loosen packaging requirements

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>
2026-07-19 16:43:32 -04:00 · 2025-10-06 09:05:34 -06:00 · 2025-10-04 16:26:11 -04:00 · 2025-10-04 16:21:50 -04:00 · 2025-10-03 20:14:04 -04:00 · 2025-10-03 20:12:53 -04:00
51 changed files with 6069 additions and 3332 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 +1,5 @@
+---
+"llama-cloud-services": patch
+---
+
+Agent data extraction citations can be undefined
@@ -0,0 +1,5 @@
+---
+"llama-cloud-services-py": patch
+---
+
+Loosen packaging dep requirement
@@ -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": []
+}
@@ -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"

@@ -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 }}
@@ -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@v4
+
+      - uses: pnpm/action-setup@v3
+
+      - name: Setup Node.js
+        uses: actions/setup-node@v5
+        with:
+          node-version: "22"
+          cache: "pnpm"
+
+      - name: Setup Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@v3
+
+      - 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:
@@ -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
+}
@@ -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 --no-js --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 +1,19 @@
+# llama-cloud-services-py
+
+## 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,
@@ -157,10 +162,14 @@ 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]:
@@ -185,6 +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,
@@ -211,9 +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(
-            deployment_name=self.deployment_name,
-            collection=self.collection,
+        raw = await self.untyped_search(
            filter=filter,
            order_by=order_by,
            offset=offset,
@@ -228,6 +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,
@@ -254,7 +295,37 @@ 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(
+        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,
@@ -265,11 +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,
-        )
@@ -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])

@@ -116,10 +115,10 @@ 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(
@@ -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)
@@ -4,7 +4,10 @@ import re
 from pydantic import BaseModel, Field, SerializeAsAny
 from typing import Dict, Any, List, Optional

-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

@@ -258,6 +261,24 @@ 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_dollar_signs(text: str) -> str:
+            """Escape dollar signs in text to prevent Jupyter from interpreting them as LaTeX.
+
+            Args:
+                text: The text to escape
+
+            Returns:
+                Text with dollar signs escaped
+            """
+            return text.replace("$", r"\$")
+
+        return escape_dollar_signs(text)
+
    def get_markdown_documents(self, split_by_page: bool = False) -> List[Document]:
        """
        Get the markdown documents from the job.
@@ -268,17 +289,22 @@ 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 +354,10 @@ 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 +1,8 @@
+# llama_parse
+
+## 0.6.72
+
+### Patch Changes
+
+- Updated dependencies [ad6734b]
+  - llama-cloud-services-py@0.6.72
@@ -0,0 +1,20 @@
+{
+  "name": "llama_parse",
+  "version": "0.6.72",
+  "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 = "0.6.67"
+version = "0.6.72"
 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>=0.6.67"]
+dependencies = ["llama-cloud-services>=0.6.72"]

 [project.scripts]
 llama-parse = "llama_parse.cli.main:parse"
@@ -0,0 +1,10 @@
+{
+  "name": "llama-cloud-services-py",
+  "version": "0.6.72",
+  "private": false,
+  "license": "MIT",
+  "scripts": {},
+  "devDependencies": {
+    "changesets": "^1.0.2"
+  }
+}
@@ -19,7 +19,7 @@ dev = [

 [project]
 name = "llama-cloud-services"
-version = "0.6.67"
+version = "0.6.72"
 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==0.1.42",
+  "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]
@@ -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,
@@ -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"}
@@ -56,7 +56,7 @@ 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",
        deployment_name="test-agent",
@@ -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
@@ -1582,21 +1582,21 @@ wheels = [

