INSTRUCT: "meta.llama3-8b-instruct-v1:0",
+  META_LLAMA3_70B_INSTRUCT: "meta.llama3-70b-instruct-v1:0",
+  META_LLAMA3_1_8B_INSTRUCT: "meta.llama3-1-8b-instruct-v1:0",
+  META_LLAMA3_1_70B_INSTRUCT: "meta.llama3-1-70b-instruct-v1:0",
+  META_LLAMA3_1_405B_INSTRUCT: "meta.llama3-1-405b-instruct-v1:0",
+  META_LLAMA3_2_1B_INSTRUCT: "meta.llama3-2-1b-instruct-v1:0",
+  META_LLAMA3_2_3B_INSTRUCT: "meta.llama3-2-3b-instruct-v1:0",
+  META_LLAMA3_2_11B_INSTRUCT: "meta.llama3-2-11b-instruct-v1:0",
+  META_LLAMA3_2_90B_INSTRUCT: "meta.llama3-2-90b-instruct-v1:0",
+  META_LLAMA3_3_70B_INSTRUCT: "meta.llama3-3-70b-instruct-v1:0",
+  MISTRAL_7B_INSTRUCT: "mistral.mistral-7b-instruct-v0:2",
+  MISTRAL_MIXTRAL_7B_INSTRUCT: "mistral.mixtral-8x7b-instruct-v0:1",
+  MISTRAL_MIXTRAL_LARGE_2402: "mistral.mistral-large-2402-v1:0",
+  AMAZON_NOVA_PREMIER_1: "amazon.nova-premier-v1:0",
+  AMAZON_NOVA_PRO_1: "amazon.nova-pro-v1:0",
+  AMAZON_NOVA_LITE_1: "amazon.nova-lite-v1:0",
+  AMAZON_NOVA_MICRO_1: "amazon.nova-micro-v1:0",
+};
+
+export type BEDROCK_MODELS =
+  (typeof BEDROCK_MODELS)[keyof typeof BEDROCK_MODELS];
+
+export const INFERENCE_BEDROCK_MODELS = {
+  US_ANTHROPIC_CLAUDE_3_HAIKU: "us.anthropic.claude-3-haiku-20240307-v1:0",
+  US_ANTHROPIC_CLAUDE_3_5_HAIKU: "us.anthropic.claude-3-5-haiku-20241022-v1:0",
+  US_ANTHROPIC_CLAUDE_3_OPUS: "us.anthropic.claude-3-opus-20240229-v1:0",
+  US_ANTHROPIC_CLAUDE_3_SONNET: "us.anthropic.claude-3-sonnet-20240229-v1:0",
+  US_ANTHROPIC_CLAUDE_3_5_SONNET:
+    "us.anthropic.claude-3-5-sonnet-20240620-v1:0",
+  US_ANTHROPIC_CLAUDE_3_5_SONNET_V2:
+    "us.anthropic.claude-3-5-sonnet-20241022-v2:0",
+  US_ANTHROPIC_CLAUDE_3_7_SONNET:
+    "us.anthropic.claude-3-7-sonnet-20250219-v1:0",
+  US_META_LLAMA_3_2_1B_INSTRUCT: "us.meta.llama3-2-1b-instruct-v1:0",
+  US_META_LLAMA_3_2_3B_INSTRUCT: "us.meta.llama3-2-3b-instruct-v1:0",
+  US_META_LLAMA_3_2_11B_INSTRUCT: "us.meta.llama3-2-11b-instruct-v1:0",
+  US_META_LLAMA_3_2_90B_INSTRUCT: "us.meta.llama3-2-90b-instruct-v1:0",
+  US_META_LLAMA_3_3_70B_INSTRUCT: "us.meta.llama3-3-70b-instruct-v1:0",
+  US_AMAZON_NOVA_PREMIER_1: "us.amazon.nova-premier-v1:0",
+  US_AMAZON_NOVA_PRO_1: "us.amazon.nova-pro-v1:0",
+  US_AMAZON_NOVA_LITE_1: "us.amazon.nova-lite-v1:0",
+  US_AMAZON_NOVA_MICRO_1: "us.amazon.nova-micro-v1:0",
+
+  EU_ANTHROPIC_CLAUDE_3_HAIKU: "eu.anthropic.claude-3-haiku-20240307-v1:0",
+  EU_ANTHROPIC_CLAUDE_3_5_HAIKU: "eu.anthropic.claude-3-5-haiku-20240307-v1:0",
+  EU_ANTHROPIC_CLAUDE_3_SONNET: "eu.anthropic.claude-3-sonnet-20240229-v1:0",
+  EU_ANTHROPIC_CLAUDE_3_5_SONNET:
+    "eu.anthropic.claude-3-5-sonnet-20240620-v1:0",
+  EU_ANTHROPIC_CLAUDE_3_7_SONNET:
+    "eu.anthropic.claude-3-7-sonnet-20250219-v1:0",
+  EU_META_LLAMA_3_2_1B_INSTRUCT: "eu.meta.llama3-2-1b-instruct-v1:0",
+  EU_META_LLAMA_3_2_3B_INSTRUCT: "eu.meta.llama3-2-3b-instruct-v1:0",
+  EU_AMAZON_NOVA_PREMIER_1: "eu.amazon.nova-premier-v1:0",
+  EU_AMAZON_NOVA_PRO_1: "eu.amazon.nova-pro-v1:0",
+  EU_AMAZON_NOVA_LITE_1: "eu.amazon.nova-lite-v1:0",
+  EU_AMAZON_NOVA_MICRO_1: "eu.amazon.nova-micro-v1:0",
+};
+
+export type INFERENCE_BEDROCK_MODELS =
+  (typeof INFERENCE_BEDROCK_MODELS)[keyof typeof INFERENCE_BEDROCK_MODELS];
+
+export const INFERENCE_TO_BEDROCK_MAP: Record<
+  INFERENCE_BEDROCK_MODELS,
+  BEDROCK_MODELS
+> = {
+  [INFERENCE_BEDROCK_MODELS.US_ANTHROPIC_CLAUDE_3_HAIKU]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_HAIKU,
+  [INFERENCE_BEDROCK_MODELS.US_ANTHROPIC_CLAUDE_3_OPUS]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_OPUS,
+  [INFERENCE_BEDROCK_MODELS.US_ANTHROPIC_CLAUDE_3_SONNET]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_SONNET,
+  [INFERENCE_BEDROCK_MODELS.US_ANTHROPIC_CLAUDE_3_5_SONNET]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET,
+  [INFERENCE_BEDROCK_MODELS.US_ANTHROPIC_CLAUDE_3_7_SONNET]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_7_SONNET,
+  [INFERENCE_BEDROCK_MODELS.US_ANTHROPIC_CLAUDE_3_5_SONNET_V2]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET_V2,
+  [INFERENCE_BEDROCK_MODELS.US_ANTHROPIC_CLAUDE_3_5_HAIKU]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_HAIKU,
+  [INFERENCE_BEDROCK_MODELS.US_META_LLAMA_3_2_1B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_2_1B_INSTRUCT,
+  [INFERENCE_BEDROCK_MODELS.US_META_LLAMA_3_2_3B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_2_3B_INSTRUCT,
+  [INFERENCE_BEDROCK_MODELS.US_META_LLAMA_3_2_11B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_2_11B_INSTRUCT,
+  [INFERENCE_BEDROCK_MODELS.US_META_LLAMA_3_2_90B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_2_90B_INSTRUCT,
+  [INFERENCE_BEDROCK_MODELS.US_META_LLAMA_3_3_70B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_3_70B_INSTRUCT,
+
+  [INFERENCE_BEDROCK_MODELS.US_AMAZON_NOVA_PREMIER_1]:
+    BEDROCK_MODELS.AMAZON_NOVA_PREMIER_1,
+  [INFERENCE_BEDROCK_MODELS.US_AMAZON_NOVA_PRO_1]:
+    BEDROCK_MODELS.AMAZON_NOVA_PRO_1,
+  [INFERENCE_BEDROCK_MODELS.US_AMAZON_NOVA_LITE_1]:
+    BEDROCK_MODELS.AMAZON_NOVA_LITE_1,
+  [INFERENCE_BEDROCK_MODELS.US_AMAZON_NOVA_MICRO_1]:
+    BEDROCK_MODELS.AMAZON_NOVA_MICRO_1,
+
+  [INFERENCE_BEDROCK_MODELS.EU_ANTHROPIC_CLAUDE_3_HAIKU]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_HAIKU,
+  [INFERENCE_BEDROCK_MODELS.EU_ANTHROPIC_CLAUDE_3_SONNET]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_SONNET,
+  [INFERENCE_BEDROCK_MODELS.EU_ANTHROPIC_CLAUDE_3_5_SONNET]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET,
+  [INFERENCE_BEDROCK_MODELS.EU_ANTHROPIC_CLAUDE_3_7_SONNET]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_7_SONNET,
+  [INFERENCE_BEDROCK_MODELS.EU_META_LLAMA_3_2_1B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_2_1B_INSTRUCT,
+  [INFERENCE_BEDROCK_MODELS.EU_META_LLAMA_3_2_3B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_2_3B_INSTRUCT,
+
+  [INFERENCE_BEDROCK_MODELS.EU_AMAZON_NOVA_PREMIER_1]:
+    BEDROCK_MODELS.AMAZON_NOVA_PREMIER_1,
+  [INFERENCE_BEDROCK_MODELS.EU_AMAZON_NOVA_PRO_1]:
+    BEDROCK_MODELS.AMAZON_NOVA_PRO_1,
+  [INFERENCE_BEDROCK_MODELS.EU_AMAZON_NOVA_LITE_1]:
+    BEDROCK_MODELS.AMAZON_NOVA_LITE_1,
+  [INFERENCE_BEDROCK_MODELS.EU_AMAZON_NOVA_MICRO_1]:
+    BEDROCK_MODELS.AMAZON_NOVA_MICRO_1,
+};
+
+/*
+ * Values taken from https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters.html#model-parameters-claude
+ */
+
+const COMPLETION_MODELS = {
+  [BEDROCK_MODELS.AMAZON_TITAN_TG1_LARGE]: 8000,
+  [BEDROCK_MODELS.AMAZON_TITAN_TEXT_EXPRESS_V1]: 8000,
+  [BEDROCK_MODELS.AI21_J2_GRANDE_INSTRUCT]: 8000,
+  [BEDROCK_MODELS.AI21_J2_JUMBO_INSTRUCT]: 8000,
+  [BEDROCK_MODELS.AI21_J2_MID]: 8000,
+  [BEDROCK_MODELS.AI21_J2_MID_V1]: 8000,
+  [BEDROCK_MODELS.AI21_J2_ULTRA]: 8000,
+  [BEDROCK_MODELS.AI21_J2_ULTRA_V1]: 8000,
+  [BEDROCK_MODELS.COHERE_COMMAND_TEXT_V14]: 4096,
+};
+
+const CHAT_ONLY_MODELS = {
+  [BEDROCK_MODELS.ANTHROPIC_CLAUDE_INSTANT_1]: 100000,
