chore: version packages (#982 )

Add ImageNode support in TypeScript (#969 )
chore: version packages (#981 )
2026-07-19 16:43:32 -04:00 · 2025-10-16 16:58:29 -06:00 · 2025-10-16 16:56:28 -06:00 · 2025-10-15 12:01:34 -06:00 · 2025-10-15 11:57:34 -06:00 · 2025-10-14 19:03:36 -06:00
69 changed files with 12992 additions and 6588 deletions
@@ -0,0 +1,8 @@
+# Changesets
+
+Hello and welcome! This folder has been automatically generated by `@changesets/cli`, a build tool that works
+with multi-package repos, or single-package repos to help you version and publish your code. You can
+find the full documentation for it [in our repository](https://github.com/changesets/changesets)
+
+We have a quick list of common questions to get you started engaging with this project in
+[our documentation](https://github.com/changesets/changesets/blob/main/docs/common-questions.md)
@@ -0,0 +1,11 @@
+{
+  "$schema": "https://unpkg.com/@changesets/config@3.1.1/schema.json",
+  "changelog": "@changesets/cli/changelog",
+  "commit": false,
+  "fixed": [],
+  "linked": [],
+  "access": "restricted",
+  "baseBranch": "main",
+  "updateInternalDependencies": "patch",
+  "ignore": []
+}
@@ -27,7 +27,7 @@ jobs:
      - uses: actions/checkout@v5

      - name: Install uv
-        uses: astral-sh/setup-uv@v6
+        uses: astral-sh/setup-uv@v7
        with:
          version: ${{ env.UV_VERSION }}

@@ -21,7 +21,7 @@ jobs:
      - uses: pnpm/action-setup@v4

      - name: Setup Node.js
-        uses: actions/setup-node@v4
+        uses: actions/setup-node@v5
        with:
          node-version-file: "ts/llama_cloud_services/.nvmrc"

@@ -30,12 +30,12 @@ jobs:

      # Initializes the CodeQL tools for scanning.
      - name: Initialize CodeQL
-        uses: github/codeql-action/init@v3
+        uses: github/codeql-action/init@v4
        with:
          languages: python
          dependency-caching: true

      - name: Perform CodeQL Analysis
-        uses: github/codeql-action/analyze@v3
+        uses: github/codeql-action/analyze@v4
        with:
          category: "/language:python"
@@ -22,7 +22,7 @@ jobs:
        with:
          fetch-depth: ${{ github.event_name == 'pull_request' && 2 || 0 }}
      - name: Install uv
-        uses: astral-sh/setup-uv@v6
+        uses: astral-sh/setup-uv@v7
        with:
          version: ${{ env.UV_VERSION }}

@@ -31,7 +31,7 @@ jobs:

      - uses: pnpm/action-setup@v4
      - name: Setup Node.js
-        uses: actions/setup-node@v4
+        uses: actions/setup-node@v5
        with:
          node-version-file: "ts/llama_cloud_services/.nvmrc"
      - name: Install dependencies
@@ -1,66 +0,0 @@
-name: Publish Release - Python
-
-on:
-  push:
-    tags:
-      - "v*"
-
-  workflow_dispatch:
-
-env:
-  UV_VERSION: "0.7.20"
-
-jobs:
-  build-n-publish:
-    name: Build and publish to PyPI
-    if: github.repository == 'run-llama/llama_cloud_services'
-    runs-on: ubuntu-latest
-
-    steps:
-      - uses: actions/checkout@v5
-
-      - name: Install uv
-        uses: astral-sh/setup-uv@v6
-        with:
-          version: ${{ env.UV_VERSION }}
-
-      - name: Set up Python
-        run: uv python install
-
-      - name: Display Python version
-        run: python --version
-
-      - name: Build
-        working-directory: py
-        run: uv build
-
-      - name: Test installing built package
-        shell: bash
-        working-directory: py
-        run: |
-          uv venv
-          uv pip install dist/*.whl
-
-      - name: Publish package
-        shell: bash
-        working-directory: py
-        run: uv publish --token ${{ secrets.LLAMA_PARSE_PYPI_TOKEN }}
-
-      - name: Build and publish llama-parse
-        working-directory: py/llama_parse/
-        run: |
-          uv build
-          uv publish --token ${{ secrets.LLAMA_PARSE_PYPI_TOKEN }}
-
-      - name: Create GitHub Release
-        id: create_release
-        uses: actions/create-release@v1
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} # This token is provided by Actions, you do not need to create your own token
-        with:
-          tag_name: ${{ github.ref }}
-          release_name: ${{ github.ref }} - LlamaCloud Services PY
-          artifacts: "py/**/dist/*"
-          generateReleaseNotes: true
-          draft: false
-          prerelease: false
@@ -1,52 +0,0 @@
-name: Publish Release - TypeScript
-
-on:
-  push:
-    tags:
-      - "llama-cloud-services@*"
-
-jobs:
-  build-and-publish:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout Repo
-        uses: actions/checkout@v5
-
-      - uses: pnpm/action-setup@v4
-
-      - name: Setup Node.js
-        uses: actions/setup-node@v4
-        with:
-          node-version-file: "ts/llama_cloud_services/.nvmrc"
-
-      - name: Install dependencies
-        run: pnpm install --no-frozen-lockfile
-
-      - name: Run Build
-        working-directory: ts/llama_cloud_services/
-        run: pnpm build
-
-      - name: Build tarball
-        run: |
-          pnpm pack
-        working-directory: ts/llama_cloud_services
-
-      - name: Setup npm authentication
-        run: echo "//registry.npmjs.org/:_authToken=${NPM_TOKEN}" > ~/.npmrc
-        env:
-          NPM_TOKEN: ${{ secrets.NPM_TOKEN }}
-
-      - name: Release
-        working-directory: ts/llama_cloud_services
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          NPM_TOKEN: ${{ secrets.NPM_TOKEN }}
-        run: pnpm publish --access public --no-git-checks
-
-      - name: Create release
-        uses: ncipollo/release-action@v1
-        with:
-          artifacts: "ts/llama_cloud_services/llama-cloud-services*.tgz"
-          name: Release ${{ github.ref_name }} - LlamaCloud Services TS
-          generateReleaseNotes: true
-          token: ${{ secrets.GITHUB_TOKEN }}
@@ -22,7 +22,7 @@ jobs:
        with:
          fetch-depth: 0
      - name: Install uv
-        uses: astral-sh/setup-uv@v6
+        uses: astral-sh/setup-uv@v7
        with:
          version: ${{ env.UV_VERSION }}

@@ -26,7 +26,7 @@ jobs:
        with:
          fetch-depth: 0
      - name: Install uv
-        uses: astral-sh/setup-uv@v6
+        uses: astral-sh/setup-uv@v7
        with:
          version: ${{ env.UV_VERSION }}

@@ -24,7 +24,7 @@ jobs:
      - uses: actions/checkout@v5
      - uses: pnpm/action-setup@v4
      - name: Setup Node.js
-        uses: actions/setup-node@v4
+        uses: actions/setup-node@v5
        with:
          node-version-file: "ts/llama_cloud_services/.nvmrc"
      - name: Install dependencies
@@ -0,0 +1,61 @@
+name: Version Bump and Release
+
+on:
+  push:
+    branches:
+      - main
+
+concurrency: ${{ github.workflow }}-${{ github.ref }}
+
+jobs:
+  release:
+    name: Release
+    runs-on: ubuntu-latest
+    # Only run on main branch pushes
+    if: github.ref == 'refs/heads/main'
+    steps:
+      - name: Checkout Repo
+        uses: actions/checkout@v5
+
+      - uses: pnpm/action-setup@v4
+
+      - name: Setup Node.js
+        uses: actions/setup-node@v5
+        with:
+          node-version: "22"
+          cache: "pnpm"
+
+      - name: Setup Python
+        uses: actions/setup-python@v6
+        with:
+          python-version: "3.11"
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@v7
+
+      - name: Install dependencies
+        run: pnpm install
+
+      - name: Add auth token to .npmrc file
+        run: |
+          cat << EOF >> ".npmrc"
+            //registry.npmjs.org/:_authToken=$NPM_TOKEN
+          EOF
+        env:
+          NPM_TOKEN: ${{ secrets.NPM_TOKEN }}
+
+      - name: Create Release Pull Request or Publish packages
+        id: changesets
+        uses: changesets/action@v1
+        with:
+          commit: "chore: version packages"
+          title: "chore: version packages"
+          # Custom version script
+          version: pnpm -w run version
+          # Custom publish script
+          publish: pnpm -w run publish
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          NPM_TOKEN: ${{ secrets.NPM_TOKEN }}
+          UV_PUBLISH_TOKEN: ${{ secrets.PYPI_TOKEN }}
+          LLAMA_PARSE_PYPI_TOKEN: ${{ secrets.LLAMA_PARSE_PYPI_TOKEN }}
@@ -9,3 +9,4 @@ __pycache__/
 node_modules/
 .turbo/
 dist/
+.npmrc
@@ -29,7 +29,7 @@ repos:
      - id: black-jupyter
        name: black-src
        alias: black
-        exclude: ".*uv.lock"
+        exclude: ".*uv.lock|examples/extract/solar_panel_e2e_comparison.ipynb"
  - repo: https://github.com/pre-commit/mirrors-mypy
    rev: v1.0.1
    hooks:
@@ -7,7 +7,7 @@
   "source": [
    "# Extraction and Analysis over a Fidelity Multi-Fund Annual Report\n",
    "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services-demo/blob/main/examples/extract/asset_manager_fund_analysis.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/extract/asset_manager_fund_analysis.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
    "\n",
    "In this notebook we show you how to create an agentic document workflow over a complex document that contains annual reports for multiple funds - each fund reports financials in a standardized reporting structure, and it's all consolidated in the same document.\n",
    "\n",
@@ -7,7 +7,7 @@
   "source": [
    "# Automotive Equity Research: A Multi-Step Agentic Workflow\n",
    "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services-demo/blob/main/examples/extract/automotive_sector_analysis.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/extract/automotive_sector_analysis.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
    "\n",
    "This notebook demonstrates an end‑to‑end agentic workflow using LlamaExtract and the LlamaIndex event‑driven workflow framework for automotive sector analysis.\n",
    "\n",
@@ -1035,7 +1035,7 @@
 ],
 "metadata": {
  "kernelspec": {
-   "display_name": ".venv",
+   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
@@ -1052,5 +1052,5 @@
  }
 },
 "nbformat": 4,
- "nbformat_minor": 2
+ "nbformat_minor": 4
 }
@@ -0,0 +1,765 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Complete Parse → Classify → Extract Workflow with LlamaCloud Services\n",
+    "\n",
+    "This notebook demonstrates the complete workflow for processing documents using LlamaCloud services:\n",
+    "1. **Parse** - Extract and convert documents to markdown\n",
+    "2. **Classify** - Categorize documents based on their content\n",
+    "3. **Extract** - Extract structured data using the markdown as input via SourceText\n",
+    "\n",
+    "## Overview of the Workflow\n",
+    "\n",
+    "### 1. Parse Phase\n",
+    "- Use `LlamaParse` to convert documents (PDFs, Word docs, etc.) into structured formats\n",
+    "- Extract markdown content that preserves document structure\n",
+    "- Get both raw text and markdown representations\n",
+    "\n",
+    "### 2. Classify Phase\n",
+    "- Use `ClassifyClient` to categorize documents based on content\n",
+    "- Apply classification rules to route documents appropriately\n",
+    "- Handle different document types with specific processing logic\n",
+    "\n",
+    "### 3. Extract Phase\n",
+    "- Use `LlamaExtract` with `SourceText` to extract structured data\n",
+    "- Pass the markdown content as input for more accurate extraction\n",
+    "- Define custom schemas for structured data extraction\n",
+    "\n",
+    "Let's walk through each step with practical examples."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Setup and Installation"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Install required packages\n",
+    "!pip install llama-cloud-services\n",
+    "!pip install python-dotenv"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "✅ API key configured\n"
+     ]
+    }
+   ],
+   "source": [
+    "import os\n",
+    "import nest_asyncio\n",
+    "from getpass import getpass\n",
+    "from dotenv import load_dotenv\n",
+    "\n",
+    "# Load environment variables\n",
+    "load_dotenv()\n",
+    "nest_asyncio.apply()\n",
+    "\n",
+    "# Set up API key\n",
+    "os.environ[\"LLAMA_CLOUD_API_KEY\"] = \"\"  # edit it\n",
+    "\n",
+    "# Setup Base URL\n",
+    "# os.envrion[\"LLAMA_CLOUD_BASE_URL\"] = \"https://api.cloud.eu.llamaindex.ai/\" # update if necessay\n",
+    "\n",
+    "print(\"✅ API key configured\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Download Sample Documents\n",
+    "\n",
+    "Let's download some sample documents to work with:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "📁 financial_report.pdf already exists\n",
+      "📁 technical_spec.pdf already exists\n",
+      "\n",
+      "📂 Sample documents ready!\n"
+     ]
+    }
+   ],
+   "source": [
+    "import requests\n",
+    "import os\n",
+    "\n",
+    "# Create directory for sample documents\n",
+    "os.makedirs(\"sample_docs\", exist_ok=True)\n",
+    "\n",
+    "# Download sample documents\n",
+    "docs_to_download = {\n",
+    "    \"financial_report.pdf\": \"https://raw.githubusercontent.com/run-llama/llama_index/main/docs/docs/examples/data/10k/uber_2021.pdf\",\n",
+    "    \"technical_spec.pdf\": \"https://www.ti.com/lit/ds/symlink/lm317.pdf\",\n",
+    "}\n",
+    "\n",
+    "for filename, url in docs_to_download.items():\n",
+    "    filepath = f\"sample_docs/{filename}\"\n",
+    "    if not os.path.exists(filepath):\n",
+    "        print(f\"Downloading {filename}...\")\n",
+    "        response = requests.get(url)\n",
+    "        if response.status_code == 200:\n",
+    "            with open(filepath, \"wb\") as f:\n",
+    "                f.write(response.content)\n",
+    "            print(f\"✅ Downloaded {filename}\")\n",
+    "        else:\n",
+    "            print(f\"❌ Failed to download {filename}\")\n",
+    "    else:\n",
+    "        print(f\"📁 {filename} already exists\")\n",
+    "\n",
+    "print(\"\\n📂 Sample documents ready!\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Phase 1: Document Parsing\n",
+    "\n",
+    "First, let's parse our documents using LlamaParse to extract clean markdown content."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "🔄 Parsing documents...\n",
+      "Started parsing the file under job_id 8a8c76f9-354d-4275-91d8-312ff1adc762\n",
+      "...✅ Parsed financial report (Job ID: 8a8c76f9-354d-4275-91d8-312ff1adc762)\n",
+      "Started parsing the file under job_id 7e603448-ed80-4d18-948b-6801ed51c41b\n",
+      "✅ Parsed technical spec (Job ID: 7e603448-ed80-4d18-948b-6801ed51c41b)\n",
+      "\n",
+      "📄 Parsing complete!\n"
+     ]
+    }
+   ],
+   "source": [
+    "from llama_cloud_services.parse.base import LlamaParse\n",
+    "from llama_cloud_services.parse.utils import ResultType\n",
+    "import asyncio\n",
+    "\n",
+    "# Initialize the parser\n",
+    "parser = LlamaParse(\n",
+    "    result_type=ResultType.MD,  # Get markdown output\n",
+    "    verbose=True,\n",
+    "    language=\"en\",\n",
+    "    # Premium mode for better accuracy\n",
+    "    premium_mode=True,\n",
+    "    # Extract tables as HTML for better structure\n",
+    "    output_tables_as_HTML=True,\n",
+    "    # Parse only first few pages for demo\n",
+    ")\n",
+    "\n",
+    "print(\"🔄 Parsing documents...\")\n",
+    "\n",
+    "# Parse the financial report\n",
+    "financial_result = await parser.aparse(\"sample_docs/financial_report.pdf\")\n",
+    "print(f\"✅ Parsed financial report (Job ID: {financial_result.job_id})\")\n",
+    "\n",
+    "# Parse the technical specification\n",
+    "technical_result = await parser.aparse(\"sample_docs/technical_spec.pdf\")\n",
+    "print(f\"✅ Parsed technical spec (Job ID: {technical_result.job_id})\")\n",
+    "\n",
+    "print(\"\\n📄 Parsing complete!\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Extract Markdown Content\n",
+    "\n",
+    "Now let's get the markdown content from our parsed documents:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "📋 Financial Report Markdown (first 500 chars):\n",
+      "\n",
+      "\n",
+      "# UNITED STATES\n",
+      "# SECURITIES AND EXCHANGE COMMISSION\n",
+      "Washington, D.C. 20549\n",
+      "\n",
+      "## FORM 10-K\n",
+      "\n",
+      "(Mark One)\n",
+      "\n",
+      "☒ ANNUAL REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934\n",
+      "For the fiscal year ended December 31, 2021\n",
+      "OR\n",
+      "☐ TRANSITION REPORT PURSUANT TO SECTION 13 OR 15(d) OF THE SECURITIES EXCHANGE ACT OF 1934\n",
+      "For the transition period from_____ to _____\n",
+      "Commission File Number: 001-38902\n",
+      "\n",
+      "# UBER TECHNOLOGIES, INC.\n",
+      "(Exact name of registrant as specified in its charter)\n",
+      "\n",
+      "Delaware\n",
+      "...\n",
+      "\n",
+      "📋 Technical Spec Markdown (first 500 chars):\n",
+      "\n",
+      "\n",
+      "LM317\n",
+      "SLVS044Z – SEPTEMBER 1997 – REVISED APRIL 2025\n",
+      "\n",
+      "# LM317 3-Pin Adjustable Regulator\n",
+      "\n",
+      "## 1 Features\n",
+      "\n",
+      "• Output voltage range:\n",
+      "  – Adjustable: 1.25V to 37V\n",
+      "• Output current: 1.5A\n",
+      "• Line regulation: 0.01%/V (typ)\n",
+      "• Load regulation: 0.1% (typ)\n",
+      "• Internal short-circuit current limiting\n",
+      "• Thermal overload protection\n",
+      "• Output safe-area compensation (new chip)\n",
+      "• PSRR: 80dB at 120Hz for CADJ = 10μF (new chip)\n",
+      "• Packages:\n",
+      "  – 4-pin, SOT-223 (DCY)\n",
+      "  – 3-pin, TO-263 (KTT)\n",
+      "  – 3-pin, TO-220 (KCS, KCT),\n",
+      "...\n",
+      "\n",
+      "📏 Financial report markdown length: 1348671 characters\n",
+      "📏 Technical spec markdown length: 90971 characters\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Get markdown content from parsed documents\n",
+    "financial_markdown = await financial_result.aget_markdown()\n",
+    "technical_markdown = await technical_result.aget_markdown()\n",
+    "\n",
+    "print(\"📋 Financial Report Markdown (first 500 chars):\")\n",
+    "print(financial_markdown[:500])\n",
+    "print(\"...\\n\")\n",
+    "\n",
+    "print(\"📋 Technical Spec Markdown (first 500 chars):\")\n",
+    "print(technical_markdown[:500])\n",
+    "print(\"...\\n\")\n",
+    "\n",
+    "print(f\"📏 Financial report markdown length: {len(financial_markdown)} characters\")\n",
+    "print(f\"📏 Technical spec markdown length: {len(technical_markdown)} characters\")\n",
+    "\n",
+    "document_texts = [financial_markdown, technical_markdown]"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Phase 2: Document Classification\n",
+    "\n",
+    "Next, let's classify our documents based on their content using the ClassifyClient."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "🏷️  Setting up document classification...\n",
+      "📝 Created 3 classification rules\n"
+     ]
+    }
+   ],
+   "source": [
+    "from llama_cloud_services.beta.classifier.client import ClassifyClient\n",
+    "from llama_cloud.types import ClassifierRule\n",
+    "from llama_cloud_services.files.client import FileClient\n",
+    "from llama_cloud.client import AsyncLlamaCloud\n",
+    "\n",
+    "# Initialize the classify client\n",
+    "api_key = os.environ[\"LLAMA_CLOUD_API_KEY\"]\n",
+    "classify_client = ClassifyClient.from_api_key(api_key)\n",
+    "\n",
+    "print(\"🏷️  Setting up document classification...\")\n",
+    "\n",
+    "# Define classification rules\n",
+    "classification_rules = [\n",
+    "    ClassifierRule(\n",
+    "        type=\"financial_document\",\n",
+    "        description=\"Documents containing financial data, revenue, expenses, SEC filings, or financial statements\",\n",
+    "    ),\n",
+    "    ClassifierRule(\n",
+    "        type=\"technical_specification\",\n",
+    "        description=\"Technical datasheets, component specifications, engineering documents, or technical manuals\",\n",
+    "    ),\n",
+    "    ClassifierRule(\n",
+    "        type=\"general_document\",\n",