 [[package]]
 name = "llama-cloud"
-version = "0.1.42"
+version = "0.1.43"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
    { name = "certifi" },
    { name = "httpx" },
    { name = "pydantic" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/21/04/ae0694b582d6aab4d6e7957febb7bff048897ac231ad80ba1bd71547d944/llama_cloud-0.1.42.tar.gz", hash = "sha256:485aa0e364ea648e3aaa3b2c54af7bcb6f2242c50b4f86ec022e137413fff464", size = 112480, upload-time = "2025-09-16T20:25:42.631Z" }
+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/6a/61/85d115699a59d03f0783e119aaf6d534fca95dbe1a4531a8056e6a4774ed/llama_cloud-0.1.42-py3-none-any.whl", hash = "sha256:4ed3edde4a277ff52eeb831188c8476eb079b5e4605ad3142157a0f054b27d96", size = 311857, upload-time = "2025-09-16T20:25:41.479Z" },
+    { 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 = "0.6.66"
+version = "0.6.72"
 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 = "==0.1.42" },
+    { 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()
@@ -1,5 +1,11 @@
 # llama-cloud-services

+## 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": "0.3.6",
+  "version": "0.3.7",
  "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();
@@ -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,
@@ -112,6 +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
   */
@@ -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 {
@@ -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
 */
@@ -1530,27 +1530,6 @@ export const Body_run_job_on_file_api_v1_extraction_jobs_file_postSchema = {
  title: "Body_run_job_on_file_api_v1_extraction_jobs_file_post",
 } as const;

-export const Body_run_job_test_user_api_v1_extraction_jobs_test_postSchema = {
-  properties: {
-    job_create: {
-      $ref: "#/components/schemas/ExtractJobCreate",
-    },
-    extract_settings: {
-      anyOf: [
-        {
-          $ref: "#/components/schemas/LlamaExtractSettings",
-        },
-        {
-          type: "null",
-        },
-      ],
-    },
-  },
-  type: "object",
-  required: ["job_create"],
-  title: "Body_run_job_test_user_api_v1_extraction_jobs_test_post",
-} as const;
-
 export const Body_screenshot_api_parsing_screenshot_postSchema = {
  properties: {
    file: {
@@ -2796,30 +2775,6 @@ export const Body_upload_file_api_v1_parsing_upload_postSchema = {
  title: "Body_upload_file_api_v1_parsing_upload_post",
 } as const;

-export const Body_upload_file_v2_api_v2alpha1_parse_upload_postSchema = {
-  properties: {
-    configuration: {
-      type: "string",
-      title: "Configuration",
-    },
-    file: {
-      anyOf: [
-        {
-          type: "string",
-          format: "binary",
-        },
-        {
-          type: "null",
-        },
-      ],
-      title: "File",
-    },
-  },
-  type: "object",
-  required: ["configuration"],
-  title: "Body_upload_file_v2_api_v2alpha1_parse_upload_post",
-} as const;
-
 export const BoxAuthMechanismSchema = {
  type: "string",
  enum: ["developer_token", "ccg"],
@@ -3180,12 +3135,6 @@ export const ChatMessageSchema = {
  title: "ChatMessage",
 } as const;

-export const ChunkModeSchema = {
-  type: "string",
-  enum: ["PAGE", "DOCUMENT", "SECTION", "GROUPED_PAGES"],
-  title: "ChunkMode",
-} as const;
-
 export const ClassificationResultSchema = {
  properties: {
    reasoning: {
@@ -5486,6 +5435,13 @@ export const CustomClaimsSchema = {
      description: "Whether the user is allowed to delete organizations.",
      default: false,
    },
+    allowed_spreadsheet: {
+      type: "boolean",
+      title: "Allowed Spreadsheet",
+      description:
+        "Whether the user is allowed to access the spreadsheet feature.",
+      default: false,
+    },
  },
  type: "object",
  title: "CustomClaims",
@@ -6213,6 +6169,54 @@ export const DeleteParamsSchema = {
  description: "Schema for the parameters of a delete job.",
 } as const;