+  [BEDROCK_MODELS.ANTHROPIC_CLAUDE_1]: 100000,
+  [BEDROCK_MODELS.ANTHROPIC_CLAUDE_2]: 100000,
+  [BEDROCK_MODELS.ANTHROPIC_CLAUDE_2_1]: 200000,
+  [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_SONNET]: 200000,
+  [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_HAIKU]: 200000,
+  [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_OPUS]: 200000,
+  [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET]: 200000,
+  [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET_V2]: 200000,
+  [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_HAIKU]: 200000,
+  [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_7_SONNET]: 200000,
+  [BEDROCK_MODELS.META_LLAMA2_13B_CHAT]: 2048,
+  [BEDROCK_MODELS.META_LLAMA2_70B_CHAT]: 4096,
+  [BEDROCK_MODELS.META_LLAMA3_8B_INSTRUCT]: 8192,
+  [BEDROCK_MODELS.META_LLAMA3_70B_INSTRUCT]: 8192,
+  [BEDROCK_MODELS.META_LLAMA3_1_8B_INSTRUCT]: 128000,
+  [BEDROCK_MODELS.META_LLAMA3_1_70B_INSTRUCT]: 128000,
+  [BEDROCK_MODELS.META_LLAMA3_1_405B_INSTRUCT]: 128000,
+  [BEDROCK_MODELS.META_LLAMA3_2_1B_INSTRUCT]: 131000,
+  [BEDROCK_MODELS.META_LLAMA3_2_3B_INSTRUCT]: 131000,
+  [BEDROCK_MODELS.META_LLAMA3_2_11B_INSTRUCT]: 128000,
+  [BEDROCK_MODELS.META_LLAMA3_2_90B_INSTRUCT]: 128000,
+  [BEDROCK_MODELS.META_LLAMA3_3_70B_INSTRUCT]: 128000,
+  [BEDROCK_MODELS.MISTRAL_7B_INSTRUCT]: 32000,
+  [BEDROCK_MODELS.MISTRAL_MIXTRAL_7B_INSTRUCT]: 32000,
+  [BEDROCK_MODELS.MISTRAL_MIXTRAL_LARGE_2402]: 32000,
+  [BEDROCK_MODELS.AMAZON_NOVA_PREMIER_1]: 300000,
+  [BEDROCK_MODELS.AMAZON_NOVA_PRO_1]: 300000,
+  [BEDROCK_MODELS.AMAZON_NOVA_LITE_1]: 300000,
+  [BEDROCK_MODELS.AMAZON_NOVA_MICRO_1]: 130000,
+};
+
+const BEDROCK_FOUNDATION_LLMS = { ...COMPLETION_MODELS, ...CHAT_ONLY_MODELS };
+
+/*
+ * Only the following models support streaming as
+ * per result of Bedrock.Client.list_foundation_models
+ * https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/bedrock/client/list_foundation_models.html
+ */
+export const STREAMING_MODELS = new Set([
+  BEDROCK_MODELS.AMAZON_TITAN_TG1_LARGE,
+  BEDROCK_MODELS.AMAZON_TITAN_TEXT_EXPRESS_V1,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_INSTANT_1,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_1,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_2,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_2_1,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_SONNET,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_HAIKU,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_OPUS,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET_V2,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_HAIKU,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_7_SONNET,
+  BEDROCK_MODELS.META_LLAMA2_13B_CHAT,
+  BEDROCK_MODELS.META_LLAMA2_70B_CHAT,
+  BEDROCK_MODELS.META_LLAMA3_8B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_70B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_1_8B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_1_70B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_1_405B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_1B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_3B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_11B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_90B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_3_70B_INSTRUCT,
+  BEDROCK_MODELS.MISTRAL_7B_INSTRUCT,
+  BEDROCK_MODELS.MISTRAL_MIXTRAL_7B_INSTRUCT,
+  BEDROCK_MODELS.MISTRAL_MIXTRAL_LARGE_2402,
+  BEDROCK_MODELS.AMAZON_NOVA_PREMIER_1,
+  BEDROCK_MODELS.AMAZON_NOVA_PRO_1,
+  BEDROCK_MODELS.AMAZON_NOVA_LITE_1,
+  BEDROCK_MODELS.AMAZON_NOVA_MICRO_1,
+]);
+
+export const TOOL_CALL_MODELS: BEDROCK_MODELS[] = [
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_SONNET,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_HAIKU,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_OPUS,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET_V2,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_HAIKU,
+  BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_7_SONNET,
+  BEDROCK_MODELS.META_LLAMA3_1_405B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_1B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_3B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_11B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_90B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_3_70B_INSTRUCT,
+  BEDROCK_MODELS.AMAZON_NOVA_PREMIER_1,
+  BEDROCK_MODELS.AMAZON_NOVA_PRO_1,
+  BEDROCK_MODELS.AMAZON_NOVA_LITE_1,
+  BEDROCK_MODELS.AMAZON_NOVA_MICRO_1,
+];
+
+const getProvider = (model: string): Provider => {
+  const providerName = model.split(".")[0];
+  if (!providerName) {
+    throw new Error(`Model ${model} is not supported`);
+  }
+  if (!(providerName in PROVIDERS)) {
+    throw new Error(
+      `Provider ${providerName} for model ${model} is not supported`,
+    );
+  }
+  return PROVIDERS[providerName]!;
+};
+
+export type BedrockModelParams = {
+  model: BEDROCK_MODELS | INFERENCE_BEDROCK_MODELS;
+  temperature?: number;
+  topP?: number;
+  maxTokens?: number;
+};
+
+export const BEDROCK_MODEL_MAX_TOKENS: Partial<Record<BEDROCK_MODELS, number>> =
+  {
+    [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_SONNET]: 4096,
+    [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_HAIKU]: 4096,
+    [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_OPUS]: 4096,
+    [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET]: 4096,
+    [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET_V2]: 8192,
+    [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_HAIKU]: 8192,
+    [BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_7_SONNET]: 8192,
+    [BEDROCK_MODELS.META_LLAMA2_13B_CHAT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA2_70B_CHAT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_8B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_70B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_1_8B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_1_70B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_1_405B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_2_1B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_2_3B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_2_11B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_2_90B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_3_70B_INSTRUCT]: 2048,
+  };
+
+const DEFAULT_BEDROCK_PARAMS = {
+  temperature: 0.1,
+  topP: 1,
+  maxTokens: 1024, // required by anthropic
+};
+
+export type BedrockParams = BedrockRuntimeClientConfig & BedrockModelParams;
+
+/**
+ * ToolCallLLM for Bedrock
+ */
+export class Bedrock extends ToolCallLLM<BedrockAdditionalChatOptions> {
+  private client: BedrockRuntimeClient;
+  protected actualModel: BEDROCK_MODELS | INFERENCE_BEDROCK_MODELS;
+  model: BEDROCK_MODELS;
+  temperature: number;
+  topP: number;