+    "        description=\"General business documents, contracts, or other unspecified document types\",\n",
+    "    ),\n",
+    "]\n",
+    "\n",
+    "print(f\"📝 Created {len(classification_rules)} classification rules\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Phase 3: Structured Data Extraction using SourceText\n",
+    "\n",
+    "Now comes the key part - using the markdown content as input for structured data extraction via SourceText."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "⚙️  LlamaExtract initialized\n"
+     ]
+    }
+   ],
+   "source": [
+    "from llama_cloud_services.extract.extract import LlamaExtract, SourceText\n",
+    "from llama_cloud.types import ExtractConfig, ExtractMode\n",
+    "from pydantic import BaseModel, Field\n",
+    "from typing import List, Optional\n",
+    "\n",
+    "# Initialize LlamaExtract\n",
+    "llama_extract = LlamaExtract(api_key=api_key, verbose=True)\n",
+    "\n",
+    "print(\"⚙️  LlamaExtract initialized\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Define Extraction Schemas\n",
+    "\n",
+    "Let's define different schemas for different document types:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "📋 Extraction schemas defined\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Schema for financial documents\n",
+    "class FinancialMetrics(BaseModel):\n",
+    "    company_name: str = Field(description=\"Name of the company\")\n",
+    "    document_type: str = Field(\n",
+    "        description=\"Type of financial document (10-K, 10-Q, annual report, etc.)\"\n",
+    "    )\n",
+    "    fiscal_year: int = Field(description=\"Fiscal year of the report\")\n",
+    "    revenue_2021: str = Field(description=\"Total revenue in 2021\")\n",
+    "    net_income_2021: str = Field(description=\"Net income in 2021\")\n",
+    "    key_business_segments: List[str] = Field(\n",
+    "        default=[], description=\"Main business segments or divisions\"\n",
+    "    )\n",
+    "    risk_factors: List[str] = Field(\n",
+    "        default=[], description=\"Key risk factors mentioned\"\n",
+    "    )\n",
+    "\n",
+    "\n",
+    "# Schema for technical specifications\n",
+    "class VoltageRange(BaseModel):\n",
+    "    min_voltage: Optional[float] = Field(description=\"Minimum voltage\")\n",
+    "    max_voltage: Optional[float] = Field(description=\"Maximum voltage\")\n",
+    "    unit: str = Field(default=\"V\", description=\"Voltage unit\")\n",
+    "\n",
+    "\n",
+    "class TechnicalSpec(BaseModel):\n",
+    "    component_name: str = Field(description=\"Name of the technical component\")\n",
+    "    manufacturer: Optional[str] = Field(description=\"Manufacturer name\")\n",
+    "    part_number: Optional[str] = Field(description=\"Part or model number\")\n",
+    "    description: str = Field(description=\"Brief description of the component\")\n",
+    "    operating_voltage: Optional[VoltageRange] = Field(\n",
+    "        description=\"Operating voltage range\"\n",
+    "    )\n",
+    "    maximum_current: Optional[float] = Field(\n",
+    "        description=\"Maximum current rating in amperes\"\n",
+    "    )\n",
+    "    key_features: List[str] = Field(\n",
+    "        default=[], description=\"Key features and capabilities\"\n",
+    "    )\n",
+    "    applications: List[str] = Field(default=[], description=\"Typical applications\")\n",
+    "\n",
+    "\n",
+    "print(\"📋 Extraction schemas defined\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Complete Workflow Summary\n",
+    "\n",
+    "Let's create a function that demonstrates the complete workflow:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "🔧 Workflow function defined!\n"
+     ]
+    }
+   ],
+   "source": [
+    "import tempfile\n",
+    "from pathlib import Path\n",
+    "from llama_cloud import ExtractConfig\n",
+    "\n",
+    "\n",
+    "async def complete_document_workflow(markdown_content: str):\n",
+    "    \"\"\"\n",
+    "    Complete workflow: Parse → Classify → Extract\n",
+    "    \"\"\"\n",
+    "    print(f\"🚀 Starting complete workflow\")\n",
+    "    print(\"=\" * 60)\n",
+    "\n",
+    "    # Step 1: Classify\n",
+    "    print(\"🏷️  Step 2: Classifying document...\")\n",
+    "\n",
+    "    with tempfile.NamedTemporaryFile(\n",
+    "        mode=\"w\", suffix=\".md\", delete=False, encoding=\"utf-8\"\n",
+    "    ) as tmp:\n",
+    "        tmp.write(markdown_content)\n",
+    "        temp_path = Path(tmp.name)\n",
+    "\n",
+    "    print(temp_path)\n",
+    "\n",
+    "    classification = await classify_client.aclassify_file_path(\n",
+    "        rules=classification_rules, file_input_path=str(temp_path)\n",
+    "    )\n",
+    "    doc_type = classification.items[0].result.type\n",
+    "    confidence = classification.items[0].result.confidence\n",
+    "    print(f\"   ✅ Classified as: {doc_type} (confidence: {confidence:.2f})\")\n",
+    "\n",
+    "    # Step 2: Extract based on classification\n",
+    "    print(\"🔍 Step 3: Extracting structured data using SourceText...\")\n",
+    "    source_text = SourceText(\n",
+    "        text_content=markdown_content,\n",
+    "        filename=f\"{os.path.basename(temp_path)}_markdown.md\",\n",
+    "    )\n",
+    "\n",
+    "    # Choose schema based on classification\n",
+    "    if \"financial\" in doc_type.lower():\n",
+    "        schema = FinancialMetrics\n",
+    "        print(\"   📊 Using FinancialMetrics schema\")\n",
+    "    elif \"technical\" in doc_type.lower():\n",
+    "        schema = TechnicalSpec\n",
+    "        print(\"   🔧 Using TechnicalSpec schema\")\n",
+    "    else:\n",
+    "        schema = FinancialMetrics  # Default fallback\n",
+    "        print(\"   📊 Using default FinancialMetrics schema\")\n",
+    "\n",
+    "    extract_config = ExtractConfig(\n",
+    "        extraction_mode=\"BALANCED\",\n",
+    "    )\n",
+    "\n",
+    "    extraction_result = llama_extract.extract(\n",
+    "        data_schema=schema, config=extract_config, files=source_text\n",
+    "    )\n",
+    "\n",
+    "    print(\"   ✅ Extraction complete!\")\n",
+    "\n",
+    "    return {\n",
+    "        \"file_path\": temp_path,\n",
+    "        \"markdown_length\": len(markdown_content),\n",
+    "        \"classification\": doc_type,\n",
+    "        \"confidence\": confidence,\n",
+    "        \"extracted_data\": extraction_result.data,\n",
+    "        \"markdown_sample\": markdown_content[:200] + \"...\"\n",
+    "        if len(markdown_content) > 200\n",
+    "        else markdown_content,\n",
+    "    }\n",
+    "\n",
+    "\n",
+    "print(\"🔧 Workflow function defined!\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Run Complete Workflow on Both Documents"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "🚀 Starting complete workflow\n",
+      "============================================================\n",
+      "🏷️  Step 2: Classifying document...\n",
+      "/var/folders/g6/4b5lpp5974gcpr890ybhbw4r0000gn/T/tmpos3b62tm.md\n",
+      "   ✅ Classified as: financial_document (confidence: 1.00)\n",
+      "🔍 Step 3: Extracting structured data using SourceText...\n",
+      "   📊 Using FinancialMetrics schema\n",
+      "..   ✅ Extraction complete!\n",
+      "\n",
+      "============================================================\n",
+      "\n",
+      "🚀 Starting complete workflow\n",
+      "============================================================\n",
+      "🏷️  Step 2: Classifying document...\n",
+      "/var/folders/g6/4b5lpp5974gcpr890ybhbw4r0000gn/T/tmpppz9ub_m.md\n",
+      "   ✅ Classified as: technical_specification (confidence: 1.00)\n",
+      "🔍 Step 3: Extracting structured data using SourceText...\n",
+      "   🔧 Using TechnicalSpec schema\n",
+      "   ✅ Extraction complete!\n",
+      "\n",
+      "============================================================\n",
+      "\n",
+      "📋 Processed 2 documents successfully!\n"
+     ]
+    }
+   ],
+   "source": [
+    "# Process both documents through the complete workflow\n",
+    "results = []\n",
+    "\n",
+    "for doc_text in document_texts:\n",
+    "    try:\n",
+    "        result = await complete_document_workflow(doc_text)\n",
+    "        results.append(result)\n",
+    "        print(\"\\n\" + \"=\" * 60 + \"\\n\")\n",
+    "    except Exception as e:\n",
+    "        print(f\"❌ Error processing {doc_path}: {str(e)}\")\n",
+    "        print(\"\\n\" + \"=\" * 60 + \"\\n\")\n",
+    "\n",
+    "print(f\"📋 Processed {len(results)} documents successfully!\")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Final Results Summary"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "📈 COMPLETE WORKFLOW RESULTS SUMMARY\n",
+      "======================================================================\n",
+      "\n",
+      "📄 Document 1: tmpos3b62tm.md\n",
+      "   📊 Classification: financial_document (confidence: 1.00)\n",
+      "   📝 Markdown length: 1,348,671 characters\n",
+      "   📋 Markdown sample: \n",
+      "\n",
+      "# UNITED STATES\n",
+      "# SECURITIES AND EXCHANGE COMMISSION\n",
+      "Washington, D.C. 20549\n",
+      "\n",
+      "## FORM 10-K\n",
+      "\n",
+      "(Mark O...\n",
+      "   🎯 Extracted fields: 7 fields\n",
+      "   • company_name: Uber Technologies, Inc.\n",
+      "   • document_type: Annual Report on Form 10-K\n",
+      "   • fiscal_year: 2021\n",
+      "   • revenue_2021: $21,764\n",
+      "   • net_income_2021: $(496)\n",
+      "   • key_business_segments: ['Mobility', 'Delivery', 'Freight', 'All Other (including former New Mobility, e-bikes, e-scooters, Advanced Technologies Group and other technology programs)']\n",
+      "   • risk_factors: [\"The company faces numerous risk factors across its business operations and environment. The COVID-19 pandemic and related mitigation measures have adversely affected parts of the business, including reduced demand for Mobility offerings and creating ongoing uncertainties. The company's operational and financial performance is influenced by competitive pressure in the mobility, delivery, and logistics industries, characterized by well-established alternatives, low barriers to entry, and low switching costs. Driver classification risks exist if Drivers are deemed employees, workers, or quasi-employees rather than independent contractors, exposing the company to legal actions and financial liabilities globally. Competition challenges require the company to sometimes lower fares, offer incentives, and promotions, which impacts profitability. There are significant operating losses historically with substantial future operating expense increases anticipated, and the ability to achieve or maintain profitability is uncertain. Network value depends on maintaining critical mass among Drivers, consumers, merchants, shippers, and carriers, and failures to do so diminish platform attractiveness. Brand and reputation maintenance is critical, with exposure to negative publicity, media coverage, and risks from associated companies' brands or licensed brands in joint ventures.\\n\\nOperational risks include historical workplace culture and compliance challenges, management complexity due to rapid growth, technological infrastructure issues potentially causing disruptions or poor user experience, and security or data privacy breaches that could impact revenue and reputation. Platform users may engage in or be subjected to criminal, violent, or dangerous activity leading to safety incidents and legal actions. New offerings and technologies investments are inherently risky without guaranteed benefits. Economic conditions, inflation, and increased costs (fuel, food, labor, energy) may negatively impact results. Regulatory risks are extensive and global, involving payment and financial services compliance, licensing, anti-money laundering laws, data privacy (GDPR, CCPA, LGPD), and labor laws. Legal and regulatory investigations and inquiries, including antitrust, FCPA, labor classification, data protection, and intellectual property matters, pose risks of fines, penalties, operational changes, and increased costs.\\n\\nGeopolitical and jurisdictional risks include operating limitations or bans in some locations, currency exchange risk, and complex evolving regulations with the potential for fines and loss of licenses or permits. Insurance risks include potential inadequacy of reserves, liability exposure from accidents or impersonation, and insurer insolvency. Driver qualification requirements and background checks may increase costs or fail to expose all relevant information, with associated insurance cost risks and potential for courtroom or regulatory challenges to pricing models.\\n\\nFinancial risks comprise significant accumulated deficits, requirement for additional capital with uncertain availability, debt obligations, tax exposure including uncertain positions and observed changes in tax laws, and volatility in common stock price with no expected cash dividends. Accounting judgments and estimates involve critical assumptions affecting reported financial metrics related to goodwill, revenue recognition, incentive accruals, and stock-based compensation. Cybersecurity risks include exposures to malware, ransomware, phishing, and other cyberattacks. Climate change presents physical and transitional risks that may impact operations and costs, and failure to meet climate commitments may have operational and reputational consequences.\\n\\nOther risks include potential liability under anti-corruption and anti-terrorism laws, adverse effects from defaults under debt agreements, limitations in takeover actions due to corporate governance provisions, and the impact of non-GAAP financial measure limitations. Overall, these diverse and interconnected risk factors contribute to significant uncertainty regarding the company's future business prospects, operating results, and financial condition.\"]\n",
+      "\n",
+      "📄 Document 2: tmpppz9ub_m.md\n",
+      "   📊 Classification: technical_specification (confidence: 1.00)\n",
+      "   📝 Markdown length: 90,971 characters\n",
+      "   📋 Markdown sample: \n",
+      "\n",
+      "LM317\n",
+      "SLVS044Z – SEPTEMBER 1997 – REVISED APRIL 2025\n",
+      "\n",
+      "# LM317 3-Pin Adjustable Regulator\n",
+      "\n",
+      "## 1 Fea...\n",
+      "   🎯 Extracted fields: 8 fields\n",
+      "   • component_name: LM317\n",
+      "   • manufacturer: Texas Instruments\n",
+      "   • part_number: LM317\n",
+      "   • description: The LM317 is an adjustable three-pin, positive-voltage regulator capable of supplying up to 1.5A over an output voltage range of 1.25V to 37V. It features line and load regulation, internal current limiting, thermal overload protection, and safe operating area compensation.\n",
+      "   • operating_voltage: {'min_voltage': 1.25, 'max_voltage': 37.0, 'unit': 'V'}\n",
+      "   • maximum_current: 1.5\n",
+      "   • key_features: ['Adjustable output voltage: 1.25V to 37V', 'Output current up to 1.5A', 'Line regulation: 0.01%/V (typical)', 'Load regulation: 0.1% (typical)', 'Internal short-circuit current limiting', 'Thermal overload protection', 'Output safe-area compensation', 'High power supply rejection ratio (PSRR): 80dB at 120Hz (new chip)', 'Available in SOT-223, TO-263, and TO-220 packages']\n",
+      "   • applications: ['Multifunction printers', 'AC drive power stage modules', 'Electricity meters', 'Servo drive control modules', 'Merchant network and server power supply units']\n",
+      "\n",
+      "✨ Workflow completed successfully!\n",
+      "\n",
+      "📚 Key Learnings:\n",
+      "   • Parse: Converted documents to clean markdown format\n",
+      "   • Classify: Automatically categorized document types\n",
+      "   • Extract: Used SourceText with markdown for structured data extraction\n",
+      "   • The markdown content provides much better context for extraction than raw PDFs\n"
+     ]
+    }
+   ],
+   "source": [
+    "print(\"📈 COMPLETE WORKFLOW RESULTS SUMMARY\")\n",
+    "print(\"=\" * 70)\n",
+    "\n",
+    "for i, result in enumerate(results, 1):\n",
+    "    print(f\"\\n📄 Document {i}: {os.path.basename(result['file_path'])}\")\n",
+    "    print(\n",
+    "        f\"   📊 Classification: {result['classification']} (confidence: {result['confidence']:.2f})\"\n",
+    "    )\n",
+    "    print(f\"   📝 Markdown length: {result['markdown_length']:,} characters\")\n",
+    "    print(f\"   📋 Markdown sample: {result['markdown_sample'][:100]}...\")\n",
+    "    print(f\"   🎯 Extracted fields: {len(result['extracted_data'])} fields\")\n",
+    "\n",
+    "    # Print all key–value pairs\n",
+    "    extracted = result[\"extracted_data\"]\n",
+    "    for key, value in extracted.items():\n",
+    "        print(f\"   • {key}: {value}\")\n",
+    "\n",
+    "print(\"\\n✨ Workflow completed successfully!\")\n",
+    "print(\"\\n📚 Key Learnings:\")\n",
+    "print(\"   • Parse: Converted documents to clean markdown format\")\n",
+    "print(\"   • Classify: Automatically categorized document types\")\n",
+    "print(\"   • Extract: Used SourceText with markdown for structured data extraction\")\n",
+    "print(\n",
+    "    \"   • The markdown content provides much better context for extraction than raw PDFs\"\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Conclusion\n",
+    "\n",
+    "This notebook demonstrated the complete **Parse → Classify → Extract** workflow using LlamaCloud services:\n",
+    "\n",
+    "### Key Components:\n",
+    "\n",
+    "1. **LlamaParse** (`llama_cloud_services.parse.base.LlamaParse`):\n",
+    "   - Converts documents to clean, structured markdown\n",
+    "   - Preserves document structure and formatting\n",
+    "   - Handles various file types (PDF, DOCX, etc.)\n",
+    "\n",
+    "2. **ClassifyClient** (`llama_cloud_services.beta.classifier.client.ClassifyClient`):\n",
+    "   - Automatically categorizes documents based on content\n",
+    "   - Uses customizable rules for classification\n",
+    "   - Provides confidence scores for classifications\n",
+    "\n",
+    "3. **LlamaExtract with SourceText** (`llama_cloud_services.extract.extract.LlamaExtract`, `SourceText`):\n",
+    "   - Extracts structured data using custom Pydantic schemas\n",
+    "   - **SourceText** allows using markdown content as input instead of raw files\n",
+    "   - Provides much better extraction accuracy when using processed markdown\n",
+    "\n",
+    "### Workflow Benefits:\n",
+    "\n",
+    "- **Better Accuracy**: Using markdown from parsing provides cleaner, more structured input for extraction\n",
+    "- **Automatic Routing**: Classification allows different processing logic for different document types\n",
+    "- **Structured Output**: Custom schemas ensure consistent, structured data extraction\n",
+    "- **Flexible Input**: SourceText supports text content, file paths, and bytes\n",
+    "\n",
+    "### Key Insights:\n",
+    "\n",
+    "1. **SourceText is the bridge**: It allows you to pass the clean markdown content from parsing directly to extraction\n",
+    "2. **Markdown improves extraction**: Pre-processed markdown provides much better context than raw PDFs\n",
+    "3. **Classification enables smart routing**: Different document types can use different extraction schemas\n",
+    "4. **End-to-end automation**: The entire workflow can be automated for production use\n",
+    "\n",
+    "This approach is ideal for production document processing pipelines where you need to:\n",
+    "- Process various document types automatically\n",
+    "- Extract structured data consistently\n",
+    "- Maintain high accuracy and reliability\n",
+    "- Handle documents at scale\n",
+    "\n",
+    "The combination of these three services provides a powerful, flexible document processing pipeline that can handle complex, real-world document processing requirements."