+export const DeleteRequestSchema = {
+  properties: {
+    deployment_name: {
+      type: "string",
+      title: "Deployment Name",
+      description: "The agent deployment's name to delete data for",
+    },
+    collection: {
+      type: "string",
+      title: "Collection",
+      description: "The logical agent data collection to delete from",
+      default: "default",
+    },
+    filter: {
+      anyOf: [
+        {
+          additionalProperties: {
+            $ref: "#/components/schemas/FilterOperation",
+          },
+          type: "object",
+        },
+        {
+          type: "null",
+        },
+      ],
+      title: "Filter",
+      description: "Optional filters to select which items to delete",
+    },
+  },
+  type: "object",
+  required: ["deployment_name"],
+  title: "DeleteRequest",
+  description: "API request body for bulk deleting agent data by query",
+} as const;
+
+export const DeleteResponseSchema = {
+  properties: {
+    deleted_count: {
+      type: "integer",
+      title: "Deleted Count",
+    },
+  },
+  type: "object",
+  required: ["deleted_count"],
+  title: "DeleteResponse",
+  description: "API response for bulk delete operation",
+} as const;
+
 export const DirectRetrievalParamsSchema = {
  properties: {
    mode: {
@@ -6946,6 +6950,20 @@ export const ExtractConfigSchema = {
      description: "Whether to invalidate the cache for the extraction.",
      default: false,
    },
+    num_pages_context: {
+      anyOf: [
+        {
+          type: "integer",
+          minimum: 1,
+        },
+        {
+          type: "null",
+        },
+      ],
+      title: "Num Pages Context",
+      description:
+        "Number of pages to pass as context on long document extraction.",
+    },
    page_range: {
      anyOf: [
        {
@@ -7202,6 +7220,7 @@ export const ExtractModelsSchema = {
    "openai-gpt-5-mini",
    "gemini-2.0-flash",
    "gemini-2.5-flash",
+    "gemini-2.5-flash-lite",
    "gemini-2.5-pro",
    "openai-gpt-4o",
    "openai-gpt-4o-mini",
@@ -7849,6 +7868,52 @@ export const ExtractTargetSchema = {
  title: "ExtractTarget",
 } as const;

+export const ExtractedTableSchema = {
+  properties: {
+    table_id: {
+      type: "integer",
+      title: "Table Id",
+      description: "Unique identifier for this table within the file",
+    },
+    sheet_name: {
+      type: "string",
+      title: "Sheet Name",
+      description: "Worksheet name where table was found",
+    },
+    row_span: {
+      type: "integer",
+      title: "Row Span",
+      description: "Number of rows in the table",
+    },
+    col_span: {
+      type: "integer",
+      title: "Col Span",
+      description: "Number of columns in the table",
+    },
+    has_headers: {
+      type: "boolean",
+      title: "Has Headers",
+      description: "Whether the table has header rows",
+    },
+    metadata_json: {
+      anyOf: [
+        {
+          type: "string",
+        },
+        {
+          type: "null",
+        },
+      ],
+      title: "Metadata Json",
+      description: "JSON metadata with detailed table information",
+    },
+  },
+  type: "object",
+  required: ["table_id", "sheet_name", "row_span", "col_span", "has_headers"],
+  title: "ExtractedTable",
+  description: "A single extracted table from a spreadsheet",
+} as const;
+
 export const FailPageModeSchema = {
  type: "string",
  enum: ["raw_text", "blank_page", "error_message"],
@@ -10828,140 +10893,6 @@ export const LegacyParseJobConfigSchema = {
  description: "Configuration for llamaparse job",
 } as const;