+  maxTokens?: number;
+  provider: Provider;
+  topK?: number;
+
+  // there should be no check for env variables. Bedrock can be authenticated in various ways
+  // AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY and AWS_REGION are the env variables used directly by the sdk
+  constructor({
+    temperature,
+    topP,
+    maxTokens,
+    model,
+    ...params
+  }: BedrockParams) {
+    super();
+    this.actualModel = model;
+    this.model = INFERENCE_TO_BEDROCK_MAP[model] ?? model;
+    this.provider = getProvider(this.model);
+    this.maxTokens = maxTokens ?? DEFAULT_BEDROCK_PARAMS.maxTokens;
+    this.temperature = temperature ?? DEFAULT_BEDROCK_PARAMS.temperature;
+    this.topP = topP ?? DEFAULT_BEDROCK_PARAMS.topP;
+    this.client = new BedrockRuntimeClient(params);
+  }
+
+  get supportToolCall(): boolean {
+    return TOOL_CALL_MODELS.includes(this.model);
+  }
+
+  get metadata(): LLMMetadata {
+    // NOTE, Anthropic supports top_k but LLMMetadata does not
+    return {
+      model: this.model,
+      temperature: this.temperature,
+      topP: this.topP,
+      maxTokens: this.maxTokens,
+      contextWindow: BEDROCK_FOUNDATION_LLMS[this.model] ?? 128000,
+      tokenizer: undefined,
+      structuredOutput: false,
+    };
+  }
+
+  protected async nonStreamChat(
+    params: BedrockChatParamsNonStreaming,
+  ): Promise<BedrockChatNonStreamResponse> {
+    if (!this.supportToolCall && params.tools?.length) {
+      console.warn(`The model "${this.model}" doesn't support ToolCall`);
+    }
+    const input = this.provider.getRequestBody(
+      this.metadata,
+      params.messages,
+      params.tools,
+      params.additionalChatOptions,
+    );
+    const command = new InvokeModelCommand(input);
+    command.input.modelId = this.actualModel;
+
+    const response = await this.client.send(command);
+    let options: ToolCallLLMMessageOptions = {};
+    if (this.supportToolCall) {
+      const tools = this.provider.getToolsFromResponse(response);
+      if (tools.length) {
+        options = { toolCall: tools };
+      }
+    }
+    return {
+      raw: response,
+      message: {
+        role: "assistant",
+        content: this.provider.getTextFromResponse(response),
+        options,
+      },
+    };
+  }
+
+  protected async *streamChat(
+    params: BedrockChatParamsStreaming,
+  ): BedrockChatStreamResponse {
+    if (!STREAMING_MODELS.has(this.model))
+      throw new Error(`The model: ${this.model} does not support streaming`);
+
+    if (!this.supportToolCall && params.tools?.length) {
+      console.warn(`The model "${this.model}" doesn't support ToolCall`);
+    }
+
+    const input = this.provider.getRequestBody(
+      this.metadata,
+      params.messages,
+      params.tools,
+      params.additionalChatOptions,
+    );
+    const command = new InvokeModelWithResponseStreamCommand(input);
+    command.input.modelId = this.actualModel;
+
+    const response = await this.client.send(command);
+    if (response.body) yield* this.provider.reduceStream(response.body);
+  }
+
+  chat(params: BedrockChatParamsStreaming): Promise<BedrockChatStreamResponse>;
+  chat(
+    params: BedrockChatParamsNonStreaming,
+  ): Promise<BedrockChatNonStreamResponse>;
+  @wrapLLMEvent
+  async chat(
+    params: BedrockChatParamsStreaming | BedrockChatParamsNonStreaming,
+  ): Promise<BedrockChatStreamResponse | BedrockChatNonStreamResponse> {
+    if (params.stream) {
+      return this.streamChat(params);
+    }
+    return this.nonStreamChat(params);
+  }
+
+  complete(
+    params: LLMCompletionParamsStreaming,
+  ): Promise<AsyncIterable<CompletionResponse>>;
+  complete(
+    params: LLMCompletionParamsNonStreaming,
+  ): Promise<CompletionResponse>;
+  async complete(
+    params: LLMCompletionParamsStreaming | LLMCompletionParamsNonStreaming,
+  ): Promise<CompletionResponse | AsyncIterable<CompletionResponse>> {
+    const message: ChatMessage = {
+      role: "user",
+      content: mapMessageContentToMessageContentDetails(params.prompt),
+    };
+
+    const input = this.provider.getRequestBody(this.metadata, [message]);
+
+    if (params.stream) {
+      const command = new InvokeModelWithResponseStreamCommand(input);
+      const response = await this.client.send(command);
+      if (response.body)
+        return streamConverter(response.body, (response) => {
+          return {
+            text: this.provider.getTextFromStreamResponse(response),
+            raw: response,
+          };
+        });
+    }
+
+    const command = new InvokeModelCommand(input);
+    const response = await this.client.send(command);
+    return {
+      text: this.provider.getTextFromResponse(response),
+      raw: response,
+    };
+  }
+}
@@ -0,0 +1,3 @@
+export const TOKENS = {
+  TOOL_CALL: "<|python_tag|>",
+};
@@ -0,0 +1,153 @@
+import type {
+  InvokeModelCommandInput,
+  InvokeModelWithResponseStreamCommandInput,
+  ResponseStream,
+} from "@aws-sdk/client-bedrock-runtime";
+import type {
+  BaseTool,
+  ChatMessage,
+  LLMMetadata,
+  ToolCall,
+  ToolCallLLMMessageOptions,
+} from "@llamaindex/core/llms";
+import { toUtf8 } from "../utils";
+import type { MetaNoneStreamingResponse, MetaStreamEvent } from "./types";
+
+import { randomUUID } from "@llamaindex/env";
+import { Provider, type BedrockChatStreamResponse } from "../provider";
+import { TOKENS } from "./constants";
+import {
+  mapChatMessagesToMetaLlama2Messages,
+  mapChatMessagesToMetaLlama3Messages,
+} from "./utils";
+
+export class MetaProvider extends Provider<MetaStreamEvent> {
+  getResultFromResponse(
+    // eslint-disable-next-line @typescript-eslint/no-explicit-any
+    response: Record<string, any>,
+  ): MetaNoneStreamingResponse {
+    return JSON.parse(toUtf8(response.body));
+  }
+
+  getToolsFromResponse<ToolContent>(
+    // eslint-disable-next-line @typescript-eslint/no-explicit-any
+    response: Record<string, any>,
+  ): ToolContent[] {
+    const result = this.getResultFromResponse(response);
+    if (!result.generation.trim().startsWith(TOKENS.TOOL_CALL)) return [];
+    const tool = JSON.parse(
+      result.generation.trim().split(TOKENS.TOOL_CALL)[1]!,
+    );
+    return [
+      {
+        id: randomUUID(),
+        name: tool.name,
+        input: tool.parameters,
+      } as ToolContent,
+    ];
+  }
+
+  // eslint-disable-next-line @typescript-eslint/no-explicit-any
+  getTextFromResponse(response: Record<string, any>): string {
+    const result = this.getResultFromResponse(response);
+    if (result.generation.trim().startsWith(TOKENS.TOOL_CALL)) return "";
+    return result.generation;
+  }
+
+  // eslint-disable-next-line @typescript-eslint/no-explicit-any
+  getTextFromStreamResponse(response: Record<string, any>): string {
+    const event = this.getStreamingEventResponse(response);
+    if (event?.generation) {
+      return event.generation;