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}
@@ -7,7 +7,7 @@
   "source": [
    "# Dynamic Section Retrieval with LlamaParse\n",
    "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services-demo/blob/main/examples/parse/advanced_rag/dynamic_section_retrieval.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/advanced_rag/dynamic_section_retrieval.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
    "\n",
    "This notebook showcases a concept called \"dynamic section retrieval\".\n",
    "\n",
@@ -6,7 +6,7 @@
   "source": [
    "# Advanced RAG with LlamaParse\n",
    "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_parse/blob/main/examples/demo_advanced.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_parse/blob/main/examples/parse/demo_advanced.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
    "\n",
    "This notebook is a complete walkthrough for using LlamaParse with advanced indexing/retrieval techniques in LlamaIndex over the Apple 10K Filing. \n",
    "\n",
@@ -6,7 +6,7 @@
   "source": [
    "# RAG with Excel Spreadsheet using LlamaPrase\n",
    "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/demo_excel.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/demo_excel.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
    "\n",
    "This notebook shows you using LlamaParse with Excel Spreadsheet.\n",
    "\n",
@@ -7,7 +7,7 @@
   "source": [
    "# Download Charts\n",
    "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/demo_get_charts.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/demo_get_charts.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
    "\n",
    "This notebook demonstrates how to download charts from a document using the result object.\n",
    "\n",
@@ -6,7 +6,7 @@
   "source": [
    "# LlamaParse - Fast checking Insurance Contract for Coverage\n",
    "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/demo_insurance.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/demo_insurance.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
    "\n",
    "In this notebook we will look at how LlamaParse can be used to extract structured coverage information from an insurance policy.\n",
    "\n",
@@ -36,7 +36,7 @@
   "cell_type": "markdown",
   "metadata": {},
   "source": [
-    "## Download an insurance policy fron IRDAI\n",
+    "## Download an insurance policy from IRDAI\n",
    "\n",
    "The Insurance Regulatory and Development Authority of India (IRDAI) maintains a great resource: https://policyholder.gov.in/web/guest/non-life-insurance-products where all insurance policies available in India are publicly available for download! Let's download a complex health insurance policy as an example."
   ]
@@ -228,11 +228,11 @@
    "    result_type=\"markdown\",\n",
    "    system_prompt_append=\"\"\"\n",
    "This document is an insurance policy.\n",
-    "When a benefits/coverage/exlusion is describe in the document ammend to it add a text in the follwing benefits string format (where coverage could be an exclusion).\n",
+    "When a benefits/coverage/exlusion is describe in the document amend to it add a text in the following benefits string format (where coverage could be an exclusion).\n",
    "\n",
    "For {nameofrisk} and in this condition {whenDoesThecoverageApply} the coverage is {coverageDescription}. \n",
    "                                        \n",
-    "If the document contain a benefits TABLE that describe coverage amounts, do not ouput it as a table, but instead as a list of benefits string.\n",
+    "If the document contain a benefits TABLE that describe coverage amounts, do not output it as a table, but instead as a list of benefits string.\n",
    "                                       \n",
    "\"\"\",\n",
    ").aparse(\"./policy.pdf\")\n",
@@ -7,7 +7,7 @@
   "source": [
    "# LlamaParse `JobResult` Tour\n",
    "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/demo_json.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/demo_json.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
    "\n",
    "The `JobResult` object is the main object returned by the LlamaParse API. It contains all the information about the job, including the parsed data, metadata, and any errors.\n",
    "\n",
@@ -9,7 +9,7 @@
    "\n",
    "LlamaParse supports users to specify a `language` parameter before uploading documents, giving users better OCR capabilities over non-English PDFs, parsing images into more accurate representations.\n",
    "\n",
-    "You can specify 80+ different languages: see this file for a full list of supported languages: https://github.com/run-llama/llama_cloud_services/blob/main/llama_parse/base.py.\n",
+    "You can specify 80+ different languages: see this file for a full list of supported languages: https://github.com/run-llama/llama_cloud_services/blob/main/py/llama_cloud_services/parse/base.py.\n",
    "\n",
    "This notebook shows a demo of this in action. \n",
    "\n",
@@ -4,7 +4,7 @@
   "cell_type": "markdown",
   "metadata": {},
   "source": [
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/excel/o1_excel_rag.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/excel/o1_excel_rag.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
   ]
  },
  {
@@ -740,7 +740,7 @@
   "cell_type": "markdown",
   "metadata": {},
   "source": [
-    "In this example, these pages aren't going to be that different when parsed, but we can verify which pages triggered auto-made by looking at the [JSON output](https://github.com/run-llama/llama_cloud_services/blob/main/examples/demo_json_tour.ipynb) of LlamaParse:"
+    "In this example, these pages aren't going to be that different when parsed, but we can verify which pages triggered auto-made by looking at the [JSON output](https://github.com/run-llama/llama_cloud_services/blob/main/examples/parse/demo_json_tour.ipynb) of LlamaParse:"
   ]
  },
  {
@@ -5,9 +5,16 @@
  "private": true,
  "keywords": [],
  "author": "",
+  "scripts": {
+    "pre-commit-version": "pnpm changeset",
+    "version": "./scripts/changeset-version.py version",
+    "publish": "./scripts/changeset-version.py publish --tag"
+  },
  "devDependencies": {
    "prettier": "^3.6.2",
-    "lint-staged": "^15.4.2"
+    "lint-staged": "^15.4.2",
+    "@changesets/cli": "^2.29.5",
+    "changesets": "^1.0.2"
  },
  "lint-staged": {
    "ts/llama_cloud_services/src/**/*.{ts,tsx,js,jsx}": [
@@ -147,7 +147,7 @@ documents = SimpleDirectoryReader(
 ).load_data()
 ```

-Full documentation for `SimpleDirectoryReader` can be found on the [LlamaIndex Documentation](https://docs.llamaindex.ai/en/stable/module_guides/loading/simpledirectoryreader.html).
+Full documentation for `SimpleDirectoryReader` can be found on the [LlamaIndex Documentation](https://developers.llamaindex.ai/python/framework/module_guides/loading/simpledirectoryreader/).

 ## Examples

@@ -1,2 +1,4 @@
 packages:
-  - "ts/**"
+  - "ts/*"
+  - "py"
+  - "py/*"
@@ -0,0 +1,44 @@
+# llama-cloud-services-py
+
+## 0.6.76
+
+### Patch Changes
+
+- 4f24f53: Add aggressive_table_extraction flag in python sdk
+
+## 0.6.75
+
+### Patch Changes
+
+- f81532e: Safest types possible for parse
+
+## 0.6.74
+
+### Patch Changes
+
+- 1bf5223: Fix default bbox values
+- 24166dc: Now only escape single dollar signs - preserve double for latex equations
+
+## 0.6.73
+
+### Patch Changes
+
+- e6a7939: Loosen packaging dep requirement
+
+## 0.6.72
+
+### Patch Changes
+
+- ad6734b: Fixup and test versioning
+
+## 0.6.71
+
+### Patch Changes
+
+- 51011b9: Escape dollar signs in jupyter notebooks
+
+## 0.6.70
+
+### Patch Changes
+
+- d028397: Update llama-cloud api version, and integrate with agent data deletion
@@ -1,6 +1,11 @@
 import os
 from typing import Any, Dict, Generic, List, Optional, Type

+from llama_cloud import (
+    AgentData,
+    PaginatedResponseAgentData,
+    PaginatedResponseAggregateGroup,
+)
 from llama_cloud.client import AsyncLlamaCloud
 from tenacity import (
    WrappedFn,
@@ -86,7 +91,7 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
            client=llama_client,
            type=ExtractedPerson,
            collection="extracted_people",
-            agent_url_id="person-extraction-agent"
+            deployment_name="person-extraction-agent"
        )

        # Create data
@@ -109,10 +114,12 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
        self,
        type: Type[AgentDataT],
        collection: str = "default",
-        agent_url_id: Optional[str] = None,
+        deployment_name: Optional[str] = None,
        client: Optional[AsyncLlamaCloud] = None,
        token: Optional[str] = None,
        base_url: Optional[str] = None,
+        # deprecated, use deployment_name instead
+        agent_url_id: Optional[str] = None,
    ):
        """
        Initialize the AsyncAgentDataClient.
@@ -123,11 +130,11 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
            collection: Named collection within the agent for organizing data.
                Defaults to "default". Collections allow logical separation of
                different data types or workflows within the same agent.
-            agent_url_id: Unique identifier for the agent. This normally appears in the
-                url of an agent within the llama cloud platform. If not provided,
-                will attempt to use the LLAMA_DEPLOY_DEPLOYMENT_NAME environment
-                variable. Data can only be added to an already existing agent in the
-                platform.
+            deployment_name: Unique identifier for the agent deployment. This normally
+                appears in the URL of an agent within the Llama Cloud platform. If not
+                provided, will attempt to use the LLAMA_DEPLOY_DEPLOYMENT_NAME
+                environment variable. Data can only be added to an already existing
+                agent in the platform.
            client: AsyncLlamaCloud client instance for API communication. If not provided, will
                construct one from the provided api token and base url
            token: Llama Cloud API token. Reads from LLAMA_CLOUD_API_KEY if not provided
@@ -135,15 +142,14 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
                defaults to https://api.cloud.llamaindex.ai

        Raises:
-            ValueError: If agent_url_id is not provided and the
+            ValueError: If deployment_name is not provided and the
                LLAMA_DEPLOY_DEPLOYMENT_NAME environment variable is not set

        Note:
            The client automatically applies retry logic to all API calls with
            exponential backoff for timeout, connection, and HTTP status errors.
        """
-
-        self.agent_url_id = agent_url_id or get_default_agent_id()
+        self.deployment_name = deployment_name or agent_url_id or get_default_agent_id()

        self.collection = collection
        if not client:
@@ -156,15 +162,19 @@ class AsyncAgentDataClient(Generic[AgentDataT]):

    @agent_data_retry
    async def get_item(self, item_id: str) -> TypedAgentData[AgentDataT]:
-        raw_data = await self.client.beta.get_agent_data(
+        raw_data = await self.untyped_get_item(item_id)
+        return TypedAgentData.from_raw(raw_data, self.type)
+
+    @agent_data_retry
+    async def untyped_get_item(self, item_id: str) -> AgentData:
+        return await self.client.beta.get_agent_data(
            item_id=item_id,
        )
-        return TypedAgentData.from_raw(raw_data, validator=self.type)

    @agent_data_retry
    async def create_item(self, data: AgentDataT) -> TypedAgentData[AgentDataT]:
        raw_data = await self.client.beta.create_agent_data(
-            agent_slug=self.agent_url_id,
+            deployment_name=self.deployment_name,
            collection=self.collection,
            data=data.model_dump(),
        )
@@ -184,6 +194,21 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
    async def delete_item(self, item_id: str) -> None:
        await self.client.beta.delete_agent_data(item_id=item_id)

+    @agent_data_retry
+    async def delete(
+        self, filter: Optional[Dict[str, Dict[ComparisonOperator, Any]]] = None
+    ) -> int:
+        """
+        Delete agent data by query, similar to search.
+        Returns the number of deleted items.
+        """
+        response = await self.client.beta.delete_agent_data_by_query_api_v_1_beta_agent_data_delete_post(
+            deployment_name=self.deployment_name,
+            collection=self.collection,
+            filter=filter,
+        )
+        return response.deleted_count
+
    @agent_data_retry
    async def search(
        self,
@@ -210,9 +235,7 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
            offset: Number of items to skip from the beginning. Defaults to 0.
            include_total: Whether to include the total count in the response. Defaults to False to improve performance. It's recommended to only request on the first page.
        """
-        raw = await self.client.beta.search_agent_data_api_v_1_beta_agent_data_search_post(
-            agent_slug=self.agent_url_id,
-            collection=self.collection,
+        raw = await self.untyped_search(
            filter=filter,
            order_by=order_by,
            offset=offset,
@@ -227,6 +250,25 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
            total=raw.total_size,
        )

+    @agent_data_retry
+    async def untyped_search(
+        self,
+        filter: Optional[Dict[str, Dict[ComparisonOperator, Any]]] = None,
+        order_by: Optional[str] = None,
+        offset: Optional[int] = None,
+        page_size: Optional[int] = None,
+        include_total: bool = False,
+    ) -> PaginatedResponseAgentData:
+        return await self.client.beta.search_agent_data_api_v_1_beta_agent_data_search_post(
+            deployment_name=self.deployment_name,
+            collection=self.collection,
+            filter=filter,
+            order_by=order_by,
+            offset=offset,
+            page_size=page_size,
+            include_total=include_total,
+        )
+
    @agent_data_retry
    async def aggregate(
        self,
@@ -253,8 +295,38 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
            offset: Number of groups to skip from the beginning. Defaults to 0.
            page_size: Maximum number of groups to return per page.
        """
-        raw = await self.client.beta.aggregate_agent_data_api_v_1_beta_agent_data_aggregate_post(
-            agent_slug=self.agent_url_id,
+        raw = await self.untyped_aggregate(
+            filter=filter,
+            group_by=group_by,
+            count=count,
+            first=first,
+            order_by=order_by,
+            offset=offset,
+            page_size=page_size,
+        )
+
+        return TypedAggregateGroupItems(
+            items=[
+                TypedAggregateGroup.from_raw(grp, validator=self.type)
+                for grp in raw.items
+            ],
+            has_more=raw.next_page_token is not None,
+            total=raw.total_size,
+        )
+
+    @agent_data_retry
+    async def untyped_aggregate(
+        self,
+        filter: Optional[Dict[str, Dict[ComparisonOperator, Any]]] = None,
+        group_by: Optional[List[str]] = None,
+        count: Optional[bool] = None,
+        first: Optional[bool] = None,
+        order_by: Optional[str] = None,
+        offset: Optional[int] = None,
+        page_size: Optional[int] = None,
+    ) -> PaginatedResponseAggregateGroup:
+        return await self.client.beta.aggregate_agent_data_api_v_1_beta_agent_data_aggregate_post(
+            deployment_name=self.deployment_name,
            collection=self.collection,
            page_size=page_size,
            filter=filter,
@@ -264,11 +336,3 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
            first=first,
            offset=offset,
        )
-        return TypedAggregateGroupItems(
-            items=[
-                TypedAggregateGroup.from_raw(item, validator=self.type)
-                for item in raw.items
-            ],
-            has_more=raw.next_page_token is not None,
-            total=raw.total_size,
-        )
@@ -10,7 +10,7 @@ CRUD operations, search capabilities, filtering, and aggregation functionality
 for managing agent-generated data at scale.

 Key Concepts:
- Agent Slug: Unique identifier for an agent instance
+- Deployment Name: Unique identifier for an agent deployment
 - Collection: Named grouping of data within an agent (defaults to "default"). Data within a collection should be of the same type.
 - Agent Data: Individual structured data records with metadata and timestamps

@@ -26,7 +26,7 @@ Example Usage:
        client=async_llama_cloud,
        type=Person,
        collection="people",
-        agent_url_id="my-extraction-agent-xyz"
+        deployment_name="my-extraction-agent-xyz"
    )

    # Create typed data
@@ -56,7 +56,6 @@ from typing import (

 # Type variable for user-defined data models
 AgentDataT = TypeVar("AgentDataT", bound=BaseModel)
-
 # Type variable for extracted data (can be dict or Pydantic model)
 ExtractedT = TypeVar("ExtractedT", bound=Union[BaseModel, dict])

@@ -78,7 +77,7 @@ class TypedAgentData(BaseModel, Generic[AgentDataT]):

    Attributes:
        id: Unique identifier for this data record
-        agent_url_id: Identifier of the agent that created this data
+        deployment_name: Identifier of the agent deployment that created this data
        collection: Named collection within the agent (used for organization)
        data: The actual structured data payload (typed as AgentDataT)
        created_at: Timestamp when the record was first created
@@ -94,8 +93,8 @@ class TypedAgentData(BaseModel, Generic[AgentDataT]):
    """

    id: Optional[str] = Field(description="Unique identifier for this data record")
-    agent_url_id: str = Field(
-        description="Identifier of the agent that created this data"
+    deployment_name: str = Field(
+        description="Identifier of the agent deployment that created this data"
    )
    collection: Optional[str] = Field(
        description="Named collection within the agent for data organization"
@@ -116,15 +115,15 @@ class TypedAgentData(BaseModel, Generic[AgentDataT]):
        Args:
            raw_data: Raw agent data from the API
            validator: Pydantic model class to validate the data field
-
        Returns:
            TypedAgentData instance with validated data
        """
+
        data: AgentDataT = validator.model_validate(raw_data.data)

        return cls(
            id=raw_data.id,
-            agent_url_id=raw_data.agent_slug,
+            deployment_name=raw_data.deployment_name,
            collection=raw_data.collection,
            data=data,
            created_at=raw_data.created_at,
@@ -222,12 +221,16 @@ def parse_extracted_field_metadata(
    return {
        k: _parse_extracted_field_metadata_recursive(v)
        for k, v in field_metadata.items()
-        if k not in _METADATA_FIELDS_SIBLING_TO_LEAF
-        and k not in _ADDITIONAL_ROOT_METADATA_FIELDS
+        if not _is_reasoning_field(k, v) and k not in _ADDITIONAL_ROOT_METADATA_FIELDS
    }


-_METADATA_FIELDS_SIBLING_TO_LEAF = {"reasoning"}
+def _is_reasoning_field(field_name: str, field_value: Any) -> bool:
+    # There can either be a user specified reasoning field (from the schema), or a reasoning metadata field for the
+    # dict of values
+    return field_name == "reasoning" and isinstance(field_value, str)
+
+
 _ADDITIONAL_ROOT_METADATA_FIELDS = {"error"}


@@ -257,14 +260,12 @@ def _parse_extracted_field_metadata_recursive(
            except ValidationError:
                pass
        additional_fields = {
-            k: v
-            for k, v in field_value.items()
-            if k in _METADATA_FIELDS_SIBLING_TO_LEAF
+            k: v for k, v in field_value.items() if _is_reasoning_field(k, v)
        }
        return {
            k: _parse_extracted_field_metadata_recursive(v, additional_fields)
            for k, v in field_value.items()
-            if k not in _METADATA_FIELDS_SIBLING_TO_LEAF
+            if not _is_reasoning_field(k, v)
        }
    elif isinstance(field_value, list):
        return [_parse_extracted_field_metadata_recursive(item) for item in field_value]
@@ -19,14 +19,12 @@ from llama_cloud import (
    ExtractAgent as CloudExtractAgent,
    ExtractConfig,
    ExtractJob,
-    ExtractJobCreate,
    ExtractRun,
    File,
    FileData,
    ExtractMode,
    StatusEnum,
    ExtractTarget,
-    LlamaExtractSettings,
    PaginatedExtractRunsResponse,
 )
 from llama_cloud.client import AsyncLlamaCloud
@@ -463,56 +461,6 @@ class ExtractionAgent:
            )
        )

-    async def _run_extraction_test(
-        self,
-        files: Union[FileInput, List[FileInput]],
-        extract_settings: LlamaExtractSettings,
-    ) -> Union[ExtractJob, List[ExtractJob]]:
-        if not isinstance(files, list):
-            files = [files]
-            single_file = True
-        else:
-            single_file = False
-
-        upload_tasks = [self._upload_file(file) for file in files]
-        with augment_async_errors():
-            uploaded_files = await run_jobs(
-                upload_tasks,
-                workers=self.num_workers,
-                desc="Uploading files",
-                show_progress=self.show_progress,
-            )
-
-        async def run_job(file: File) -> ExtractRun:
-            job_queued = await self._client.llama_extract.run_job_test_user(
-                job_create=ExtractJobCreate(
-                    extraction_agent_id=self.id,
-                    file_id=file.id,
-                    data_schema_override=self.data_schema,
-                    config_override=self.config,
-                ),
-                extract_settings=extract_settings,
-            )
-            return await self._wait_for_job_result(job_queued.id)
-
-        job_tasks = [run_job(file) for file in uploaded_files]
-        with augment_async_errors():
-            extract_results = await run_jobs(
-                job_tasks,
-                workers=self.num_workers,
-                desc="Running extraction jobs",
-                show_progress=self.show_progress,
-            )
-
-        if self._verbose:
-            for file, job in zip(files, extract_results):
-                file_repr = (
-                    str(file) if isinstance(file, (str, Path)) else "<bytes/buffer>"
-                )
-                print(f"Running extraction for file {file_repr} under job_id {job.id}")
-
-        return extract_results[0] if single_file else extract_results
-
    async def queue_extraction(
        self,
        files: Union[FileInput, List[FileInput]],
@@ -544,12 +492,10 @@ class ExtractionAgent:

        job_tasks = [
            self._client.llama_extract.run_job(
-                request=ExtractJobCreate(
-                    extraction_agent_id=self.id,
-                    file_id=file.id,
-                    data_schema_override=self.data_schema,
-                    config_override=self.config,
-                ),
+                extraction_agent_id=self.id,
+                file_id=file.id,
+                data_schema_override=self.data_schema,
+                config_override=self.config,
            )
            for file in uploaded_files
        ]
@@ -489,6 +489,7 @@ class LlamaCloudIndex(BaseManagedIndex):
        name: str,
        project_name: str = DEFAULT_PROJECT_NAME,
        organization_id: Optional[str] = None,
+        project_id: Optional[str] = None,
        api_key: Optional[str] = None,
        base_url: Optional[str] = None,
        app_url: Optional[str] = None,
@@ -504,15 +505,15 @@ class LlamaCloudIndex(BaseManagedIndex):
        app_url = app_url or os.environ.get("LLAMA_CLOUD_APP_URL", DEFAULT_APP_URL)
        client = get_client(api_key, base_url, app_url, timeout)

-        # create project if it doesn't exist
-        project = client.projects.upsert_project(
-            organization_id=organization_id, request=ProjectCreate(name=project_name)
-        )
-        if project.id is None:
-            raise ValueError(f"Failed to create/get project {project_name}")
-
-        if verbose:
-            print(f"Created project {project.id} with name {project.name}")
+        if project_id is None:
+            # create project if it doesn't exist
+            project = client.projects.upsert_project(
+                organization_id=organization_id,
+                request=ProjectCreate(name=project_name),
+            )
+            project_id = project.id
+            if verbose:
+                print(f"Created project {project_id} with name {project_name}")

        # create pipeline
        pipeline_create = PipelineCreate(
@@ -523,7 +524,7 @@ class LlamaCloudIndex(BaseManagedIndex):
            llama_parse_parameters=llama_parse_parameters or LlamaParseParameters(),
        )
        pipeline = client.pipelines.upsert_pipeline(
-            project_id=project.id, request=pipeline_create
+            project_id=project_id, request=pipeline_create
        )
        if pipeline.id is None:
            raise ValueError(f"Failed to create/get pipeline {name}")
@@ -532,8 +533,7 @@ class LlamaCloudIndex(BaseManagedIndex):

        return cls(
            name,
-            project_name=project.name,
-            organization_id=project.organization_id,
+            project_id=project_id,
            api_key=api_key,
            base_url=base_url,
            app_url=app_url,
@@ -606,6 +606,7 @@ class LlamaCloudIndex(BaseManagedIndex):
        name: str,
        project_name: str = DEFAULT_PROJECT_NAME,
        organization_id: Optional[str] = None,
+        project_id: Optional[str] = None,
        api_key: Optional[str] = None,
        base_url: Optional[str] = None,
        app_url: Optional[str] = None,
@@ -631,6 +632,7 @@ class LlamaCloudIndex(BaseManagedIndex):
            verbose=verbose,
            embedding_config=embedding_config,
            transform_config=transform_config,
+            project_id=project_id,
        )

        app_url = app_url or os.environ.get("LLAMA_CLOUD_APP_URL", DEFAULT_APP_URL)
@@ -188,6 +188,10 @@ class LlamaParse(BasePydanticReader):
        default=False,
        description="If set to true, LlamaParse will try to detect long table and adapt the output.",
    )
+    aggressive_table_extraction: Optional[bool] = Field(
+        default=False,
+        description="If set to true, LlamaParse will try to extract tables aggressively, may lead to false positives.",
+    )
    annotate_links: Optional[bool] = Field(
        default=False,
        description="Annotate links found in the document to extract their URL.",
@@ -420,6 +424,10 @@ class LlamaParse(BasePydanticReader):
        default=False,
        description="If set to true, the parser will extract sub-tables from the spreadsheet when possible (more than one table per sheet).",
    )
+    spreadsheet_force_formula_computation: Optional[bool] = Field(
+        default=False,
+        description="If set to true, the parser will re-compute values for all spreadsheet cells containing formulas.",
+    )
    specialized_chart_parsing_agentic: Optional[bool] = Field(
        default=False,
        description="If set to true, the parser will use a specialized agentic chart parsing model to extract data from charts. This model is able to understand the chart type and extract the data accordingly.",
@@ -709,6 +717,9 @@ class LlamaParse(BasePydanticReader):
        if self.adaptive_long_table:
            data["adaptive_long_table"] = self.adaptive_long_table

+        if self.aggressive_table_extraction:
+            data["aggressive_table_extraction"] = self.aggressive_table_extraction
+
        if self.annotate_links:
            data["annotate_links"] = self.annotate_links

@@ -965,6 +976,11 @@ class LlamaParse(BasePydanticReader):
        if self.spreadsheet_extract_sub_tables:
            data["spreadsheet_extract_sub_tables"] = self.spreadsheet_extract_sub_tables

+        if self.spreadsheet_force_formula_computation:
+            data[
+                "spreadsheet_force_formula_computation"
+            ] = self.spreadsheet_force_formula_computation
+
        if self.specialized_chart_parsing_agentic:
            data[
                "specialized_chart_parsing_agentic"
@@ -1,17 +1,87 @@
 import httpx
 import os
 import re
-from pydantic import BaseModel, Field, SerializeAsAny
-from typing import Dict, Any, List, Optional
+from pydantic import BaseModel, ConfigDict, Field, SerializeAsAny, model_validator
+from typing import Dict, Any, List, Optional, get_origin, get_args

-from llama_cloud_services.parse.utils import make_api_request
+from llama_cloud_services.parse.utils import (
+    make_api_request,
+    is_jupyter,
+)
 from llama_index.core.async_utils import asyncio_run
 from llama_index.core.schema import Document, ImageDocument, ImageNode, TextNode

 PAGE_REGEX = r"page[-_](\d+)\.jpg$"

+SAFE_MODEL_CONFIGS = ConfigDict(
+    extra="allow",
+    validate_assignment=False,
+    arbitrary_types_allowed=True,
+    validate_default=False,
+)

-class JobMetadata(BaseModel):
+
+class SafeBaseModel(BaseModel):
+    """Base model that gracefully handles None values from unstable backend responses."""
+
+    model_config = SAFE_MODEL_CONFIGS
+
+    @model_validator(mode="before")
+    @classmethod
+    def coerce_none_to_defaults(cls, data: Any) -> Any:
+        """
+        Replace None values with appropriate defaults based on field type annotations.
+        This prevents validation errors when the backend returns None for non-optional fields.
+        """
+        if not isinstance(data, dict):
+            return data
+
+        # Process each field that has a None value
+        result = {}
+        for key, value in data.items():
+            if value is not None or key not in cls.model_fields:
+                result[key] = value
+                continue
+
+            # Value is None and field exists in model
+            field_info = cls.model_fields[key]
+
+            # If field has a default or default_factory, let Pydantic handle it
+            from pydantic_core import PydanticUndefined
+
+            if (
+                field_info.default is not PydanticUndefined
+                or field_info.default_factory is not None
+            ):
+                continue
+
+            # Otherwise, provide a sensible default based on the type annotation
+            annotation = field_info.annotation
+            origin = get_origin(annotation)
+
+            # Handle List types
+            if origin is list:
+                result[key] = []
+            # Handle Dict types
+            elif origin is dict:
+                result[key] = {}
+            # Handle basic types
+            elif annotation == str or (origin and str in get_args(annotation)):
+                result[key] = ""
+            elif annotation == int or (origin and int in get_args(annotation)):
+                result[key] = 0
+            elif annotation == float or (origin and float in get_args(annotation)):
+                result[key] = 0.0
+            elif annotation == bool or (origin and bool in get_args(annotation)):
+                result[key] = False
+            # If we can't determine a safe default, skip (let Pydantic try)
+            else:
+                result[key] = value
+
+        return result
+
+
+class JobMetadata(SafeBaseModel):
    """Metadata about the job."""

    job_pages: int = Field(default=0, description="The number of pages in the job.")
@@ -24,19 +94,31 @@ class JobMetadata(BaseModel):
    )


-class BBox(BaseModel):
+class BBox(SafeBaseModel):
    """A bounding box."""

-    x: float = Field(description="The x-coordinate of the bounding box.")
-    y: float = Field(description="The y-coordinate of the bounding box.")
-    w: float = Field(description="The width of the bounding box.")
-    h: float = Field(description="The height of the bounding box.")
+    x: Optional[float] = Field(
+        default=None,
+        description="The x-coordinate of the bounding box.",
+    )
+    y: Optional[float] = Field(
+        default=None,
+        description="The y-coordinate of the bounding box.",
+    )
+    w: Optional[float] = Field(
+        default=None,
+        description="The width of the bounding box.",
+    )
+    h: Optional[float] = Field(
+        default=None,
+        description="The height of the bounding box.",
+    )


-class PageItem(BaseModel):
+class PageItem(SafeBaseModel):
    """An item in a page."""

-    type: str = Field(description="The type of the item.")
+    type: str = Field(default="", description="The type of the item.")
    lvl: Optional[int] = Field(
        default=None, description="The level of indentation of the item."
    )
@@ -58,10 +140,10 @@ class PageItem(BaseModel):
    )


-class ImageItem(BaseModel):
+class ImageItem(SafeBaseModel):
    """An image in a page."""

-    name: str = Field(description="The name of the image.")
+    name: str = Field(default="", description="The name of the image.")
    height: Optional[float] = Field(
        default=None, description="The height of the image."
    )
@@ -81,22 +163,28 @@ class ImageItem(BaseModel):
    type: Optional[str] = Field(default=None, description="The type of the image.")


-class LayoutItem(BaseModel):
+class LayoutItem(SafeBaseModel):
    """The layout of a page."""

-    image: str = Field(description="The name of the image containing the layout item")
-    confidence: float = Field(description="The confidence of the layout item.")
-    label: str = Field(description="The label of the layout item.")
+    image: str = Field(
+        default="", description="The name of the image containing the layout item"
+    )
+    confidence: float = Field(
+        default=0.0, description="The confidence of the layout item."
+    )
+    label: str = Field(default="", description="The label of the layout item.")
    bbox: Optional[BBox] = Field(
        default=None, description="The bounding box of the layout item."
    )
-    isLikelyNoise: bool = Field(description="Whether the layout item is likely noise.")
+    isLikelyNoise: bool = Field(
+        default=False, description="Whether the layout item is likely noise."
+    )


-class ChartItem(BaseModel):
+class ChartItem(SafeBaseModel):
    """A chart in a page."""

-    name: str = Field(description="The name of the chart.")
+    name: str = Field(default="", description="The name of the chart.")
    x: Optional[float] = Field(
        default=None, description="The x-coordinate of the chart."
    )
@@ -109,7 +197,7 @@ class ChartItem(BaseModel):
    )


-class Page(BaseModel):
+class Page(SafeBaseModel):
    """A page of the document."""

    page: int = Field(default=0, description="The page number.")
@@ -164,7 +252,7 @@ class Page(BaseModel):
    )


-class JobResult(BaseModel):
+class JobResult(SafeBaseModel):
    """The raw JSON result from the LlamaParse API."""

    pages: List[Page] = Field(
@@ -258,6 +346,29 @@ class JobResult(BaseModel):
        documents = await self.aget_text_documents(split_by_page)
        return [TextNode(text=doc.text, metadata=doc.metadata) for doc in documents]

+    def _format_markdown_for_notebook(self, text: Optional[str]) -> Optional[str]:
+        """Format markdown text for Jupyter notebook display by escaping dollar signs."""
+        if text is None:
+            return None
+
+        def escape_single_dollar_signs(text: str) -> str:
+            """Escape single dollar signs in text to prevent Jupyter from interpreting them as LaTeX.
+
+            Preserves all strings of dollar signs greater than length 1,
+            especially preserving double dollar signs ($$) which denote LaTeX equations.
+
+            Args:
+                text: The text to escape
+
+            Returns:
+                Text with single dollar signs escaped
+            """
+            # Replace single $ with \$, but preserve $$
+            # Use negative lookahead and lookbehind to match $ not preceded or followed by $
+            return re.sub(r"(?<!\$)\$(?!\$)", r"\$", text)
+
+        return escape_single_dollar_signs(text)
+
    def get_markdown_documents(self, split_by_page: bool = False) -> List[Document]:
        """
        Get the markdown documents from the job.
@@ -268,17 +379,22 @@ class JobResult(BaseModel):
        if split_by_page:
            return [
                Document(
-                    text=page.md,
+                    text=self._format_markdown_for_notebook(page.md)
+                    if is_jupyter()
+                    else page.md,
                    metadata={"page_number": page.page, "file_name": self.file_name},
                )
                for page in self.pages
            ]
        else:
+            text = self._page_separator.join(
+                [page.md if page.md is not None else "" for page in self.pages]
+            )
            return [
                Document(
-                    text=self._page_separator.join(
-                        [page.md if page.md is not None else "" for page in self.pages]
-                    ),
+                    text=self._format_markdown_for_notebook(text)
+                    if is_jupyter()
+                    else text,
                    metadata={"file_name": self.file_name},
                )
            ]
@@ -328,7 +444,10 @@ class JobResult(BaseModel):
        """
        url = f"{self._base_url}/api/v1/parsing/job/{self.job_id}/result/raw/markdown"
        response = await make_api_request(self._client, "GET", url)
-        return response.content.decode("utf-8")
+        markdown = response.content.decode("utf-8")
+        return (
+            self._format_markdown_for_notebook(markdown) if is_jupyter() else markdown
+        )

    def get_text(self) -> str:
        """
@@ -1,3 +1,4 @@
+import functools
 import httpx
 import itertools
 import logging
@@ -356,6 +357,17 @@ def partition_pages(
            return


+@functools.lru_cache(maxsize=1)
+def is_jupyter() -> bool:
+    """Check if we're running in a Jupyter environment."""