+    }
+    return "";
+  }
+
+  async *reduceStream(
+    stream: AsyncIterable<ResponseStream>,
+  ): BedrockChatStreamResponse {
+    const collecting: string[] = [];
+    let toolId: string | undefined = undefined;
+    for await (const response of stream) {
+      const event = this.getStreamingEventResponse(response);
+      const delta = this.getTextFromStreamResponse(response);
+
+      // odd quirk of llama3.1, start token is \n\n
+      if (
+        !toolId &&
+        !event?.generation.trim() &&
+        event?.generation_token_count === 1 &&
+        event?.prompt_token_count !== null
+      )
+        continue;
+
+      if (delta.startsWith(TOKENS.TOOL_CALL)) {
+        toolId = randomUUID();
+        const parts = delta.split(TOKENS.TOOL_CALL).filter((part) => part);
+        collecting.push(...parts);
+        continue;
+      }
+
+      let options: undefined | ToolCallLLMMessageOptions = undefined;
+      if (toolId && event?.stop_reason === "stop") {
+        if (delta) collecting.push(delta);
+        const tool = JSON.parse(collecting.join(""));
+        options = {
+          toolCall: [
+            {
+              id: toolId,
+              name: tool.name,
+              input: tool.parameters,
+            } as ToolCall,
+          ],
+        };
+      } else if (toolId && !event?.stop_reason) {
+        collecting.push(delta);
+        continue;
+      }
+
+      if (!delta && !options) continue;
+
+      yield {
+        delta: options ? "" : delta,
+        options,
+        raw: response,
+      };
+    }
+  }
+
+  getRequestBody<T extends ChatMessage>(
+    metadata: LLMMetadata,
+    messages: T[],
+    tools: BaseTool[] = [],
+  ): InvokeModelCommandInput | InvokeModelWithResponseStreamCommandInput {
+    let prompt: string = "";
+    let images: string[] = [];
+    if (metadata.model.startsWith("meta.llama3")) {
+      const mapped = mapChatMessagesToMetaLlama3Messages({
+        messages,
+        tools,
+        model: metadata.model,
+      });
+      prompt = mapped.prompt;
+      images = mapped.images;
+    } else if (metadata.model.startsWith("meta.llama2")) {
+      prompt = mapChatMessagesToMetaLlama2Messages(messages);
+    } else {
+      throw new Error(`Meta model ${metadata.model} is not supported`);
+    }
+
+    return {
+      modelId: metadata.model,
+      contentType: "application/json",
+      accept: "application/json",
+      body: JSON.stringify({
+        prompt,
+        images: images.length ? images : undefined,
+        max_gen_len: metadata.maxTokens,
+        temperature: metadata.temperature,
+        top_p: metadata.topP,
+      }),
+    };
+  }
+}
@@ -0,0 +1,21 @@
+import type { InvocationMetrics } from "../types";
+
+export type MetaTextContent = string;
+
+export type MetaMessage = {
+  role: "user" | "assistant" | "system" | "ipython";
+  content: MetaTextContent;
+};
+
+type MetaResponse = {
+  generation: string;
+  prompt_token_count: number;
+  generation_token_count: number;
+  stop_reason: "stop" | "length";
+};
+
+export type MetaStreamEvent = MetaResponse & {
+  "amazon-bedrock-invocationMetrics": InvocationMetrics;
+};
+
+export type MetaNoneStreamingResponse = MetaResponse;
@@ -0,0 +1,273 @@
+import type {
+  BaseTool,
+  ChatMessage,
+  LLMMetadata,
+  MessageContentTextDetail,
+  ToolCallLLMMessageOptions,
+} from "@llamaindex/core/llms";
+import { extractDataUrlComponents } from "../utils";
+import { TOKENS } from "./constants";
+import type { MetaMessage } from "./types";
+
+const getToolCallInstructionString = (tool: BaseTool): string => {
+  return `Use the function '${tool.metadata.name}' to '${tool.metadata.description}'`;
+};
+
+const getToolCallParametersString = (tool: BaseTool): string => {
+  return JSON.stringify({
+    name: tool.metadata.name,
+    description: tool.metadata.description,
+    parameters: tool.metadata.parameters
+      ? Object.entries(tool.metadata.parameters.properties).map(
+          ([name, definition]) => ({ [name]: definition }),
+        )
+      : {},
+  });
+};
+
+// ported from https://github.com/meta-llama/llama-agentic-system/blob/main/llama_agentic_system/system_prompt.py
+// NOTE: using json instead of the above xml style tool calling works more reliability
+export const getToolsPrompt_3_1 = (tools?: BaseTool[]) => {
+  if (!tools?.length) return "";
+
+  const customToolParams = tools.map((tool) => {
+    return [
+      getToolCallInstructionString(tool),
+      getToolCallParametersString(tool),
+    ].join("\n\n");
+  });
+
+  return `
+Environment: node
+
+# Tool Instructions
+- Never use ipython, always use javascript in node
+
+Cutting Knowledge Date: December 2023
+Today Date: ${new Date().toLocaleString("en-US", { year: "numeric", month: "long" })} 
+
+You have access to the following functions:
+
+${customToolParams}
+
+Think very carefully before calling functions.
+
+If a you choose to call a function ONLY reply in the following json format:
+{
+  "name": function_name,
+  "parameters": parameters,
+}
+where
+
+{
+  "name": function_name,
+  "parameters": parameters, => a JSON dict with the function argument name as key and function argument value as value.
+}
+
+Here is an example,
+
+{
+  "name": "example_function_name",
+  "parameters": {"example_name": "example_value"}
+}
+
+Reminder:
+- Function calls MUST follow the specified format
+- Required parameters MUST be specified
+- Only call one function at a time
+- Put the entire function call reply on one line
+- Always add your sources when using search results to answer the user query
+  `;
+};
+
+export const getToolsPrompt_3_2 = (tools?: BaseTool[]) => {
+  if (!tools?.length) return "";
+  return `
+You are an expert in composing functions. You are given a question and a set of possible functions.
+Based on the question, you will need to make one or more function/tool calls to achieve the purpose.
+If none of the function can be used, point it out. If the given question lacks the parameters required by the function,
+also point it out. You should only return the function call in tools call sections.
+
+If you decide to invoke any of the function(s), you MUST put it in the format of and start with the token: ${TOKENS.TOOL_CALL}:
+{
+  "name": function_name,
+  "parameters": parameters,
+}
+where
+
+{
+  "name": function_name,
+  "parameters": parameters, => a JSON dict with the function argument name as key and function argument value as value.
+}
+
+Here is an example,
+
+{
+  "name": "example_function_name",
+  "parameters": {"example_name": "example_value"}
+}
+
+Reminder:
+- Function calls MUST follow the specified format
+- Required parameters MUST be specified
+- Only call one function at a time
+- You SHOULD NOT include any other text in the response
+- Put the entire function call reply on one line
+
+Here is a list of functions in JSON format that you can invoke.