+    try:
+        from IPython import get_ipython
+
+        return get_ipython().__class__.__name__ == "ZMQInteractiveShell"
+    except (ImportError, AttributeError):
+        return False
+
+
 def extract_tables_from_json_results(
    json_results: List[dict], download_path: str
 ) -> List[str]:
@@ -0,0 +1,37 @@
+# llama_parse
+
+## 0.6.76
+
+### Patch Changes
+
+- Updated dependencies [4f24f53]
+  - llama-cloud-services-py@0.6.76
+
+## 0.6.75
+
+### Patch Changes
+
+- Updated dependencies [f81532e]
+  - llama-cloud-services-py@0.6.75
+
+## 0.6.74
+
+### Patch Changes
+
+- Updated dependencies [1bf5223]
+- Updated dependencies [24166dc]
+  - llama-cloud-services-py@0.6.74
+
+## 0.6.73
+
+### Patch Changes
+
+- Updated dependencies [e6a7939]
+  - llama-cloud-services-py@0.6.73
+
+## 0.6.72
+
+### Patch Changes
+
+- Updated dependencies [ad6734b]
+  - llama-cloud-services-py@0.6.72
@@ -0,0 +1,20 @@
+{
+  "name": "llama_parse",
+  "version": "0.6.76",
+  "description": "",
+  "main": "index.js",
+  "private": false,
+  "scripts": {
+    "test": "echo \"Error: no test specified\" && exit 1"
+  },
+  "dependencies": {
+    "llama-cloud-services-py": "workspace:*"
+  },
+  "keywords": [],
+  "author": "",
+  "license": "ISC",
+  "packageManager": "pnpm@10.11.1",
+  "devDependencies": {
+    "changesets": "^1.0.2"
+  }
+}
@@ -11,13 +11,13 @@ dev = [

 [project]
 name = "llama-parse"
-version = "0.6.65"
+version = "0.6.76"
 description = "Parse files into RAG-Optimized formats."
 authors = [{name = "Logan Markewich", email = "logan@llamaindex.ai"}]
 requires-python = ">=3.9,<4.0"
 readme = "README.md"
 license = "MIT"
-dependencies = ["llama-cloud-services>=0.6.64"]
+dependencies = ["llama-cloud-services>=0.6.76"]

 [project.scripts]
 llama-parse = "llama_parse.cli.main:parse"
@@ -0,0 +1,10 @@
+{
+  "name": "llama-cloud-services-py",
+  "version": "0.6.76",
+  "private": false,
+  "license": "MIT",
+  "scripts": {},
+  "devDependencies": {
+    "changesets": "^1.0.2"
+  }
+}
@@ -19,7 +19,7 @@ dev = [

 [project]
 name = "llama-cloud-services"
-version = "0.6.65"
+version = "0.6.76"
 description = "Tailored SDK clients for LlamaCloud services."
 authors = [{name = "Logan Markewich", email = "logan@runllama.ai"}]
 requires-python = ">=3.9,<4.0"
@@ -27,14 +27,14 @@ readme = "README.md"
 license = "MIT"
 dependencies = [
  "llama-index-core>=0.12.0",
-  "llama-cloud==0.1.41",
+  "llama-cloud==0.1.43",
  "pydantic>=2.8,!=2.10",
  "click>=8.1.7,<9",
  "python-dotenv>=1.0.1,<2",
  "eval-type-backport>=0.2.0,<0.3 ; python_version < '3.10'",
  "platformdirs>=4.3.7,<5",
  "tenacity>=8.5.0, <10.0",
-  "packaging>=25.0"
+  "packaging>=23.0"
 ]

 [project.scripts]
@@ -68,7 +68,7 @@ async def test_agent_data_crud_operations():
        client=client,
        type=ExampleData,
        collection=f"test-collection-{test_id[:8]}",
-        agent_url_id=LLAMA_DEPLOY_DEPLOYMENT_NAME,
+        deployment_name=LLAMA_DEPLOY_DEPLOYMENT_NAME,
    )

    # Create test data
@@ -1,16 +1,13 @@
 import os
 import pytest

-from llama_cloud_services.extract import LlamaExtract, ExtractionAgent
-from time import perf_counter
+from llama_cloud_services.extract import LlamaExtract
 from collections import namedtuple
 import json
 import uuid
 from llama_cloud.types import (
    ExtractConfig,
    ExtractMode,
-    LlamaParseParameters,
-    LlamaExtractSettings,
 )
 from tests.extract.util import load_test_dotenv

@@ -122,27 +119,3 @@ def extraction_agent(test_case: BenchmarkTestCase, extractor: LlamaExtract):
    # Create new agent
    agent = extractor.create_agent(agent_name, schema, config=test_case.config)
    yield agent
-
-
-@pytest.mark.skipif(
-    "CI" in os.environ or not LLAMA_CLOUD_API_KEY,
-    reason="LLAMA_CLOUD_API_KEY not set or CI environment not suitable for benchmarking",
-)
-@pytest.mark.parametrize("test_case", get_test_cases(), ids=lambda x: x.name)
-@pytest.mark.asyncio(loop_scope="session")
-async def test_extraction(
-    test_case: BenchmarkTestCase, extraction_agent: ExtractionAgent
-) -> None:
-    start = perf_counter()
-    result = await extraction_agent._run_extraction_test(
-        test_case.input_file,
-        extract_settings=LlamaExtractSettings(
-            llama_parse_params=LlamaParseParameters(
-                invalidate_cache=True,
-                do_not_cache=True,
-            )
-        ),
-    )
-    end = perf_counter()
-    print(f"Time taken: {end - start} seconds")
-    print(result)
@@ -7,7 +7,7 @@ from pathlib import Path


 def load_test_dotenv():
-    load_dotenv(Path(__file__).parent.parent.parent / ".env.dev", override=True)
+    load_dotenv(Path(__file__).parent.parent.parent.parent / ".env.dev", override=True)


 def json_subset_match_score(expected: Any, actual: Any) -> float:
@@ -304,6 +304,9 @@ async def test_page_screenshot_retrieval(index_name: str, local_file: str):
    not base_url or not api_key, reason="No platform base url or api key set"
 )
@pytest.mark.asyncio
+@pytest.mark.skip(
+    reason="Consistently failing with FAILED tests/index/test_index.py::test_page_figure_retrieval - assert 0 > 0 +  where 0 = len([])"
+)
 async def test_page_figure_retrieval(index_name: str, local_figures_file: str):
    index = await LlamaCloudIndex.acreate_index(
        name=index_name,
@@ -6,6 +6,40 @@ from llama_cloud_services import LlamaParse
 from llama_cloud_services.parse.types import JobResult


+def test_format_parse_result_markdown_for_notebook():
+    """Test the _format_markdown_for_notebook function.
+    Right now, the only work it does is escape single dollar signs."""
+    result = JobResult(job_id="test", file_name="test.pdf", job_result={})
+
+    # Test None input
+    assert result._format_markdown_for_notebook(None) is None
+
+    # Test single dollar sign gets escaped
+    assert result._format_markdown_for_notebook("This costs $5") == "This costs \\$5"
+
+    # Test double dollar signs are preserved (LaTeX equations)
+    assert (
+        result._format_markdown_for_notebook("$$x^2 + y^2 = z^2$$")
+        == "$$x^2 + y^2 = z^2$$"
+    )
+
+    # Test mixed single and double dollar signs
+    text = "This costs $5, but $$E = mc^2$$ is priceless"
+    expected = "This costs \\$5, but $$E = mc^2$$ is priceless"
+    assert result._format_markdown_for_notebook(text) == expected
+
+    # Test multiple single dollar signs
+    assert result._format_markdown_for_notebook("$10 and $20") == "\\$10 and \\$20"
+
+    # Test three or more consecutive dollar signs (preserve them)
+    assert result._format_markdown_for_notebook("$$$") == "$$$"
+
+    # Test adjacent dollar signs with text in between
+    text = "$$inline$$ and $separate"
+    expected = "$$inline$$ and \\$separate"
+    assert result._format_markdown_for_notebook(text) == expected
+
+
@pytest.fixture
 def file_path() -> str:
    return "tests/test_files/attention_is_all_you_need.pdf"
@@ -0,0 +1,158 @@
+import pytest
+from typing import Any, Dict, List, Optional
+from pydantic import BaseModel
+from datetime import datetime
+
+from llama_cloud.types.agent_data import AgentData
+from llama_cloud.types.aggregate_group import AggregateGroup
+
+from llama_cloud_services.beta.agent_data.client import AsyncAgentDataClient
+
+
+class Person(BaseModel):
+    name: str
+    age: int
+
+
+class FakeBeta:
+    def __init__(self) -> None:
+        self._get_item_response: Optional[AgentData] = None
+        self._search_items: List[AgentData] = []
+        self._aggregate_items: List[AggregateGroup] = []
+        self._total_size: Optional[int] = None
+        self._next_page_token: Optional[str] = None
+
+    # Single get
+    async def get_agent_data(self, item_id: str) -> AgentData:
+        assert self._get_item_response is not None, "_get_item_response not set"
+        return self._get_item_response
+
+    # Search
+    async def search_agent_data_api_v_1_beta_agent_data_search_post(
+        self,
+        *,
+        deployment_name: str,
+        collection: str,
+        filter: Optional[Dict[str, Any]] = None,
+        order_by: Optional[str] = None,
+        offset: Optional[int] = None,
+        page_size: Optional[int] = None,
+        include_total: bool = False,
+    ) -> Any:
+        class Resp:
+            def __init__(
+                self,
+                items: List[AgentData],
+                total_size: Optional[int],
+                next_page_token: Optional[str],
+            ) -> None:
+                self.items = items
+                self.total_size = total_size
+                self.next_page_token = next_page_token
+
+        return Resp(self._search_items, self._total_size, self._next_page_token)
+
+    # Aggregate
+    async def aggregate_agent_data_api_v_1_beta_agent_data_aggregate_post(
+        self,
+        *,
+        deployment_name: str,
+        collection: str,
+        page_size: Optional[int] = None,
+        filter: Optional[Dict[str, Any]] = None,
+        order_by: Optional[str] = None,
+        group_by: Optional[List[str]] = None,
+        count: Optional[bool] = None,
+        first: Optional[bool] = None,
+        offset: Optional[int] = None,
+    ) -> Any:
+        class Resp:
+            def __init__(
+                self,
+                items: List[AggregateGroup],
+                total_size: Optional[int],
+                next_page_token: Optional[str],
+            ) -> None:
+                self.items = items
+                self.total_size = total_size
+                self.next_page_token = next_page_token
+
+        return Resp(self._aggregate_items, self._total_size, self._next_page_token)
+
+
+class FakeClient:
+    def __init__(self) -> None:
+        self.beta = FakeBeta()
+
+
+def make_agent_data(data: Dict[str, Any]) -> AgentData:
+    return AgentData(
+        id="id-1",
+        deployment_name="dep",
+        collection="col",
+        data=data,
+        created_at=datetime.now(),
+        updated_at=datetime.now(),
+    )
+
+
+def make_group(
+    group_key: Dict[str, Any],
+    first_item: Optional[Dict[str, Any]],
+    count: Optional[int] = None,
+) -> AggregateGroup:
+    return AggregateGroup(group_key=group_key, count=count, first_item=first_item)
+
+
+@pytest.mark.asyncio
+async def test_untyped_get_item_valid_to_dict() -> None:
+    client = FakeClient()
+    client.beta._get_item_response = make_agent_data({"name": "Alice", "age": 30})
+
+    adc = AsyncAgentDataClient(type=Person, client=client, deployment_name="dep")
+    item = await adc.untyped_get_item("id-1")
+    assert item.data == {"name": "Alice", "age": 30}
+
+
+@pytest.mark.asyncio
+async def test_untyped_get_item_invalid_retains_dict() -> None:
+    client = FakeClient()
+    # age wrong type; will fail validation and should be returned as dict
+    client.beta._get_item_response = make_agent_data({"name": "Bob", "age": "x"})
+
+    adc = AsyncAgentDataClient(type=Person, client=client, deployment_name="dep")
+    item = await adc.untyped_get_item("id-1")
+    assert item.data == {"name": "Bob", "age": "x"}
+
+
+@pytest.mark.asyncio
+async def test_untyped_search_mixed_items() -> None:
+    client = FakeClient()
+    client.beta._search_items = [
+        make_agent_data({"name": "Carol", "age": 22}),
+        make_agent_data({"name": "Dave", "age": "bad"}),
+    ]
+    client.beta._total_size = 2
+
+    adc = AsyncAgentDataClient(type=Person, client=client, deployment_name="dep")
+    results = await adc.untyped_search(include_total=True)
+    assert len(results.items) == 2
+    assert results.items[0].data == {"name": "Carol", "age": 22}
+    assert results.items[1].data == {"name": "Dave", "age": "bad"}
+    assert results.total_size == 2
+
+
+@pytest.mark.asyncio
+async def test_untyped_aggregate_first_item_dict() -> None:
+    client = FakeClient()
+    client.beta._aggregate_items = [
+        make_group({"k": 1}, {"name": "Eve", "age": 40}),
+        make_group({"k": 2}, {"name": "Frank", "age": "bad"}),
+    ]
+    client.beta._total_size = 2
+
+    adc = AsyncAgentDataClient(type=Person, client=client, deployment_name="dep")
+    results = await adc.untyped_aggregate(group_by=["k"], first=True)
+    assert len(results.items) == 2
+    assert results.items[0].first_item == {"name": "Eve", "age": 40}
+    assert results.items[1].first_item == {"name": "Frank", "age": "bad"}
@@ -38,7 +38,7 @@ def test_typed_agent_data_from_raw():
    """Test TypedAgentData.from_raw class method."""
    raw_data = AgentData(
        id="456",
-        agent_slug="extraction-agent",
+        deployment_name="extraction-agent",
        collection="employees",
        data={"name": "Jane Smith", "age": 25, "email": "jane@company.com"},
        created_at=datetime.now(),
@@ -48,7 +48,7 @@ def test_typed_agent_data_from_raw():
    typed_data = TypedAgentData.from_raw(raw_data, Person)

    assert typed_data.id == "456"
-    assert typed_data.agent_url_id == "extraction-agent"
+    assert typed_data.deployment_name == "extraction-agent"
    assert typed_data.collection == "employees"
    assert typed_data.data.name == "Jane Smith"
    assert typed_data.data.age == 25
@@ -56,10 +56,10 @@ def test_typed_agent_data_from_raw():


 def test_typed_agent_data_from_raw_validation_error():
-    """Test TypedAgentData.from_raw with invalid data."""
+    """Test TypedAgentData.from_raw with invalid data now raises InvalidTypedAgentData."""
    raw_data = AgentData(
        id="789",
-        agent_slug="test-agent",
+        deployment_name="test-agent",
        collection="people",
        data={"name": "Invalid Person", "age": "not_a_number"},  # Invalid age
        created_at=datetime.now(),
@@ -613,3 +613,51 @@ def test_parses_field_metadata_with_error_field():
    }
    assert parsed.metadata.get("field_errors") == "This is an error"
    assert parsed.metadata.get("job_id") == "job-123"
+
+
+REASONING_IN_SCHEMA = {
+    "majority_opinion": {
+        "type": {
+            "citation": [
+                {
+                    "page": 4,
+                    "matching_text": "BARRETT, J., delivered the opinion for a unanimous Court.",
+                },
+                {"page": 11, "matching_text": "Opinion of the Court"},
+            ],
+            "parsing_confidence": 1.0,
+            "extraction_confidence": 0.9999998919950147,
+            "confidence": 0.9999998919950147,
+        },
+        "reasoning": {
+            "citation": [
+                {
+                    "page": 15,
+                    "matching_text": "We hold that §5110(b)(1) is not subject to equitable tolling and affirm the judg...",
+                }
+            ],
+            "parsing_confidence": 1.0,
+            "extraction_confidence": 0.414292785946868,
+            "confidence": 0.414292785946868,
+        },
+    },
+    "reasoning": {
+        "citation": [
+            {
+                "page": 15,
+                "matching_text": "We hold that §5110(b)(1) is not subject to equitable tolling and affirm the judg...",
+            }
+        ],
+        "parsing_confidence": 1.0,
+        "extraction_confidence": 0.414292785946868,
+        "confidence": 0.414292785946868,
+    },
+}
+
+
+def test_field_conflict_in_schema():
+    extracted = parse_extracted_field_metadata(REASONING_IN_SCHEMA)
+    assert isinstance(extracted["reasoning"], ExtractedFieldMetadata)
+    assert isinstance(
+        extracted["majority_opinion"]["reasoning"], ExtractedFieldMetadata
+    )
@@ -118,10 +118,8 @@ async def test_extraction_agent_aextract_accepts_llama_file(
    dummy_llama_extract_iface = SimpleNamespace()

    async def fake_run_job(**kwargs):
-        # Ensure we are receiving a request with the right file_id
-        request = kwargs.get("request")
-        assert hasattr(request, "file_id")
-        assert request.file_id == llama_file.id
+        file_id = kwargs.get("file_id")
+        assert file_id == llama_file.id
        return SimpleNamespace(id="job_42")

    dummy_llama_extract_iface.run_job = fake_run_job
@@ -1,16 +1,155 @@
 import pytest
+from unittest.mock import MagicMock, patch

+import llama_cloud_services.index.base as base
+from llama_cloud import (
+    PipelineEmbeddingConfig_ManagedOpenaiEmbedding,
+    Project,
+    Pipeline,
+    CloudDocument,
+)
+from llama_index.core.constants import DEFAULT_PROJECT_NAME
 from llama_index.core.indices.managed.base import BaseManagedIndex
-from llama_cloud_services.index import (
-    LlamaCloudIndex,
+from llama_index.core.schema import Document
+from llama_cloud_services.index import LlamaCloudIndex
+
+
+# Simple test data as values, not fixtures
+TEST_PROJECT = Project(id="proj-123", name="test-project", organization_id="org-123")
+
+EMBEDDING_CONFIG = PipelineEmbeddingConfig_ManagedOpenaiEmbedding(
+    type="MANAGED_OPENAI_EMBEDDING"
+)
+TEST_PIPELINE = Pipeline(
+    id="pipe-456",
+    name="test-pipeline",
+    project_id="proj-123",
+    embedding_config=PipelineEmbeddingConfig_ManagedOpenaiEmbedding(
+        type="MANAGED_OPENAI_EMBEDDING"
+    ),
 )


-def test_class():
+@pytest.fixture
+def mock_client() -> MagicMock:
+    """Mock client with sensible defaults."""