+
+${JSON.stringify(tools)}
+`;
+};
+
+export const mapChatRoleToMetaRole = (
+  role: ChatMessage["role"],
+): MetaMessage["role"] => {
+  if (role === "assistant") return "assistant";
+  if (role === "user") return "user";
+  return "system";
+};
+
+export const mapChatMessagesToMetaMessages = <
+  T extends ChatMessage<ToolCallLLMMessageOptions>,
+>(
+  messages: T[],
+): MetaMessage[] => {
+  return messages.flatMap((msg) => {
+    if (msg.options && "toolCall" in msg.options) {
+      return msg.options.toolCall.map((call) => ({
+        role: "assistant",
+        content: JSON.stringify({
+          id: call.id,
+          name: call.name,
+          parameters: call.input,
+        }),
+      }));
+    }
+
+    if (msg.options && "toolResult" in msg.options) {
+      return {
+        role: "ipython",
+        content: JSON.stringify(msg.options.toolResult),
+      };
+    }
+
+    let content: string = "";
+    if (typeof msg.content === "string") {
+      content = msg.content;
+    } else if (msg.content.length) {
+      content = (msg.content[0] as MessageContentTextDetail).text;
+    }
+    return {
+      role: mapChatRoleToMetaRole(msg.role),
+      content,
+    };
+  });
+};
+
+/**
+ * Documentation at https://llama.meta.com/docs/model-cards-and-prompt-formats/meta-llama-3
+ */
+export const mapChatMessagesToMetaLlama3Messages = <T extends ChatMessage>({
+  messages,
+  model,
+  tools,
+}: {
+  messages: T[];
+  model: LLMMetadata["model"];
+  tools?: BaseTool[];
+}): { prompt: string; images: string[] } => {
+  const images: string[] = [];
+  const textMessages: T[] = [];
+
+  messages.forEach((message) => {
+    if (Array.isArray(message.content)) {
+      message.content.forEach((content) => {
+        if (content.type === "image_url") {
+          const { base64 } = extractDataUrlComponents(content.image_url.url);
+          images.push(base64);
+        } else {
+          textMessages.push(message);
+        }
+      });
+    } else {
+      textMessages.push(message);
+    }
+  });
+
+  const parts: string[] = [];
+
+  let toolsPrompt = "";
+  if (model.startsWith("meta.llama3-2")) {
+    toolsPrompt = getToolsPrompt_3_2(tools);
+  } else if (model.startsWith("meta.llama3-1")) {
+    toolsPrompt = getToolsPrompt_3_1(tools);
+  }
+  if (toolsPrompt) {
+    parts.push(
+      "<|begin_of_text|>",
+      "<|start_header_id|>system<|end_header_id|>",
+      toolsPrompt,
+      "<|eot_id|>",
+    );
+  }
+
+  const mapped = mapChatMessagesToMetaMessages(messages).map((message) => {
+    return [
+      "<|start_header_id|>",
+      message.role,
+      "<|end_header_id|>",
+      message.content,
+      "<|eot_id|>",
+    ].join("\n");
+  });
+
+  parts.push(
+    "<|begin_of_text|>",
+    ...mapped,
+    "<|start_header_id|>assistant<|end_header_id|>",
+  );
+
+  const prompt = parts.join("\n");
+  return { prompt, images };
+};
+
+/**
+ * Documentation at https://llama.meta.com/docs/model-cards-and-prompt-formats/meta-llama-2
+ */
+export const mapChatMessagesToMetaLlama2Messages = <T extends ChatMessage>(
+  messages: T[],
+): string => {
+  const mapped = mapChatMessagesToMetaMessages(messages);
+  let output = "<s>";
+  let insideInst = false;
+  let needsStartAgain = false;
+  for (const message of mapped) {
+    if (needsStartAgain) {
+      output += "<s>";
+      needsStartAgain = false;
+    }
+    const text = message.content;
+    if (message.role === "system") {
+      if (!insideInst) {
+        output += "[INST] ";
+        insideInst = true;
+      }
+      output += `<<SYS>>\n${text}\n<</SYS>>\n`;
+    } else if (message.role === "user") {
+      output += text;
+      if (insideInst) {
+        output += " [/INST]";
+        insideInst = false;
+      }
+    } else if (message.role === "assistant") {
+      if (insideInst) {
+        output += " [/INST]";
+        insideInst = false;
+      }
+      output += ` ${text} </s>\n`;
+      needsStartAgain = true;
+    }
+  }
+  return output;
+};
@@ -0,0 +1,63 @@
+import {
+  type InvokeModelCommandInput,
+  type InvokeModelWithResponseStreamCommandInput,
+  ResponseStream,
+} from "@aws-sdk/client-bedrock-runtime";
+import {
+  type BaseTool,
+  type ChatMessage,
+  type ChatResponseChunk,
+  type LLMMetadata,
+  type ToolCallLLMMessageOptions,
+} from "@llamaindex/core/llms";
+import { streamConverter } from "@llamaindex/core/utils";
+import { toUtf8 } from "./utils";
+
+export type BedrockAdditionalChatOptions = Record<string, unknown>;
+
+export type BedrockChatStreamResponse = AsyncIterable<
+  ChatResponseChunk<ToolCallLLMMessageOptions>
+>;
+
+export abstract class Provider<ProviderStreamEvent extends object = object> {
+  // eslint-disable-next-line @typescript-eslint/no-explicit-any
+  abstract getTextFromResponse(response: Record<string, any>): string;
+
+  // Return tool calls from none streaming calls
+  abstract getToolsFromResponse<T extends object = object>(
+    // eslint-disable-next-line @typescript-eslint/no-explicit-any
+    response: Record<string, any>,
+  ): T[];
+
+  getStreamingEventResponse(
+    // eslint-disable-next-line @typescript-eslint/no-explicit-any
+    response: Record<string, any>,
+  ): ProviderStreamEvent | undefined {
+    return response.chunk?.bytes
+      ? (JSON.parse(toUtf8(response.chunk?.bytes)) as ProviderStreamEvent)
+      : undefined;
+  }
+
+  async *reduceStream(
+    stream: AsyncIterable<ResponseStream>,
+  ): BedrockChatStreamResponse {
+    yield* streamConverter(stream, (response) => {
+      return {
+        delta: this.getTextFromStreamResponse(response),
+        raw: response,
+      };
+    });
+  }
+
+  // eslint-disable-next-line @typescript-eslint/no-explicit-any
+  getTextFromStreamResponse(response: Record<string, any>): string {
+    return this.getTextFromResponse(response);
+  }
+
+  abstract getRequestBody<T extends ChatMessage>(
+    metadata: LLMMetadata,
+    messages: T[],
+    tools?: BaseTool[],
+    options?: BedrockAdditionalChatOptions,
+  ): InvokeModelCommandInput | InvokeModelWithResponseStreamCommandInput;
+}
@@ -0,0 +1,6 @@
+export type InvocationMetrics = {
+  inputTokenCount: number;
+  outputTokenCount: number;
+  invocationLatency: number;
+  firstByteLatency: number;
+};
@@ -0,0 +1,23 @@
+export const toUtf8 = (input: Uint8Array): string =>
+  new TextDecoder("utf-8").decode(input);
+
+export const extractDataUrlComponents = (
+  dataUrl: string,
+): {
+  mimeType: string;
+  base64: string;
+} => {
+  const parts = dataUrl.split(";base64,");
+
+  if (parts.length !== 2 || !parts[0]!.startsWith("data:")) {
+    throw new Error("Invalid data URL");
+  }
+
+  const mimeType = parts[0]!.slice(5);
+  const base64 = parts[1]!;
+
+  return {
+    mimeType,
+    base64,
+  };
+};
@@ -0,0 +1,10 @@
+import type {
+  MessageContent,
+  MessageContentDetail,
+} from "@llamaindex/core/llms";
+
+export const mapMessageContentToMessageContentDetails = (
+  content: MessageContent,
+): MessageContentDetail[] => {
+  return Array.isArray(content) ? content : [{ type: "text", text: content }];
+};
@@ -0,0 +1,165 @@
+import type { KnowledgeBaseVectorSearchConfiguration } from "@aws-sdk/client-bedrock-agent-runtime";
+import {
+  BedrockAgentRuntimeClient,
+  type BedrockAgentRuntimeClientConfig,
+  type RetrievalFilter,
+  RetrieveCommand,
+  type SearchType,
+} from "@aws-sdk/client-bedrock-agent-runtime";
+import type { QueryBundle } from "@llamaindex/core/query-engine";
+import { BaseRetriever } from "@llamaindex/core/retriever";
+import { Document, type NodeWithScore } from "@llamaindex/core/schema";
+import { extractText } from "@llamaindex/core/utils";
+
+/**
+ * Interface for the arguments required to initialize an
+ * AmazonKnowledgeBaseRetriever instance.
+ */
+export interface AmazonKnowledgeBaseRetrieverArgs {
+  knowledgeBaseId: string;
+  topK: number;
+  region: string;
+  clientOptions?: BedrockAgentRuntimeClientConfig;
+  filter?: RetrievalFilter;
+  overrideSearchType?: SearchType;
+}
+
+/**
+ * Class for interacting with Amazon Bedrock Knowledge Bases, a RAG workflow oriented service
+ * Extends the BaseRetriever class.
+ * @example
+ * ```typescript
+ * const retriever = new AmazonKnowledgeBaseRetriever({
+ *   topK: 10,
+ *   knowledgeBaseId: "YOUR_KNOWLEDGE_BASE_ID",
+ *   region: "us-east-2",
+ *   clientOptions: {
+ *     credentials: {
+ *       accessKeyId: "YOUR_ACCESS_KEY_ID",
+ *       secretAccessKey: "YOUR_SECRET_ACCESS_KEY",
+ *     },
+ *   },
+ * });
+ *
+ * const docs = await retriever.retrieve({query: "How are clouds formed?"});
+ * ```
+ */
+export class AmazonKnowledgeBaseRetriever extends BaseRetriever {
+  static lc_name() {
+    return "AmazonKnowledgeBaseRetriever";
+  }
+
+  lc_namespace = ["llamaindex", "retrievers", "amazon_bedrock_knowledge_base"];
+
+  knowledgeBaseId: string;
+
+  topK: number;
+
+  bedrockAgentRuntimeClient: BedrockAgentRuntimeClient;
+
+  filter: RetrievalFilter | undefined;
+
+  overrideSearchType: SearchType | undefined;
+
+  constructor({
+    knowledgeBaseId,
+    topK = 10,
+    clientOptions,
+    region,
+    filter,
+    overrideSearchType,
+  }: AmazonKnowledgeBaseRetrieverArgs) {
+    super();
+
+    this.topK = topK;
+    this.filter = filter;
+    this.overrideSearchType = overrideSearchType;
+    this.bedrockAgentRuntimeClient = new BedrockAgentRuntimeClient({
+      region,
+      ...clientOptions,
+    });
+    this.knowledgeBaseId = knowledgeBaseId;
+  }
+
+  /**
+   * Cleans the result text by replacing sequences of whitespace with a
+   * single space and removing ellipses.
+   * @param resText The result text to clean.
+   * @returns The cleaned result text.
+   */
+  cleanResult(resText: string) {
+    const res = resText.replace(/\s+/g, " ").replace(/\.\.\./g, "");
+    return res;
+  }
+
+  async queryKnowledgeBase(
+    query: QueryBundle,
+    topK: number,
+    filter?: RetrievalFilter,
+    overrideSearchType?: SearchType,
+  ): Promise<NodeWithScore[]> {
+    const retrieveCommand = new RetrieveCommand({
+      knowledgeBaseId: this.knowledgeBaseId,
+      retrievalQuery: {
+        text: extractText(query),
+      },
+      retrievalConfiguration: {
+        vectorSearchConfiguration: {
+          numberOfResults: topK,
+          overrideSearchType,
+          filter,
+        } as KnowledgeBaseVectorSearchConfiguration,
+      },
+    });
+
+    const retrieveResponse =
+      await this.bedrockAgentRuntimeClient.send(retrieveCommand);
+
+    return (
+      retrieveResponse.retrievalResults?.map((result) => {
+        let source;
+        switch (result.location?.type) {
+          case "CONFLUENCE":
+            source = result.location?.confluenceLocation?.url;
+            break;
+          case "S3":
+            source = result.location?.s3Location?.uri;
+            break;
+          case "SALESFORCE":
+            source = result.location?.salesforceLocation?.url;
+            break;
+          case "SHAREPOINT":
+            source = result.location?.sharePointLocation?.url;
+            break;
+          case "WEB":
+            source = result.location?.webLocation?.url;
+            break;
+          default:
+            source = result.location?.s3Location?.uri;
+            break;
+        }
+
+        return {
+          node: new Document({
+            text: this.cleanResult(result.content?.text || ""),
+            metadata: {
+              source,
+              score: result.score,
+              ...result.metadata,
+            },
+          }),
+          score: result.score ?? 1.0,
+        };
+      }) ?? []
+    );
+  }
+
+  async _retrieve(query: QueryBundle): Promise<NodeWithScore[]> {
+    return await this.queryKnowledgeBase(
+      query,
+      this.topK,
+      this.filter,
+      this.overrideSearchType,
+    );
+  }
+}
@@ -0,0 +1,19 @@
+{
+  "extends": "../../tsconfig.json",
+  "compilerOptions": {
+    "rootDir": "./src",
+    "outDir": "./dist/type",
+    "tsBuildInfoFile": "./dist/.tsbuildinfo",
+    "emitDeclarationOnly": true,
+    "module": "ESNext",
+    "moduleResolution": "bundler",
+    "types": ["node"]
+  },
+  "include": ["./src"],
+  "exclude": ["node_modules"],
+  "references": [
+    {
+      "path": "../llamaindex/tsconfig.json"
+    }
+  ]
+}
@@ -1,5 +1,31 @@
 # @llamaindex/core

+## 0.6.18
+
+### Patch Changes
+
+- f29799e: Add toolcall callbacks to agent workflows
+- 7224c06: Add logger and callbacks to llm.exec
+
+## 0.6.17
+
+### Patch Changes
+
+- 38da40b: feat: VectoryMemoryBlock
+
+## 0.6.16
+
+### Patch Changes
+
+- a8ec08c: fix: ensure correct message content in agent workflow
+
+## 0.6.15
+
+### Patch Changes
+
+- 7ad3411: feat: add llm.exec
+- 5da5b3c: add progress callback to embeddings
+
 ## 0.6.14

 ### Patch Changes
@@ -1,7 +1,7 @@
 {
  "name": "@llamaindex/core",
  "type": "module",
-  "version": "0.6.14",
+  "version": "0.6.18",
  "description": "LlamaIndex Core Module",
  "exports": {
    "./agent": {
@@ -59,6 +59,17 @@
      },
      "default": "./llms/dist/index.js"
    },
+    "./llms/mock": {
+      "require": {
+        "types": "./llms/mock/dist/index.d.cts",
+        "default": "./llms/mock/dist/index.cjs"
+      },
+      "import": {
+        "types": "./llms/mock/dist/index.d.ts",
+        "default": "./llms/mock/dist/index.js"
+      },
+      "default": "./llms/mock/dist/index.js"
+    },
    "./decorator": {
      "require": {
        "types": "./decorator/dist/index.d.cts",
@@ -15,6 +15,7 @@ import type {
 } from "../llms";
 import { baseToolWithCallSchema } from "../schema";
 import {
+  assertIsJSONValue,
  isAsyncIterable,
  prettifyError,
  stringifyJSONToMessageContent,
@@ -227,6 +228,7 @@ export async function callTool(
      `Tool ${tool.metadata.name} (remote:${toolCall.name}) succeeded.`,
    );
    logger.log(`Output: ${JSON.stringify(output)}`);
+    assertIsJSONValue(output);
    const toolOutput: ToolOutput = {
      tool,
      input,
@@ -17,6 +17,7 @@ export type EmbeddingInfo = {

 export type BaseEmbeddingOptions = {
  logProgress?: boolean;
+  progressCallback?: (current: number, total: number) => void;
 };

 export abstract class BaseEmbedding extends TransformComponent<
@@ -138,9 +139,11 @@ export async function batchEmbeddings<T>(
      const embeddings = await embedFunc(curBatch);

      resultEmbeddings.push(...embeddings);
-
+      if (options?.progressCallback) {
+        options?.progressCallback?.(i + 1, queue.length);
+      }
      if (options?.logProgress) {
-        console.log(`getting embedding progress: ${i} / ${queue.length}`);
+        console.log(`getting embedding progress: ${i + 1} / ${queue.length}`);
      }

      curBatch.length = 0;
@@ -1,15 +1,21 @@
+import { emptyLogger } from "@llamaindex/env";
 import { extractText } from "../utils/llms";
 import { streamConverter } from "../utils/stream";
+import { callToolToMessage, getToolCallsFromResponse } from "./tool-call";
 import type {
+  ChatMessage,
  ChatResponse,
  ChatResponseChunk,
  CompletionResponse,
+  ExecResponse,
+  ExecStreamResponse,
  LLM,
  LLMChatParamsNonStreaming,
  LLMChatParamsStreaming,
  LLMCompletionParamsNonStreaming,
  LLMCompletionParamsStreaming,
  LLMMetadata,
+  PartialToolCall,
  ToolCallLLMMessageOptions,
 } from "./type";

@@ -60,13 +66,186 @@ export abstract class BaseLLM<
      AdditionalChatOptions,
      AdditionalMessageOptions
    >,
-  ): Promise<AsyncIterable<ChatResponseChunk>>;
+  ): Promise<AsyncIterable<ChatResponseChunk<AdditionalMessageOptions>>>;
  abstract chat(
    params: LLMChatParamsNonStreaming<
      AdditionalChatOptions,
      AdditionalMessageOptions
    >,
  ): Promise<ChatResponse<AdditionalMessageOptions>>;
+
+  exec(
+    params: LLMChatParamsStreaming<
+      AdditionalChatOptions,
+      AdditionalMessageOptions
+    >,
+  ): Promise<ExecStreamResponse<AdditionalMessageOptions>>;
+  exec(
+    params: LLMChatParamsNonStreaming<
+      AdditionalChatOptions,
+      AdditionalMessageOptions
+    >,
+  ): Promise<ExecResponse<AdditionalMessageOptions>>;
+  async exec(
+    params:
+      | LLMChatParamsStreaming<AdditionalChatOptions, AdditionalMessageOptions>
+      | LLMChatParamsNonStreaming<
+          AdditionalChatOptions,
+          AdditionalMessageOptions
+        >,
+  ): Promise<
+    | ExecResponse<AdditionalMessageOptions>
+    | ExecStreamResponse<AdditionalMessageOptions>
+  > {
+    if (params.stream) {
+      return this.streamExec(params);