+    client = MagicMock()
+    client.projects.upsert_project.return_value = Project(
+        id="default-proj", name=DEFAULT_PROJECT_NAME, organization_id="default-org"
+    )
+    client.pipelines.upsert_pipeline.return_value = Pipeline(
+        id="default-pipe",
+        name="default",
+        project_id="default-proj",
+        embedding_config=EMBEDDING_CONFIG,
+    )
+    client.pipelines.upsert_batch_pipeline_documents.return_value = [
+        CloudDocument(id="doc-1", text="test", metadata={})
+    ]
+    return client
+
+
+@pytest.fixture(autouse=True)
+def base_patches(mock_client: MagicMock) -> None:
+    """Auto-applied patches for all tests."""
+    with (
+        patch.object(base, "get_client", return_value=mock_client),
+        patch.object(
+            base,
+            "resolve_project_and_pipeline",
+            return_value=(TEST_PROJECT, TEST_PIPELINE),
+        ),
+        patch.object(base.LlamaCloudIndex, "wait_for_completion"),
+    ):
+        yield
+
+
+def test_class() -> None:
    names_of_base_classes = [b.__name__ for b in LlamaCloudIndex.__mro__]
    assert BaseManagedIndex.__name__ in names_of_base_classes


-def test_conflicting_index_identifiers():
+def test_conflicting_index_identifiers() -> None:
    with pytest.raises(ValueError):
        LlamaCloudIndex(name="test", pipeline_id="test", index_id="test")
+
+
+def test_from_documents_uses_provided_project_id(mock_client: MagicMock) -> None:
+    provided_project_id = "proj-123"
+    organization_id = "org-abc"
+    index_name = "my_new_index"
+
+    # Override resolve to return project with provided ID
+    test_project = Project(
+        id=provided_project_id, name="my_project", organization_id=organization_id
+    )
+    test_pipeline = Pipeline(
+        id="pipe-xyz",
+        name=index_name,
+        project_id=provided_project_id,
+        embedding_config=EMBEDDING_CONFIG,
+    )
+
+    with patch.object(
+        base, "resolve_project_and_pipeline", return_value=(test_project, test_pipeline)
+    ):
+        docs = [Document(text="hello")]
+        index = LlamaCloudIndex.from_documents(
+            documents=docs,
+            name=index_name,
+            project_id=provided_project_id,
+        )
+
+    # Assert - project upsert not called; pipeline uses provided project_id
+    mock_client.projects.upsert_project.assert_not_called()
+    assert mock_client.pipelines.upsert_pipeline.call_count == 1
+    assert (
+        mock_client.pipelines.upsert_pipeline.call_args.kwargs["project_id"]
+        == provided_project_id
+    )
+    assert index.project.id == provided_project_id
+
+
+def test_from_documents_upserts_project_when_project_id_missing(
+    mock_client: MagicMock,
+) -> None:
+    organization_id = "org-xyz"
+    index_name = "my_new_index"
+
+    # Project is created when project_id is not provided
+    upserted_project = Project(
+        id="proj-999", name=DEFAULT_PROJECT_NAME, organization_id=organization_id
+    )
+    mock_client.projects.upsert_project.return_value = upserted_project
+
+    test_pipeline = Pipeline(
+        id="pipe-xyz",
+        name=index_name,
+        project_id=upserted_project.id,
+        embedding_config=EMBEDDING_CONFIG,
+    )
+
+    with patch.object(
+        base,
+        "resolve_project_and_pipeline",
+        return_value=(upserted_project, test_pipeline),
+    ):
+        docs = [Document(text="world")]
+        index = LlamaCloudIndex.from_documents(
+            documents=docs,
+            name=index_name,
+            organization_id=organization_id,
+        )
+
+    # Assert - project was upserted with org id and default project name
+    mock_client.projects.upsert_project.assert_called_once()
+    kwargs = mock_client.projects.upsert_project.call_args.kwargs
+    assert kwargs["organization_id"] == organization_id
+    assert kwargs["request"].name == DEFAULT_PROJECT_NAME
+
+    # Pipeline created under the upserted project id
+    assert (
+        mock_client.pipelines.upsert_pipeline.call_args.kwargs["project_id"]
+        == upserted_project.id
+    )
+    assert index.project.id == upserted_project.id
@@ -1,5 +1,5 @@
 version = 1
-revision = 2
+revision = 3
 requires-python = ">=3.9, <4.0"
 resolution-markers = [
    "python_full_version >= '3.14'",
@@ -1582,21 +1582,21 @@ wheels = [

 [[package]]
 name = "llama-cloud"
-version = "0.1.41"
+version = "0.1.43"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
    { name = "certifi" },
    { name = "httpx" },
    { name = "pydantic" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/62/6c/b2e84eebed376aea34c446cab745da5fc4e9dc53309180672299083219d5/llama_cloud-0.1.41.tar.gz", hash = "sha256:dcb741b779e3e740cd64928cfffc8ef70ed0e9bae9ef26acbe1d7e32aa737bdc", size = 109854, upload-time = "2025-09-05T22:45:13.069Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/9b/33/33a8bd3a617c071caf450ca2627969f8b28272d0692f122997c10a32247e/llama_cloud-0.1.43.tar.gz", hash = "sha256:00429f05aea515449d90cde91ef3ed3687fcd93e46f6246d08cbea02f9b397a9", size = 112992, upload-time = "2025-10-02T21:55:38.355Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/1e/4d/f0af76b389310840ce3483a92560a152025b0eefe4eee0c81102bf3317e6/llama_cloud-0.1.41-py3-none-any.whl", hash = "sha256:c847f288f0d3f4b23f47345088006deae5f2cf3f223ac1819d4c1531e9aaa13e", size = 307646, upload-time = "2025-09-05T22:45:11.597Z" },
+    { url = "https://files.pythonhosted.org/packages/2b/54/559a67542396d5660a71115b29e0160e9dd784e570e1f4ef55ad22bf5b39/llama_cloud-0.1.43-py3-none-any.whl", hash = "sha256:540605d4dd13c6536a3b75cd4d04b211f29b16d17faee9381e3793a651f1dec1", size = 311460, upload-time = "2025-10-02T21:55:37.282Z" },
 ]

 [[package]]
 name = "llama-cloud-services"
-version = "0.6.65"
+version = "0.6.73"
 source = { editable = "." }
 dependencies = [
    { name = "click", version = "8.1.8", source = { registry = "https://pypi.org/simple" }, marker = "python_full_version < '3.10'" },
@@ -1631,9 +1631,9 @@ dev = [
 requires-dist = [
    { name = "click", specifier = ">=8.1.7,<9" },
    { name = "eval-type-backport", marker = "python_full_version < '3.10'", specifier = ">=0.2.0,<0.3" },
-    { name = "llama-cloud", specifier = "==0.1.41" },
+    { name = "llama-cloud", specifier = "==0.1.43" },
    { name = "llama-index-core", specifier = ">=0.12.0" },
-    { name = "packaging", specifier = ">=25.0" },
+    { name = "packaging", specifier = ">=23.0" },
    { name = "platformdirs", specifier = ">=4.3.7,<5" },
    { name = "pydantic", specifier = ">=2.8,!=2.10" },
    { name = "python-dotenv", specifier = ">=1.0.1,<2" },
@@ -0,0 +1,290 @@
+#!/usr/bin/env -S uv run --script
+# /// script
+# dependencies = ["click", "tomlkit", "packaging"]
+# ///
+
+"""
+This is a script called by the changeset bot. Normally changeset can do the following things, but this is a mixed ts and python repo, so we need to do some extra things.
+
+There's 2 things this does:
+- Versioning: Makes changes that may be committed with the newest version.
+- Releasing/Tagging: After versions are changed, we check each package to see if its released, and if not, we release it and tag it.
+
+"""
+
+from dataclasses import dataclass
+import json
+import os
+import subprocess
+from pathlib import Path
+from typing import Any, List, cast
+import urllib.request
+import urllib.error
+import re
+
+import click
+import tomlkit
+from packaging.version import Version
+
+
+def _run_command(
+    cmd: List[str], cwd: Path | None = None, env: dict[str, str] | None = None
+) -> None:
+    """Run a command, streaming output to the console, and raise on failure."""
+    subprocess.run(cmd, check=True, text=True, cwd=cwd or Path.cwd(), env=env)
+
+
+def _run_and_capture(
+    cmd: List[str], cwd: Path | None = None, env: dict[str, str] | None = None
+) -> str:
+    """Run a command and return stdout as text, raising on failure."""
+    result = subprocess.run(
+        cmd,
+        check=True,
+        text=True,
+        cwd=cwd or Path.cwd(),
+        env=env,
+        capture_output=True,
+    )
+    return result.stdout
+
+
+@dataclass
+class Package:
+    name: str
+    version: str
+    path: Path
+
+    def python_package_name(self) -> str | None:
+        if "/py/" in str(self.path) or str(self.path).endswith("/py"):
+            return self.name.removesuffix("-py")
+        return None
+
+
+def _get_pnpm_workspace_packages() -> list[Package]:
+    """Return directories for all workspace packages from pnpm list JSON output."""
+    output = _run_and_capture(["pnpm", "list", "-r", "--depth=-1", "--json"])
+
+    data = cast(list[dict[str, Any]], json.loads(output))
+    packages: list[Package] = [
+        Package(name=data["name"], version=data["version"], path=Path(data["path"]))
+        for data in data
+    ]
+    return packages
+
+
+def _sync_package_version_with_pyproject(
+    package_dir: Path, packages: dict[str, Package], js_package_name: str
+) -> None:
+    """Sync version from package.json to pyproject.toml.
+
+    Returns True if pyproject was changed, else False.
+    """
+    pyproject_path = package_dir / "pyproject.toml"
+    if not pyproject_path.exists():
+        return
+
+    package_version = packages[js_package_name].version
+    py_doc = tomlkit.parse(pyproject_path.read_text())
+
+    by_python_name = {
+        pkg.python_package_name(): pkg
+        for pkg in packages.values()
+        if pkg.python_package_name()
+    }
+
+    current_version = py_doc["project"]["version"]
+    assert isinstance(current_version, str)
+
+    # update workspace dependency strings by replacing the first version after == or >=
+    deps = py_doc["project"]["dependencies"] or []
+    changed = False
+    for i, dep in enumerate(deps):
+        if not isinstance(dep, str):
+            continue
+        pkg = (cast(str, dep).split("==")[0]).split(">=")[0]
+        if pkg not in by_python_name:
+            continue
+        target_version = by_python_name[pkg].version
+        new_dep = re.sub(
+            r"(==|>=)\s*([0-9A-Za-z_.+-]+)",
+            lambda m: m.group(1) + target_version,
+            dep,
+            count=1,
+        )
+        if new_dep != dep:
+            deps[i] = new_dep
+            changed = True
+
+    if current_version != package_version:
+        py_doc["project"]["version"] = package_version
+        changed = True
+
+    if changed:
+        pyproject_path.write_text(tomlkit.dumps(py_doc))
+        click.echo(
+            f"Updated {pyproject_path} version to {package_version} and synced dependency specs"
+        )
+
+
+def lock_python_dependencies() -> None:
+    """Lock Python dependencies."""
+    try:
+        _run_command(["uv", "lock"])
+        click.echo("Locked Python dependencies")
+    except subprocess.CalledProcessError as e:
+        click.echo(f"Warning: Failed to lock Python dependencies: {e}", err=True)
+
+
+@click.group()
+def cli() -> None:
+    """Changeset-based version management for llama-cloud-services."""
+    pass
+
+
+@cli.command()
+def version() -> None:
+    """Apply changeset versions, then sync versions for co-located JS/Py packages.
+
+    - Runs changesets to bump package.json versions.
+    - Discovers all workspace packages via pnpm.
+    - For any directory containing both package.json and pyproject.toml, and with
+      package.json private: false, set pyproject [project].version to match the JS version.
+    - If a pyproject is updated, run `uv sync` in that directory to update its lock file.
+    """
+    # Ensure we're at the repo root
+    os.chdir(Path(__file__).parent.parent)
+
+    # First, run changeset version to update all package.json files
+    _run_command(["npx", "@changesets/cli", "version"])
+
+    # Enumerate workspace packages and perform syncs
+    packages = _get_pnpm_workspace_packages()
+    version_map = {pkg.name: pkg for pkg in packages}
+    for pkg in packages:
+        _sync_package_version_with_pyproject(pkg.path, version_map, pkg.name)
+
+
+@cli.command()
+@click.option("--tag", is_flag=True, help="Tag the packages after publishing")
+@click.option("--dry-run", is_flag=True, help="Dry run the publish")
+@click.option("--js/--no-js", default=True, help="Publish the js package")
+@click.option("--py/--no-py", default=True, help="Publish the py package")
+def publish(tag: bool, dry_run: bool, js: bool, py: bool) -> None:
+    """Publish all packages."""
+    # move to the root
+    os.chdir(Path(__file__).parent.parent)
+
+    if js:
+        if not os.getenv("NPM_TOKEN"):
+            click.echo("NPM_TOKEN is not set, skipping publish", err=True)
+            raise click.Abort("No token set")
+    if py:
+        if not os.getenv("LLAMA_PARSE_PYPI_TOKEN"):
+            click.echo("LLAMA_PARSE_PYPI_TOKEN is not set, skipping publish", err=True)
+            raise click.Abort("No token set")
+
+    # not general script. Just checks each of the 2 packages to see if they need to be published.
+    if js:
+        maybe_publish_npm(dry_run)
+    if py:
+        maybe_publish_pypi(dry_run)
+
+    if tag:
+        if dry_run:
+            click.echo("Dry run, skipping tag. Would run:")
+            click.echo("  npx @changesets/cli tag")
+            click.echo("  git push --tags")
+        else:
+            # Let changesets create JS-related tags as usual
+            _run_command(["npx", "@changesets/cli", "tag"])
+            _run_command(["git", "push", "--tags"])
+
+
+def maybe_publish_npm(dry_run: bool) -> None:
+    """Publish the ts package if it needs to be published."""
+    target_dir = Path("ts/llama_cloud_services")
+    ts_path_package = target_dir / "package.json"
+    package_json = json.loads(ts_path_package.read_text())
+    version = package_json["version"]
+
+    # Check if this version is already published on npm
+    result = subprocess.run(
+        ["npm", "view", "llama-cloud-services", "versions", "--json"],
+        check=True,
+        capture_output=True,
+        text=True,
+        cwd=target_dir,
+    )
+
+    published_versions = json.loads(result.stdout)
+    if version in published_versions:
+        click.echo(
+            f"npm  package llama-cloud-services@{version} already published, skipping"
+        )
+        return
+    click.echo(f"Publishing npm package llama-cloud-services@{version}")
+    # defer to the package.json publish script
+    if dry_run:
+        click.echo("Dry run, skipping publish. Would run:")
+        click.echo("  pnpm run publish")
+        return
+    else:
+        _run_command(["pnpm", "run", "build"], cwd=target_dir)
+        _run_command(["pnpm", "publish"], cwd=target_dir)
+
+
+def maybe_publish_pypi(dry_run: bool) -> None:
+    """Publish the py packages if they need to be published."""
+    for pyproject in list(Path("py").glob("*/pyproject.toml")) + [
+        Path("py/pyproject.toml")
+    ]:
+        name, version = current_version(pyproject)
+        if is_published(name, version):
+            click.echo(f"PyPI package {name}@{version} already published, skipping")
+            continue
+        click.echo(f"Publishing PyPI package {name}@{version}")
+
+        # Use different tokens for different packages
+        env = os.environ.copy()
+        token = os.environ["LLAMA_PARSE_PYPI_TOKEN"]
+        env["UV_PUBLISH_TOKEN"] = token
+        if dry_run:
+            summary = (token[:3] + "***") if len(token) <= 6 else token[:6] + "****"
+            click.echo(
+                f"Dry run, skipping publish. Would run with publish token {summary}:"
+            )
+            click.echo("  uv build")
+            click.echo("  uv publish")
+        else:
+            _run_command(["uv", "build"], cwd=pyproject.parent)
+            _run_command(["uv", "publish"], cwd=pyproject.parent, env=env)
+
+
+def current_version(pyproject: Path) -> tuple[str, str]:
+    """Return (package_name, version_str) taken from the given pyproject.toml."""
+    doc = tomlkit.parse(pyproject.read_text())
+    name = doc["project"]["name"]
+    version = str(Version(doc["project"]["version"]))  # normalise
+    return name, version
+
+
+def is_published(
+    name: str, version: str, index_url: str = "https://pypi.org/pypi"
+) -> bool:
+    """
+    True  → `<name>==<version>` exists on the given index
+    False → package missing *or* version missing
+    """
+    url = f"{index_url.rstrip('/')}/{name}/json"
+    try:
+        data = json.load(urllib.request.urlopen(url))
+    except urllib.error.HTTPError as e:  # 404 → package not published at all
+        if e.code == 404:
+            return False
+        raise  # any other error should surface
+    return version in data["releases"]  # keys are version strings
+
+
+if __name__ == "__main__":
+    cli()
@@ -1,226 +0,0 @@
-#!/usr/bin/env -S uv run --script
-# /// script
-# dependencies = ["click", "tomlkit"]
-# ///
-
-import click
-import subprocess
-import sys
-import tomlkit
-from pathlib import Path
-import json
-
-
-def get_current_versions() -> tuple[str, str, str, str | None]:
-    """Get current versions from both pyproject.toml files and TS package.json."""
-    # Read main pyproject.toml
-    main_content = Path("py/pyproject.toml").read_text()
-    main_doc = tomlkit.parse(main_content)
-    main_version = main_doc["project"]["version"]
-
-    # Read llama_parse/pyproject.toml
-    llama_parse_content = Path("py/llama_parse/pyproject.toml").read_text()
-    llama_parse_doc = tomlkit.parse(llama_parse_content)
-    llama_parse_version = llama_parse_doc["project"]["version"]
-    # Find llama-cloud-services dependency in the dependencies list
-    dependency_version = None
-    for dep in llama_parse_doc["project"]["dependencies"]:
-        if isinstance(dep, str) and dep.startswith("llama-cloud-services"):
-            dependency_version = (
-                dep.split("==")[1]
-                if "==" in dep
-                else dep.split(">=")[1]
-                if ">=" in dep
-                else None
-            )
-            break
-
-    # Read TypeScript package.json version via helper
-    ts_version: str = get_ts_version()
-
-    return (
-        str(main_version),
-        str(llama_parse_version),
-        str(dependency_version),
-        str(ts_version) if ts_version is not None else None,
-    )
-
-
-def validate_versions(
-    main_version: str,
-    llama_parse_version: str,
-    dependency_version: str,
-) -> list[str]:
-    """Validate that versions are consistent and return warnings."""
-    warnings = []
-
-    if main_version != llama_parse_version:
-        warnings.append(
-            f"Version mismatch: main={main_version}, llama_parse={llama_parse_version}"
-        )
-
-    # Extract version from dependency string (e.g., ">=0.6.51" -> "0.6.51")
-    if dependency_version and dependency_version.startswith(">="):
-        dep_ver = dependency_version[2:]
-        if dep_ver != main_version:
-            warnings.append(
-                f"Dependency version mismatch: dependency={dep_ver}, main={main_version}"
-            )
-
-    return warnings
-
-
-def set_version(version: str) -> None:
-    """Set version across Python projects (no TS change)."""