+    }
+    const logger = params.logger ?? emptyLogger;
+    const newMessages: ChatMessage<AdditionalMessageOptions>[] = [];
+    const response = await this.chat(params);
+    newMessages.push(response.message);
+    const toolCalls = getToolCallsFromResponse(response);
+    if (params.tools && toolCalls.length > 0) {
+      for (const toolCall of toolCalls) {
+        const toolResultMessage =
+          await callToolToMessage<AdditionalMessageOptions>(
+            params.tools,
+            toolCall,
+            logger,
+          );
+        if (toolResultMessage) {
+          newMessages.push(toolResultMessage);
+        }
+      }
+    }
+    return {
+      newMessages,
+      toolCalls,
+    };
+  }
+
+  async streamExec(
+    params: LLMChatParamsStreaming<
+      AdditionalChatOptions,
+      AdditionalMessageOptions
+    >,
+  ): Promise<ExecStreamResponse<AdditionalMessageOptions>> {
+    const logger = params.logger ?? emptyLogger;
+    const responseStream = await this.chat(params);
+    const iterator = responseStream[Symbol.asyncIterator]();
+    const first = await iterator.next();
+
+    // Set firstChunk to null if empty
+    const firstChunk = !first.done ? first.value : null;
+
+    const hasToolCallsInFirst =
+      firstChunk?.options && "toolCall" in firstChunk.options;
+
+    if (!hasToolCallsInFirst) {
+      let content = firstChunk?.delta ?? "";
+      let finished = false;
+      return {
+        stream: (async function* () {
+          if (firstChunk) {
+            yield firstChunk;
+          }
+          for await (const chunk of {
+            [Symbol.asyncIterator]: () => iterator,
+          }) {
+            content += chunk.delta;
+            yield chunk;
+          }
+          finished = true;
+        })(),
+        toolCalls: [],
+        newMessages() {
+          if (!finished) {
+            throw new Error(
+              "New messages are not ready yet. Call newMessages() after the stream is done.",
+            );
+          }
+          return content
+            ? [
+                {
+                  role: "assistant",
+                  content,
+                } as ChatMessage<AdditionalMessageOptions>,
+              ]
+            : [];
+        },
+      };
+    }
+    // Helper function to process a chunk
+    function processChunk(
+      chunk: ChatResponseChunk,
+      toolCallMap: Map<string, PartialToolCall>,
+    ): ChatResponseChunk | null {
+      if (chunk.options && "toolCall" in chunk.options) {
+        // update tool call map
+        for (const toolCall of chunk.options.toolCall as PartialToolCall[]) {
+          if (toolCall.id) {
+            toolCallMap.set(toolCall.id, toolCall);
+          }
+        }
+        // return the current full response with the tool calls
+        const toolCalls = Array.from(toolCallMap.values());
+        return {
+          ...chunk,
+          options: {
+            ...chunk.options,
+            toolCall: toolCalls,
+          },
+        };
+      }
+      return null;
+    }
+    // Collect for tool call
+    let fullResponse: ChatResponseChunk | null = null;
+    const toolCallMap = new Map<string, PartialToolCall>();
+    // Process first chunk
+    fullResponse = processChunk(firstChunk, toolCallMap);
+    // Process remaining chunks
+    while (true) {
+      const next = await iterator.next();
+      if (next.done) break;
+      const chunk = next.value;
+      const potentialFull = processChunk(chunk, toolCallMap);
+      if (potentialFull) {
+        fullResponse = potentialFull;
+      }
+    }
+    if (params.tools && fullResponse) {
+      const toolCalls = getToolCallsFromResponse(fullResponse);
+      const messages: ChatMessage<AdditionalMessageOptions>[] = [];
+      messages.push({
+        role: "assistant",
+        content: "",
+        options: {
+          toolCall: toolCalls,
+        } as AdditionalMessageOptions,
+      });
+      for (const toolCall of toolCalls) {
+        const toolResultMessage =
+          await callToolToMessage<AdditionalMessageOptions>(
+            params.tools,
+            toolCall,
+            logger,
+          );
+        if (toolResultMessage) {
+          messages.push(toolResultMessage);
+        }
+      }
+      return {
+        stream: (async function* () {})(),
+        newMessages() {
+          return messages;
+        },
+        toolCalls,
+      };
+    } else {
+      throw new Error("Cannot get tool calls from response");
+    }
+  }
 }

 export abstract class ToolCallLLM<
@@ -1,5 +1,4 @@
-// TODO: move to a test package
-import { ToolCallLLM } from "../llms/base";
+import { ToolCallLLM } from "./base";
 import type {
  ChatResponse,
  ChatResponseChunk,
@@ -9,7 +8,7 @@ import type {
  LLMCompletionParamsNonStreaming,
  LLMCompletionParamsStreaming,
  LLMMetadata,
-} from "../llms/type";
+} from "./type";

 export class MockLLM extends ToolCallLLM {
  metadata: LLMMetadata;
@@ -0,0 +1,64 @@
+import { type Logger } from "@llamaindex/env";
+import { callTool } from "../agent/utils.js";
+import { stringifyJSONToMessageContent } from "../utils";
+import type {
+  BaseTool,
+  ChatMessage,
+  ChatResponse,
+  ChatResponseChunk,
+  ToolCall,
+  ToolCallLLMMessageOptions,
+} from "./type";
+
+export const getToolCallsFromResponse = (
+  response:
+    | ChatResponse<ToolCallLLMMessageOptions>
+    | ChatResponseChunk<ToolCallLLMMessageOptions>,
+): ToolCall[] => {
+  let options;
+
+  if ("message" in response) {
+    options = response.message.options;
+  } else {
+    options = response.options;
+  }
+
+  if (options && "toolCall" in options) {
+    return (options.toolCall as ToolCall[]).map((toolCall) => ({
+      ...toolCall,
+      input:
+        // XXX: this is a hack openai returns parsed object for streaming, but not for
+        // non-streaming
+        typeof toolCall.input === "string"
+          ? JSON.parse(toolCall.input)
+          : toolCall.input,
+    }));
+  }
+  return [];
+};
+
+export const callToolToMessage = async <
+  AdditionalMessageOptions extends object = object,
+>(
+  tools: BaseTool[],
+  toolCall: ToolCall,
+  logger: Logger,
+): Promise<ChatMessage<AdditionalMessageOptions> | null> => {
+  const tool = tools?.find((t) => t.metadata.name === toolCall.name);
+
+  const toolOutput = await callTool(tool, toolCall, logger);
+
+  const toolResultMessage: ChatMessage<AdditionalMessageOptions> = {
+    role: "user",
+    content: stringifyJSONToMessageContent(toolOutput.output),
+    options: {
+      toolResult: {
+        id: toolCall.id,
+        result: toolOutput.output,
+        isError: toolOutput.isError,
+      },
+    } as AdditionalMessageOptions,
+  };
+
+  return toolResultMessage;
+};
@@ -1,3 +1,4 @@
+import type { Logger } from "@llamaindex/env";
 import type { Tokenizers } from "@llamaindex/env/tokenizers";
 import type { JSONSchemaType } from "ajv";
 import { z } from "zod";
@@ -95,6 +96,22 @@ export type ChatResponseChunk<
  options?: undefined | AdditionalMessageOptions;
 };

+export interface ExecResponse<
+  AdditionalMessageOptions extends object = object,
+> {
+  newMessages: ChatMessage<AdditionalMessageOptions>[];
+  toolCalls: ToolCall[];
+}
+
+export interface ExecStreamResponse<
+  AdditionalMessageOptions extends object = object,
+> {
+  stream: AsyncIterable<ChatResponseChunk<AdditionalMessageOptions>>;
+  // this is a function as while streaming, the assistant message is not ready yet - can be called after the stream is done
+  newMessages(): ChatMessage<AdditionalMessageOptions>[];
+  toolCalls: ToolCall[];
+}
+
 export interface CompletionResponse {
  text: string;
  /**
@@ -120,9 +137,10 @@ export interface LLMChatParamsBase<
  AdditionalMessageOptions extends object = object,
 > {
  messages: ChatMessage<AdditionalMessageOptions>[];
-  additionalChatOptions?: AdditionalChatOptions;
-  tools?: BaseTool[];
-  responseFormat?: z.ZodType | object;
+  additionalChatOptions?: AdditionalChatOptions | undefined;
+  tools?: BaseTool[] | undefined;
+  responseFormat?: z.ZodType | object | undefined;
+  logger?: Logger | undefined;
 }

 export interface LLMChatParamsStreaming<
@@ -39,7 +39,9 @@ export abstract class BaseMemoryBlock<
   *
   * @returns The memory block content as an array of ChatMessage.