-    # Update main pyproject.toml
-    main_content = Path("py/pyproject.toml").read_text()
-    main_doc = tomlkit.parse(main_content)
-    main_doc["project"]["version"] = version
-    Path("py/pyproject.toml").write_text(tomlkit.dumps(main_doc))
-
-    # Update llama_parse/pyproject.toml
-    llama_parse_content = Path("py/llama_parse/pyproject.toml").read_text()
-    llama_parse_doc = tomlkit.parse(llama_parse_content)
-    llama_parse_doc["project"]["version"] = version
-    for dep_index, dep in enumerate(llama_parse_doc["project"]["dependencies"]):
-        if isinstance(dep, str) and dep.startswith("llama-cloud-services"):
-            llama_parse_doc["project"]["dependencies"][
-                dep_index
-            ] = f"llama-cloud-services>={version}"
-            break
-    Path("py/llama_parse/pyproject.toml").write_text(tomlkit.dumps(llama_parse_doc))
-
-    click.echo(f"Updated Python versions to {version}")
-
-
-def get_ts_version() -> str:
-    """Read TypeScript package.json version (if present)."""
-    ts_package_path = Path("ts/llama_cloud_services/package.json")
-    package_data = json.loads(ts_package_path.read_text())
-    data = package_data.get("version")
-    if data is None:
-        raise RuntimeError("TypeScript package.json version not found")
-    return data
-
-
-def set_ts_version(version: str) -> None:
-    """Set TypeScript package.json version only."""
-    ts_package_path = Path("ts/llama_cloud_services/package.json")
-    package_data = json.loads(ts_package_path.read_text())
-    package_data["version"] = version
-    ts_package_path.write_text(json.dumps(package_data, indent=2) + "\n")
-    click.echo(f"Updated TypeScript package.json version to {version}")
-
-
-def get_current_branch() -> str:
-    """Get the current git branch."""
-    result = subprocess.run(
-        ["git", "branch", "--show-current"], capture_output=True, text=True, check=True
-    )
-    return result.stdout.strip()
-
-
-def create_if_not_exists(version: str) -> str:
-    """Create a git tag and push it."""
-    current_branch = get_current_branch()
-    if current_branch != "main":
-        click.echo(
-            f"Error: Not on main branch (currently on {current_branch})", err=True
-        )
-        sys.exit(1)
-
-    tag_name = f"v{version}" if version[0].isdigit() else version
-    if not tag_exists(tag_name):
-        # Create tag
-        subprocess.run(["git", "tag", tag_name], check=True)
-        click.echo(f"Created tag {tag_name}")
-    else:
-        click.echo(f"Tag {tag_name} already exists")
-    return tag_name
-
-
-def tag_exists(tag_name: str) -> bool:
-    """Check if a git tag exists."""
-    result = subprocess.run(
-        ["git", "tag", "-l", tag_name], capture_output=True, text=True, check=True
-    )
-    return tag_name in result.stdout.strip()
-
-
-def push_tag(tag_name: str) -> None:
-    """Push a git tag."""
-    subprocess.run(["git", "push", "origin", tag_name], check=True)
-    click.echo(f"Pushed tag {tag_name}")
-
-
-@click.group()
-def cli() -> None:
-    """Version management for llama-cloud-services."""
-    pass
-
-
-@cli.command()
-def get() -> None:
-    """Get current versions and show validation warnings."""
-    (
-        main_version,
-        llama_parse_version,
-        dependency_version,
-        ts_version,
-    ) = get_current_versions()
-
-    click.echo("Current versions:")
-    click.echo(f"  llama-cloud-services: {main_version}")
-    click.echo(f"  llama-parse: {llama_parse_version}")
-    click.echo(f"  dependency reference: {dependency_version}")
-    click.echo(f"  typescript package: {ts_version}")
-
-    warnings = validate_versions(main_version, llama_parse_version, dependency_version)
-    if warnings:
-        click.echo("\nValidation warnings:")
-        for warning in warnings:
-            click.echo(f"  ⚠️  {warning}")
-    else:
-        click.echo("\n✅ All versions are consistent")
-
-
-@cli.command()
-@click.argument("version")
-@click.option("--js", is_flag=True, help="Update TypeScript package.json only")
-def set(version: str, js: bool) -> None:
-    """Set version for Python, TypeScript, or both (default: Python only)."""
-
-    if js:
-        set_ts_version(version)
-        return
-    else:
-        set_version(version)
-
-
-@cli.command()
-@click.option(
-    "--version", help="Version to tag (uses current version if not specified)"
-)
-@click.option(
-    "--push",
-    is_flag=True,
-    help="Push the tag to the remote repository",
-)
-@click.option(
-    "--js",
-    is_flag=True,
-    help="tag TypeScript package.json only",
-)
-def tag(version: str | None = None, push: bool = False, js: bool = False) -> None:
-    """Create and push a git tag for the current version."""
-    if not version:
-        main_version, _, _, js_version = get_current_versions()
-        version = f"llama-cloud-services@{js_version}" if js else main_version
-
-    tag_name = create_if_not_exists(version)
-    if push:
-        push_tag(tag_name)
-
-
-if __name__ == "__main__":
-    cli()
@@ -9,10 +9,12 @@ test("LlamaIndex module resolution test", async (t) => {
    const index = new LlamaCloudIndex({
      name: "test-index",
      projectName: "Default",
+      apiKey: process.env.LLAMA_CLOUD_API_KEY || "test-key",
    });
    const reader = new LlamaParseReader({
      resultType: "markdown",
      verbose: false,
+      apiKey: process.env.LLAMA_CLOUD_API_KEY || "test-key",
    });
    ok(index !== undefined);
    ok(reader !== undefined);
@@ -24,6 +26,7 @@ test("LlamaIndex module resolution test", async (t) => {
    const index = new mod.LlamaCloudIndex({
      name: "test-index",
      projectName: "Default",
+      apiKey: process.env.LLAMA_CLOUD_API_KEY || "test-key",
    });
    ok(index !== undefined);
  });
@@ -1,5 +1,23 @@
 # llama-cloud-services

+## 0.3.9
+
+### Patch Changes
+
+- 5d4cabd: Add ImageNode support in TypeScript
+
+## 0.3.8
+
+### Patch Changes
+
+- 6e0f2f4: Agent data extraction citations can be undefined
+
+## 0.3.7
+
+### Patch Changes
+
+- d028397: Update llama-cloud api version, and integrate with agent data deletion
+
 ## v0.1.0

 First release for `llama-cloud-services`.
@@ -1,9 +1,10 @@
 {
  "name": "llama-cloud-services",
-  "version": "0.3.5",
+  "version": "0.3.9",
  "type": "module",
  "license": "MIT",
  "scripts": {
+    "get-openapi": "node ./scripts/get-openapi.js",
    "generate": "./node_modules/.bin/openapi-ts",
    "build": "pnpm run generate && bunchee",
    "dev": "bunchee --watch",
@@ -13,7 +14,8 @@
    "test": "vitest run --testTimeout=60000",
    "test:watch": "vitest --watch",
    "test:ui": "vitest --ui",
-    "test:coverage": "vitest --coverage"
+    "test:coverage": "vitest --coverage",
+    "release": "pnpm run build && pnpm publish"
  },
  "files": [
    "openapi.json",
@@ -0,0 +1,21 @@
+import fs from 'fs';
+
+async function downloadOpenApiSpec() {
+    try {
+        const response = await fetch('https://api.cloud.llamaindex.ai/api/openapi.json');
+
+        if (!response.ok) {
+            throw new Error(`HTTP error! status: ${response.status}`);
+        }
+
+        const data = await response.json();
+
+        fs.writeFileSync('openapi.json', JSON.stringify(data, null, 2));
+        console.log('Successfully downloaded openapi.json');
+    } catch (error) {
+        console.error('Error downloading OpenAPI spec:', error);
+        process.exit(1);
+    }
+}
+
+downloadOpenApiSpec();
@@ -9,10 +9,16 @@ import { DEFAULT_PROJECT_NAME } from "@llamaindex/core/global";
 import type { QueryBundle } from "@llamaindex/core/query-engine";
 import { BaseRetriever } from "@llamaindex/core/retriever";
 import type { NodeWithScore } from "@llamaindex/core/schema";
-import { jsonToNode, ObjectType } from "@llamaindex/core/schema";
+import { jsonToNode, ObjectType, ImageNode } from "@llamaindex/core/schema";
 import { extractText } from "@llamaindex/core/utils";
 import type { ClientParams, CloudConstructorParams } from "./type.js";
-import { getPipelineId, initService } from "./utils.js";
+import { getPipelineId, getProjectId, initService } from "./utils.js";
+import {
+  type PageScreenshotNodeWithScore,
+  type PageFigureNodeWithScore,
+  generateFilePageScreenshotPresignedUrlApiV1FilesIdPageScreenshotsPageIndexPresignedUrlPost,
+  generateFilePageFigurePresignedUrlApiV1FilesIdPageFiguresPageIndexFigureNamePresignedUrlPost,
+} from "./api";

 export type CloudRetrieveParams = Omit<
  RetrievalParams,
@@ -43,6 +49,95 @@ export class LlamaCloudRetriever extends BaseRetriever {
    });
  }

+  private async fetchBase64FromPresignedUrl(url: string): Promise<string> {
+    const response = await fetch(url);
+    if (!response.ok) {
+      throw new Error(
+        `Failed to fetch media from presigned URL: ${response.status} ${response.statusText}`,
+      );
+    }
+    const buffer = Buffer.from(await response.arrayBuffer());
+    return buffer.toString("base64");
+  }
+
+  private async pageScreenshotNodesToNodeWithScore(
+    nodes: PageScreenshotNodeWithScore[] | undefined,
+    projectId: string,
+  ): Promise<NodeWithScore[]> {
+    if (!nodes || nodes.length === 0) return [];
+
+    const results = await Promise.all(
+      nodes.map(async (n) => {
+        const { data: presigned } =
+          await generateFilePageScreenshotPresignedUrlApiV1FilesIdPageScreenshotsPageIndexPresignedUrlPost(
+            {
+              throwOnError: true,
+              path: {
+                id: n.node.file_id,
+                page_index: n.node.page_index,
+              },
+              query: {
+                project_id: projectId,
+                organization_id: this.organizationId ?? null,
+              },
+            },
+          );
+        const base64 = await this.fetchBase64FromPresignedUrl(presigned.url);
+        const imageNode = new ImageNode({
+          image: base64,
+          metadata: {
+            ...(n.node.metadata ?? {}),
+            file_id: n.node.file_id,
+            page_index: n.node.page_index,
+          },
+        });
+        return { node: imageNode, score: n.score } satisfies NodeWithScore;
+      }),
+    );
+
+    return results;
+  }
+
+  private async pageFigureNodesToNodeWithScore(
+    nodes: PageFigureNodeWithScore[] | undefined,
+    projectId: string,
+  ): Promise<NodeWithScore[]> {
+    if (!nodes || nodes.length === 0) return [];
+
+    const results = await Promise.all(
+      nodes.map(async (n) => {
+        const { data: presigned } =
+          await generateFilePageFigurePresignedUrlApiV1FilesIdPageFiguresPageIndexFigureNamePresignedUrlPost(
+            {
+              throwOnError: true,
+              path: {
+                id: n.node.file_id,
+                page_index: n.node.page_index,
+                figure_name: n.node.figure_name,
+              },
+              query: {
+                project_id: projectId,
+                organization_id: this.organizationId ?? null,
+              },
+            },
+          );
+        const base64 = await this.fetchBase64FromPresignedUrl(presigned.url);
+        const imageNode = new ImageNode({
+          image: base64,
+          metadata: {
+            ...(n.node.metadata ?? {}),
+            file_id: n.node.file_id,
+            page_index: n.node.page_index,
+            figure_name: n.node.figure_name,
+          },
+        });
+        return { node: imageNode, score: n.score } satisfies NodeWithScore;
+      }),
+    );
+
+    return results;
+  }
+
  // LlamaCloud expects null values for filters, but LlamaIndexTS uses undefined for empty values
  // This function converts the undefined values to null
  private convertFilter(filters?: MetadataFilters): MetadataFilters | null {
@@ -76,6 +171,35 @@ export class LlamaCloudRetriever extends BaseRetriever {
  }

  async _retrieve(query: QueryBundle): Promise<NodeWithScore[]> {
+    // Handle deprecated image retrieval flag
+    const retrieveImageNodes = (this.retrieveParams as RetrievalParams)
+      .retrieve_image_nodes;
+    if (typeof retrieveImageNodes !== "undefined") {
+      console.warn(
+        "The `retrieve_image_nodes` parameter is deprecated. Use `retrieve_page_screenshot_nodes` and `retrieve_page_figure_nodes` instead.",
+      );
+    }
+
+    const retrievePageScreenshotNodes = (this.retrieveParams as RetrievalParams)
+      .retrieve_page_screenshot_nodes;
+    const retrievePageFigureNodes = (this.retrieveParams as RetrievalParams)
+      .retrieve_page_figure_nodes;
+
+    if (retrieveImageNodes) {
+      if (
+        retrievePageScreenshotNodes === false ||
+        retrievePageFigureNodes === false
+      ) {
+        throw new Error(
+          "If `retrieve_image_nodes` is set to true, both `retrieve_page_screenshot_nodes` and `retrieve_page_figure_nodes` must also be set to true or omitted.",
+        );
+      }
+      (this.retrieveParams as RetrievalParams).retrieve_page_screenshot_nodes =
+        true;
+      (this.retrieveParams as RetrievalParams).retrieve_page_figure_nodes =
+        true;
+    }
+
    const pipelineId = await getPipelineId(
      this.pipelineName,
      this.projectName,
@@ -98,6 +222,34 @@ export class LlamaCloudRetriever extends BaseRetriever {
        },
      });

-    return this.resultNodesToNodeWithScore(results.retrieval_nodes);
+    const textNodes = this.resultNodesToNodeWithScore(results.retrieval_nodes);
+
+    const needScreenshots = (this.retrieveParams as RetrievalParams)
+      .retrieve_page_screenshot_nodes;
+    const needFigures = (this.retrieveParams as RetrievalParams)
+      .retrieve_page_figure_nodes;
+
+    if (!needScreenshots && !needFigures) {
+      return textNodes;
+    }
+
+    const projectId = await getProjectId(this.projectName, this.organizationId);
+
+    const [screenshotNodes, figureNodes] = await Promise.all([
+      needScreenshots
+        ? this.pageScreenshotNodesToNodeWithScore(
+            results.image_nodes,
+            projectId,
+          )
+        : Promise.resolve([] as NodeWithScore[]),
+      needFigures
+        ? this.pageFigureNodesToNodeWithScore(
+            results.page_figure_nodes,
+            projectId,
+          )
+        : Promise.resolve([] as NodeWithScore[]),
+    ]);
+
+    return [...textNodes, ...screenshotNodes, ...figureNodes];
  }
 }
@@ -4,6 +4,7 @@ import {
  aggregateAgentDataApiV1BetaAgentDataAggregatePost,
  createAgentDataApiV1BetaAgentDataPost,
  deleteAgentDataApiV1BetaAgentDataItemIdDelete,
+  deleteAgentDataByQueryApiV1BetaAgentDataDeletePost,
  getAgentDataApiV1BetaAgentDataItemIdGet,
  searchAgentDataApiV1BetaAgentDataSearchPost,
  updateAgentDataApiV1BetaAgentDataItemIdPut,
@@ -12,6 +13,7 @@ import {
 } from "../../client";
 import type {
  AggregateAgentDataOptions,
+  DeleteAgentDataOptions,
  SearchAgentDataOptions,
  TypedAgentData,
  TypedAgentDataItems,
@@ -25,20 +27,23 @@ import type {
 export class AgentClient<T = unknown> {
  private client: ReturnType<typeof createClient>;
  private collection: string;
-  private agentUrlId: string;
+  private deploymentName: string;

  constructor({
    client = defaultClient,
    collection = "default",
-    agentUrlId = "_public",
+    deploymentName = "_public",
+    agentUrlId,
  }: {
    client?: ReturnType<typeof createClient>;
    collection?: string;
+    deploymentName?: string;
+    // deprecated, use deploymentName instead
    agentUrlId?: string;
  }) {
    this.client = client;
    this.collection = collection;
-    this.agentUrlId = agentUrlId;
+    this.deploymentName = agentUrlId || deploymentName;
  }

  /**
@@ -48,7 +53,7 @@ export class AgentClient<T = unknown> {
    const response = await createAgentDataApiV1BetaAgentDataPost({
      throwOnError: true,
      body: {
-        agent_slug: this.agentUrlId,
+        deployment_name: this.deploymentName,
        collection: this.collection,
        data: data as Record<string, unknown>,
      },
@@ -109,6 +114,24 @@ export class AgentClient<T = unknown> {
    });
  }

+  /**
+   * Delete all matching agent data, returns the total number of deleted items
+   */
+  async delete(options: DeleteAgentDataOptions): Promise<number> {
+    const response = await deleteAgentDataByQueryApiV1BetaAgentDataDeletePost({
+      throwOnError: true,
+      body: {
+        deployment_name: this.deploymentName,
+        ...(this.collection !== undefined && {
+          collection: this.collection,
+        }),
+        ...(options.filter !== undefined && { filter: options.filter }),
+      },
+      client: this.client,
+    });
+    return response.data.deleted_count;
+  }
+
  /**
   * Search agent data
   */
@@ -118,7 +141,7 @@ export class AgentClient<T = unknown> {
    const response = await searchAgentDataApiV1BetaAgentDataSearchPost({
      throwOnError: true,
      body: {
-        agent_slug: this.agentUrlId,
+        deployment_name: this.deploymentName,
        ...(this.collection !== undefined && {
          collection: this.collection,
        }),
@@ -165,7 +188,7 @@ export class AgentClient<T = unknown> {
    const response = await aggregateAgentDataApiV1BetaAgentDataAggregatePost({
      throwOnError: true,
      body: {
-        agent_slug: this.agentUrlId,
+        deployment_name: this.deploymentName,