   */
-  abstract get(): Promise<MemoryMessage<TAdditionalMessageOptions>[]>;
+  abstract get(
+    messages?: MemoryMessage<TAdditionalMessageOptions>[],
+  ): Promise<MemoryMessage<TAdditionalMessageOptions>[]>;

  /**
   * Store the messages in the memory block.
@@ -1,3 +1,4 @@
 export { BaseMemoryBlock } from "./base";
 export { FactExtractionMemoryBlock } from "./fact";
 export { StaticMemoryBlock } from "./static";
+export { VectorMemoryBlock } from "./vector";
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
github-actions[bot]	38487da65d	Release 0.11.23 (#2136 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: marcusschiesser <17126+marcusschiesser@users.noreply.github.com>	2025-07-28 14:07:23 +08:00
Marcus Schiesser	f29799e385	feat: Add toolcall callbacks to agent workflows (#2137 )	2025-07-24 15:37:14 +08:00
Marcus Schiesser	9bca30620b	fix: docs build	2025-07-23 12:55:35 +08:00
Marcus Schiesser	7224c06409	feat: Add logger and callbacks to llm.exec (#2135 )	2025-07-23 12:37:02 +08:00
github-actions[bot]	29c7cf0989	Release 0.11.22 (#2131 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>	2025-07-23 11:30:04 +08:00
Marcus Schiesser	c65a2dc4a7	chore: Deprecate community package and link to AWS package (#2134 )	2025-07-23 11:05:50 +08:00
Terence Sim	f1c5079290	docs: updated bedrock import and supported models (#2129 ) Co-authored-by: Terence Sim <40583743+InTheAxis@users.noreply.github.com>	2025-07-23 10:40:49 +08:00
Terence Sim	9ed31958a7	chore: add logger as param to AgentWorkflow constructor (#2130 ) Co-authored-by: Terence Sim <40583743+InTheAxis@users.noreply.github.com> Co-authored-by: Marcus Schiesser <marcus.schiesser@googlemail.com>	2025-07-22 16:35:28 +08:00
github-actions[bot]	e4c7113614	Release 0.11.21 (#2128 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: marcusschiesser <17126+marcusschiesser@users.noreply.github.com>	2025-07-22 12:23:58 +08:00
Thuc Pham	38da40bc98	feat: VectoryMemoryBlock (#2110 ) Co-authored-by: Marcus Schiesser <mail@marcusschiesser.de>	2025-07-22 12:18:09 +08:00
Marcus Schiesser	4d50ca4d84	chore: add streamchat test (#2122 )	2025-07-22 11:30:01 +08:00
github-actions[bot]	8b5253a297	Release (#2127 )	2025-07-21 15:40:31 -06:00
Logan	ea15e75c89	deployment docs nits (#2126 )	2025-07-21 15:30:37 -06:00
github-actions[bot]	3be87d4670	Release 0.11.20 (#2121 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: himself65 <14026360+himself65@users.noreply.github.com>	2025-07-21 09:37:44 -07:00
Terence Sim	94da13db0d	fix: azure openai streamchat empty delta throw TypeError (#2118 ) Co-authored-by: Terence Sim <40583743+InTheAxis@users.noreply.github.com>	2025-07-21 09:16:09 -07:00
Terence Sim	acd50ea99f	chore: replaced console.log with logger type from @llamaindex/env (#2123 ) Co-authored-by: Terence Sim <40583743+InTheAxis@users.noreply.github.com>	2025-07-21 09:14:06 -07:00
Adrian Lyjak	2967d57ac0	feat: default to _public agent data (#2117 )	2025-07-21 09:07:15 -07:00
Thuc Pham	a8ec08c682	fix: ensure correct message content in agent workflow (#2114 ) Co-authored-by: Marcus Schiesser <mail@marcusschiesser.de>	2025-07-21 15:13:27 +08:00
Terence Sim	678b327051	feat: added apac bedrock models (#2119 ) Co-authored-by: Marcus Schiesser <mail@marcusschiesser.de>	2025-07-21 12:13:37 +08:00
Jeremy B. Merrill	650eeb1df3	fix: GeminiEmbedding should send batches of max 100 (#2099 ) Co-authored-by: Marcus Schiesser <marcus.schiesser@googlemail.com>	2025-07-21 12:12:42 +08:00
Laurie Voss	50f6747758	Instrumenting with Google Tag Manager (in addition to Google Analytics) (#2116 )	2025-07-20 13:18:09 -07:00
github-actions[bot]	12414a6836	Release (#2113 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: marcusschiesser <17126+marcusschiesser@users.noreply.github.com>	2025-07-18 13:54:38 +08:00
Marcus Schiesser	856dd8cca8	fix: assume new models are function call models (#2112 )	2025-07-18 12:52:43 +08:00
Jerry Cheng	d8f4f6a859	Update SupabaseVectorStore.ts to fix score calculating error (#2109 ) Co-authored-by: Marcus Schiesser <marcus.schiesser@googlemail.com>	2025-07-18 12:48:47 +08:00
Logan	f594d7034f	revamp getting started flow and main index page (#2079 ) Co-authored-by: Thuc Pham <51660321+thucpn@users.noreply.github.com> Co-authored-by: Marcus Schiesser <mail@marcusschiesser.de> Co-authored-by: thucpn <thucsh2@gmail.com>	2025-07-17 16:27:28 +08:00
github-actions[bot]	c1c58feed2	Release 0.11.19 (#2105 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com> Co-authored-by: marcusschiesser <17126+marcusschiesser@users.noreply.github.com>	2025-07-17 15:44:22 +08:00
Marcus Schiesser	7ad3411766	feat: add llm.exec (#2078 )	2025-07-17 15:36:56 +08:00
Neha Prasad	a1fdb07b96	feat: multi-turn image generation support (#2106 ) Co-authored-by: Marcus Schiesser <marcus.schiesser@googlemail.com>	2025-07-17 10:30:39 +08:00
Jeremy B. Merrill	5da5b3c89c	feat: add progress callback to embeddings (#2098 ) Co-authored-by: Marcus Schiesser <mail@marcusschiesser.de>	2025-07-16 13:49:49 +08:00
r3rer3	ddc0eafbaa	feat(anthropic): stream partial tool calls (#2100 )	2025-07-15 10:06:17 -07:00
github-actions[bot]	1782554488	Release 0.11.18 (#2103 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>	2025-07-14 15:53:20 -07:00
Adrian Lyjak	a1b1598bc6	fix(cloud): add generic types into agent data responses (#2102 ) Co-authored-by: Alex Yang <himself65@outlook.com>	2025-07-14 12:01:56 -07:00
Terry Zhao	b02847ae91	fix(notion): resolve @notionhq/client dependency conflict (#2097 )	2025-07-12 11:04:06 -07:00
Alex Yang	50acb4821e	feat(cloud): use camelCase (#2096 )	2025-07-12 10:59:46 -07:00
github-actions[bot]	47a5b94b0c	Release 0.11.17 (#2095 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>	2025-07-11 21:57:02 -07:00
Alex Yang	d2be868b93	feat(cloud): missing agent api (#2094 )	2025-07-11 20:45:22 -07:00
github-actions[bot]	50d42c4129	Release 0.11.16 (#2093 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>	2025-07-11 20:13:37 -07:00
				`@@ -1 +0,0 @@`
				`export { AgentClient, createAgentClient } from "./client";`