        ...(this.collection !== undefined && {
          collection: this.collection,
        }),
@@ -209,7 +232,7 @@ export class AgentClient<T = unknown> {
  private transformResponse(data: AgentData): TypedAgentData<T> {
    const result: TypedAgentData<T> = {
      id: data.id!,
-      agentUrlId: data.agent_slug,
+      deploymentName: data.deployment_name,
      data: data.data as T,
      createdAt: new Date(data.created_at!),
      updatedAt: new Date(data.updated_at!),
@@ -250,10 +273,10 @@ export interface AgentDataClientOptions {
  /** 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 */
+  /** If running in an agent runtime, optionally provide the window url to infer the deployment name */
  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;
+  /** Deployment name for the client, if not provided, it will be inferred from the window url, or fall back to "default" */
+  deploymentName?: string;
  /** Collection name for the client, defaults to "default" */
  collection?: string;
 }
@@ -267,22 +290,25 @@ export function createAgentDataClient<T = unknown>({
  client = defaultClient,
  windowUrl,
  env,
+  deploymentName,
  agentUrlId,
  collection = "default",
 }: {
  client?: ReturnType<typeof createClient>;
  windowUrl?: string;
  env?: Record<string, string>;
+  deploymentName?: string;
+  // deprecated, use deploymentName instead
  agentUrlId?: string;
  collection?: string;
 } = {}): AgentClient<T> {
-  if (env && !agentUrlId) {
-    agentUrlId =
+  if (env && !deploymentName) {
+    deploymentName =
      env.LLAMA_DEPLOY_DEPLOYMENT_NAME ||
      env.NEXT_PUBLIC_LLAMA_DEPLOY_DEPLOYMENT_NAME ||
      env.VITE_LLAMA_DEPLOY_DEPLOYMENT_NAME;
  }
-  if (windowUrl && !agentUrlId) {
+  if (windowUrl && !deploymentName) {
    try {
      const url = new URL(windowUrl);
      const path = url.pathname;
@@ -291,17 +317,18 @@ export function createAgentDataClient<T = unknown>({
        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];
+        deploymentName = path.split("/")[2];
      }
    } catch (error) {
      console.warn(
-        "Failed to infer agent url id from window url, falling back to default",
+        "Failed to infer deployment name from window url, falling back to default",
        error,
      );
    }
  }

  return new AgentClient({
+    ...(deploymentName && { deploymentName }),
    ...(agentUrlId && { agentUrlId }),
    collection,
    client,
@@ -38,7 +38,7 @@ export interface ExtractedFieldMetadata {
  confidence?: number;
  /** The confidence score for the field based on the extracted text only */
  extraction_confidence?: number;
-  citation: FieldCitation[];
+  citation?: FieldCitation[];
 }

 export interface FieldCitation {
@@ -87,8 +87,8 @@ export interface ExtractedData<T = unknown> {
 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 deployment name of the agent that created the data. */
+  deploymentName: string;
  /** The collection of the agent data. */
  collection?: string;
  /** The data of the agent data. Usually an ExtractedData&lt;SomeOtherType&gt; */
@@ -127,6 +127,14 @@ export interface SearchAgentDataOptions {
  includeTotal?: boolean;
 }

+/**
+ * Options for deleting agent data
+ */
+export interface DeleteAgentDataOptions {
+  /** Filter options for the deletion. */
+  filter?: Record<string, FilterOperation>;
+}
+
 /**
 * Options for aggregating agent data
 */
@@ -0,0 +1,246 @@
+import { describe, it, expect, beforeEach, vi } from "vitest";
+import { AgentClient, createAgentDataClient } from "../src/beta/agent/index.js";
+import * as sdk from "../src/client/index.js";
+
+describe("AgentClient", () => {
+  beforeEach(() => {
+    vi.restoreAllMocks();
+  });
+
+  it("createItem sends correct payload and returns typed data", async () => {
+    const spy = vi
+      .spyOn(sdk, "createAgentDataApiV1BetaAgentDataPost")
+      .mockResolvedValue({
+        data: {
+          id: "1",
+          deployment_name: "dep",
+          collection: "col",
+          data: { foo: "bar" },
+          created_at: "2024-01-01T00:00:00Z",
+          updated_at: "2024-01-01T00:00:00Z",
+        },
+      } as any);
+
+    const client = new AgentClient<{ foo: string }>({
+      deploymentName: "dep",
+      collection: "col",
+    });
+    const result = await client.createItem({ foo: "bar" });
+
+    expect(spy).toHaveBeenCalledOnce();
+    const call = spy.mock.calls[0][0];
+    expect(call.body.deployment_name).toBe("dep");
+    expect(call.body.collection).toBe("col");
+    expect(call.body.data).toEqual({ foo: "bar" });
+
+    expect(result.id).toBe("1");
+    expect(result.deploymentName).toBe("dep");
+    expect(result.collection).toBe("col");
+    expect(result.data).toEqual({ foo: "bar" });
+    expect(result.createdAt).toEqual(new Date("2024-01-01T00:00:00Z"));
+    expect(result.updatedAt).toEqual(new Date("2024-01-01T00:00:00Z"));
+  });
+
+  it("getItem returns null for 404 errors", async () => {
+    const spy = vi
+      .spyOn(sdk, "getAgentDataApiV1BetaAgentDataItemIdGet")
+      .mockImplementation(async () => {
+        const err: any = new Error("Not found");
+        err.response = { status: 404 };
+        throw err;
+      });
+
+    const client = new AgentClient({ deploymentName: "dep" });
+    const res = await client.getItem("missing-id");
+
+    expect(spy).toHaveBeenCalledOnce();
+    expect(res).toBeNull();
+  });
+
+  it("updateItem updates and returns typed data", async () => {
+    const spy = vi
+      .spyOn(sdk, "updateAgentDataApiV1BetaAgentDataItemIdPut")
+      .mockResolvedValue({
+        data: {
+          id: "123",
+          deployment_name: "dep",
+          collection: "col",
+          data: { foo: "baz" },
+          created_at: "2024-01-01T00:00:00Z",
+          updated_at: "2024-01-02T00:00:00Z",
+        },
+      } as any);
+
+    const client = new AgentClient<{ foo: string }>({
+      deploymentName: "dep",
+      collection: "col",
+    });
+    const res = await client.updateItem("123", { foo: "baz" });
+
+    expect(spy).toHaveBeenCalledOnce();
+    const call = spy.mock.calls[0][0];
+    expect(call.path.item_id).toBe("123");
+    expect(call.body.data).toEqual({ foo: "baz" });
+
+    expect(res.id).toBe("123");
+    expect(res.updatedAt).toEqual(new Date("2024-01-02T00:00:00Z"));
+  });
+
+  it("deleteItem calls delete endpoint with correct path", async () => {
+    const spy = vi
+      .spyOn(sdk, "deleteAgentDataApiV1BetaAgentDataItemIdDelete")
+      .mockResolvedValue({} as any);
+
+    const client = new AgentClient({ deploymentName: "dep" });
+    await client.deleteItem("abc");
+
+    expect(spy).toHaveBeenCalledOnce();
+    expect(spy.mock.calls[0][0].path.item_id).toBe("abc");
+  });
+
+  it("delete by query returns deleted count", async () => {
+    const spy = vi
+      .spyOn(sdk, "deleteAgentDataByQueryApiV1BetaAgentDataDeletePost")
+      .mockResolvedValue({ data: { deleted_count: 7 } } as any);
+
+    const client = new AgentClient({
+      deploymentName: "dep",
+      collection: "col",
+    });
+    const count = await client.delete({
+      filter: { status: { op: "eq", value: "accepted" } as any },
+    });
+
+    expect(spy).toHaveBeenCalledOnce();
+    const body = spy.mock.calls[0][0].body;
+    expect(body.deployment_name).toBe("dep");
+    expect(body.collection).toBe("col");
+    expect(count).toBe(7);
+  });
+
+  it("search maps items and optional fields correctly", async () => {
+    const now = "2024-01-01T00:00:00Z";
+    const spy = vi
+      .spyOn(sdk, "searchAgentDataApiV1BetaAgentDataSearchPost")
+      .mockResolvedValue({
+        data: {
+          items: [
+            {
+              id: "1",
+              deployment_name: "dep",
+              collection: "col",
+              data: { foo: "bar" },
+              created_at: now,
+              updated_at: now,
+            },
+          ],
+          total_size: 1,
+          next_page_token: "next",
+        },
+      } as any);
+
+    const client = new AgentClient<{ foo: string }>({
+      deploymentName: "dep",
+      collection: "col",
+    });
+    const result = await client.search({
+      includeTotal: true,
+      orderBy: "created_at desc",
+      pageSize: 1,
+      offset: 0,
+    });
+
+    expect(spy).toHaveBeenCalledOnce();
+    const body = spy.mock.calls[0][0].body;
+    expect(body.deployment_name).toBe("dep");
+    expect(body.collection).toBe("col");
+    expect(body.include_total).toBe(true);
+    expect(body.order_by).toBe("created_at desc");
+    expect(body.page_size).toBe(1);
+    expect(body.offset).toBe(0);
+
+    expect(result.items).toHaveLength(1);
+    expect(result.totalSize).toBe(1);
+    expect(result.nextPageToken).toBe("next");
+    expect(result.items[0].createdAt).toEqual(new Date(now));
+  });
+
+  it("aggregate maps groups and optional fields correctly", async () => {
+    const spy = vi
+      .spyOn(sdk, "aggregateAgentDataApiV1BetaAgentDataAggregatePost")
+      .mockResolvedValue({
+        data: {
+          items: [
+            {
+              group_key: { status: "accepted" },
+              count: 3,
+              first_item: { foo: "bar" },
+            },
+          ],
+          total_size: 1,
+          next_page_token: "tok",
+        },
+      } as any);
+
+    const client = new AgentClient<{ foo: string }>({
+      deploymentName: "dep",
+      collection: "col",
+    });
+    const result = await client.aggregate({
+      groupBy: ["status"],
+      count: true,
+      first: true,
+      pageSize: 1,
+      offset: 0,
+    });
+
+    expect(spy).toHaveBeenCalledOnce();
+    const body = spy.mock.calls[0][0].body;
+    expect(body.deployment_name).toBe("dep");
+    expect(body.collection).toBe("col");
+    expect(body.group_by).toEqual(["status"]);
+    expect(body.count).toBe(true);
+    expect(body.first).toBe(true);
+    expect(body.page_size).toBe(1);
+    expect(body.offset).toBe(0);
+
+    expect(result.items).toHaveLength(1);
+    expect(result.totalSize).toBe(1);
+    expect(result.nextPageToken).toBe("tok");
+    expect(result.items[0].groupKey).toEqual({ status: "accepted" });
+    expect(result.items[0].count).toBe(3);
+    expect(result.items[0].firstItem).toEqual({ foo: "bar" });
+  });
+
+  it("createAgentDataClient infers deployment name from env", async () => {
+    const spy = vi
+      .spyOn(sdk, "searchAgentDataApiV1BetaAgentDataSearchPost")
+      .mockResolvedValue({
+        data: { items: [], total_size: 0 },
+      } as any);
+
+    const client = createAgentDataClient({
+      env: { LLAMA_DEPLOY_DEPLOYMENT_NAME: "env-dep" },
+    });
+    await client.search({});
+
+    const body = spy.mock.calls[0][0].body;
+    expect(body.deployment_name).toBe("env-dep");
+  });
+
+  it("createAgentDataClient infers deployment name from windowUrl (non-local)", async () => {
+    const spy = vi
+      .spyOn(sdk, "deleteAgentDataByQueryApiV1BetaAgentDataDeletePost")
+      .mockResolvedValue({
+        data: { deleted_count: 0 },
+      } as any);
+
+    const client = createAgentDataClient({
+      windowUrl: "https://app.llamaindex.ai/deployments/abc/ui/",
+    });
+    await client.delete({});
+
+    const body = spy.mock.calls[0][0].body;
+    expect(body.deployment_name).toBe("abc");
+  });
+});
Author	SHA1	Message	Date
github-actions[bot]	d0649ece6e	chore: version packages (#982 )	2025-10-16 16:58:29 -06:00
MartijnLeplae	5d4cabd843	Add ImageNode support in TypeScript (#969 )	2025-10-16 16:56:28 -06:00
github-actions[bot]	9070a6ac16	chore: version packages (#981 )	2025-10-15 12:01:34 -06:00
Bogdan Gheorghe	4f24f537f6	Add agressive table extraction argument (#980 )	2025-10-15 11:57:34 -06:00
github-actions[bot]	8859a203e2	chore: version packages (#977 )	2025-10-14 19:03:36 -06:00
dependabot[bot]	b091364054	build(deps): bump astral-sh/setup-uv from 6 to 7 (#974 )	2025-10-14 19:02:32 -06:00
dependabot[bot]	43b1a013ca	build(deps): bump github/codeql-action from 3 to 4 (#973 )	2025-10-14 19:02:20 -06:00
Logan	f81532e7f2	safest types possible for parse (#976 )	2025-10-14 19:02:07 -06:00
github-actions[bot]	986d3987d3	chore: version packages (#965 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>	2025-10-14 08:14:49 -06:00
Logan	1bf522311f	fix default bbox values (#975 )	2025-10-14 07:44:35 -06:00
Preston Carlson	24166dcfc8	Only escape single dollar sign in notebook md (#964 ) * Limit escaping to lone dollar signs - preserve double dollar for latex equations * Updated uv.lock via make lint * Patch bump * Unit test for _format_markdown_for_notebook Test doesn't depend on getting real results/is just testing a string manipulation function, so inserting before other tests. Should move to its own file if we add additional formatting configurations	2025-10-07 08:06:03 -07:00
github-actions[bot]	bfb7f3973f	chore: version packages (#956 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>	2025-10-06 11:15:55 -04:00
dependabot[bot]	979f643c77	build(deps): bump actions/checkout from 4 to 5 (#961 )	2025-10-06 09:12:38 -06:00
dependabot[bot]	aefd89cf1b	build(deps): bump actions/setup-python from 5 to 6 (#960 )	2025-10-06 09:12:30 -06:00
dependabot[bot]	8ea2b2c64e	build(deps): bump pnpm/action-setup from 3 to 4 (#959 )	2025-10-06 09:12:20 -06:00
dependabot[bot]	4a9a2a21d8	build(deps): bump astral-sh/setup-uv from 3 to 6 (#958 )	2025-10-06 09:12:08 -06:00
Logan	e6a7939206	loosen packaging requirements (#962 )	2025-10-06 09:11:57 -06:00
Adrian Lyjak	104a03e829	fix: re-enable js publishing (#963 )	2025-10-06 11:10:46 -04:00
Terry Zhao	6e0f2f4ca0	citation can be null (#869 ) * citation can be null * Add changeset --------- Co-authored-by: Terry Zhao <terryzhao@runllama.ai> Co-authored-by: Adrian Lyjak <adrianlyjak@gmail.com>	2025-10-04 16:26:11 -04:00
dependabot[bot]	0708d11f8a	Bump actions/setup-node from 4 to 5 (#909 ) Bumps [actions/setup-node](https://github.com/actions/setup-node) from 4 to 5. - [Release notes](https://github.com/actions/setup-node/releases) - [Commits](https://github.com/actions/setup-node/compare/v4...v5) --- updated-dependencies: - dependency-name: actions/setup-node dependency-version: '5' dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>	2025-10-04 16:21:50 -04:00
github-actions[bot]	be19185503	chore: version packages (#954 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>	2025-10-03 20:14:04 -04:00
Adrian Lyjak	7571b0d6c4	Missed some things again with tag fixes (#955 ) guh	2025-10-03 20:12:53 -04:00
Adrian Lyjak	ad6734bf80	fixup tagging more better (#953 ) * fix: correct private field type in py/package.json to be recognized by pnpm * use packages more directly, make public * add bump * fix crash	2025-10-03 19:53:57 -04:00
github-actions[bot]	9ec2a8322e	chore: version packages (#952 )	2025-10-03 15:11:14 -06:00
Logan	51011b9f30	fix changeset harder (#951 )	2025-10-03 15:09:58 -06:00
Logan	09805f9e15	swap changesets (#949 )	2025-10-03 15:06:00 -06:00
Adrian Lyjak	8ced6f6eab	fix: explicitly tag. I thought the action did this (#948 )	2025-10-03 16:59:41 -04:00
Preston Carlson	081ddeca34	Escaping dollar signs in md output when running in a jupyter notebook (#945 )	2025-10-03 14:52:26 -06:00
Adrian Lyjak	2460908789	Disable npm release (#946 )	2025-10-03 16:13:16 -04:00
Adrian Lyjak	c226d6a54c	Fix more bugs in publishing (#944 )	2025-10-03 11:16:43 -04:00
Adrian Lyjak	5d4c682eb2	fix: theres just one publish token (#943 )	2025-10-03 10:56:10 -04:00
github-actions[bot]	f72d3535c8	chore: version packages (#941 ) Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>	2025-10-03 10:25:11 -04:00
Adrian Lyjak	1ea09a366e	Update llama-cloud dep (#940 )	2025-10-03 09:56:56 -04:00
Adrian Lyjak	d4bbeb6389	ignore nvmrc (#942 ) ignore npmrc	2025-10-03 00:21:32 -04:00
Adrian Lyjak	d028397603	version and release via changesets (#849 )	2025-10-03 00:08:52 -04:00
Emanuel Ferreira	35ea8476db	docs: parse -> classify -> extract (#931 )	2025-09-24 18:52:15 -03:00
Logan	3e5f7c4f1e	Update parse.md	2025-09-24 11:35:13 -06:00
Adrian Lyjak	9d9b816644	Handle reasoning field conflict (#929 ) * Handle reasoning field conflict * update version to 0.6.69	2025-09-22 11:29:11 -04:00
Adrian Lyjak	83555f76e6	Handle validation errors for agent data retrieval (#928 ) * feat: Add untyped agent data retrieval and handling Introduces methods to retrieve agent data as untyped dictionaries, handling validation errors gracefully. This allows for more flexible data access when strict typing is not required or when data may be malformed. Co-authored-by: adrian <adrian@runllama.ai> * Expose raw api result --------- Co-authored-by: Cursor Agent <cursoragent@cursor.com>	2025-09-22 11:28:49 -04:00
Adrian Lyjak	5edf5f914a	Support creating indexes in a specified project_id (#924 ) * Support creating indexes in a specified project_id * Bump	2025-09-18 11:07:07 -04:00
Adrian Lyjak	22e4975cb2	Refactor agent fields in llama_cloud_services (#921 )	2025-09-17 15:14:40 -04:00
Peter Rowlands (변기호)	bc2f04379b	py: bump version to v.0.6.66 (#920 )	2025-09-16 19:34:18 +09:00
Peter Rowlands (변기호)	f9f951d5d8	parse: expose spreadsheet_force_formula_computation option (#919 )	2025-09-16 19:28:03 +09:00
Emmanuel Ferdman	355129fea5	Fix colab broken links (#750 ) Signed-off-by: Emmanuel Ferdman <emmanuelferdman@gmail.com>	2025-09-14 23:10:21 +02:00