-export const LlamaExtractSettingsSchema = {
-  properties: {
-    max_file_size: {
-      type: "integer",
-      title: "Max File Size",
-      description: "The maximum file size (in bytes) allowed for the document.",
-      default: 104857600,
-    },
-    max_file_size_ui: {
-      type: "integer",
-      title: "Max File Size Ui",
-      description: "The maximum file size (in bytes) allowed for the document.",
-      default: 31457280,
-    },
-    max_pages: {
-      type: "integer",
-      title: "Max Pages",
-      description: "The maximum number of pages allowed for the document.",
-      default: 500,
-    },
-    chunk_mode: {
-      $ref: "#/components/schemas/ChunkMode",
-      description: "The mode to use for chunking the document.",
-      default: "SECTION",
-    },
-    max_chunk_size: {
-      type: "integer",
-      title: "Max Chunk Size",
-      description:
-        "The maximum size of the chunks (in tokens) to use for chunking the document.",
-      default: 10000,
-    },
-    extraction_agent_config: {
-      additionalProperties: {
-        $ref: "#/components/schemas/StructParseConf",
-      },
-      type: "object",
-      title: "Extraction Agent Config",
-      description: "The configuration for the extraction agent.",
-    },
-    use_multimodal_parsing: {
-      type: "boolean",
-      title: "Use Multimodal Parsing",
-      description: "Whether to use experimental multimodal parsing.",
-      default: false,
-    },
-    use_pixel_extraction: {
-      type: "boolean",
-      title: "Use Pixel Extraction",
-      description:
-        "DEPRECATED: Whether to use extraction over pixels for multimodal mode.",
-      default: false,
-    },
-    llama_parse_params: {
-      $ref: "#/components/schemas/LlamaParseParameters",
-      description: "LlamaParse related settings.",
-      default: {
-        languages: ["en"],
-        parsing_instruction: "",
-        disable_ocr: false,
-        annotate_links: true,
-        adaptive_long_table: true,
-        compact_markdown_table: false,
-        disable_reconstruction: false,
-        disable_image_extraction: false,
-        invalidate_cache: false,
-        outlined_table_extraction: true,
-        merge_tables_across_pages_in_markdown: false,
-        output_pdf_of_document: false,
-        do_not_cache: false,
-        fast_mode: false,
-        skip_diagonal_text: false,
-        preserve_layout_alignment_across_pages: false,
-        preserve_very_small_text: false,
-        gpt4o_mode: false,
-        do_not_unroll_columns: false,
-        extract_layout: false,
-        high_res_ocr: false,
-        html_make_all_elements_visible: false,
-        layout_aware: false,
-        specialized_chart_parsing_agentic: false,
-        specialized_chart_parsing_plus: false,
-        specialized_chart_parsing_efficient: false,
-        specialized_image_parsing: false,
-        precise_bounding_box: false,
-        html_remove_navigation_elements: false,
-        html_remove_fixed_elements: false,
-        guess_xlsx_sheet_name: false,
-        use_vendor_multimodal_model: false,
-        page_prefix: `<<<PAGE:{pageNumber}>>>
-
-`,
-        page_suffix: `
-
-<<<END_PAGE>>>`,
-        take_screenshot: false,
-        is_formatting_instruction: true,
-        premium_mode: false,
-        continuous_mode: false,
-        auto_mode: false,
-        auto_mode_trigger_on_table_in_page: false,
-        auto_mode_trigger_on_image_in_page: false,
-        structured_output: false,
-        extract_charts: false,
-        spreadsheet_extract_sub_tables: false,
-        spreadsheet_force_formula_computation: false,
-        inline_images_in_markdown: false,
-        strict_mode_image_extraction: false,
-        strict_mode_image_ocr: false,
-        strict_mode_reconstruction: false,
-        strict_mode_buggy_font: false,
-        save_images: true,
-        hide_headers: false,
-        hide_footers: false,
-        ignore_document_elements_for_layout_detection: false,
-        output_tables_as_HTML: false,
-        internal_is_screenshot_job: false,
-        parse_mode: "parse_page_with_llm",
-        page_error_tolerance: 0.05,
-        replace_failed_page_mode: "raw_text",
-      },
-    },
-    multimodal_parse_resolution: {
-      $ref: "#/components/schemas/MultimodalParseResolution",
-      description: "The resolution to use for multimodal parsing.",
-      default: "medium",
-    },
-  },
-  type: "object",
-  title: "LlamaExtractSettings",
-  description: `All settings for the extraction agent. Only the settings in ExtractConfig
-are exposed to the user.`,
-} as const;
-
 export const LlamaParseParametersSchema = {
  properties: {
    webhook_configurations: {
@@ -12602,12 +12533,6 @@ export const MetronomeDashboardTypeSchema = {
  title: "MetronomeDashboardType",
 } as const;

-export const MultimodalParseResolutionSchema = {
-  type: "string",
-  enum: ["medium", "high"],
-  title: "MultimodalParseResolution",
-} as const;
-
 export const NodeRelationshipSchema = {
  type: "string",
  enum: ["1", "2", "3", "4", "5"],
@@ -13430,6 +13355,48 @@ export const PaginatedResponse_QuotaConfiguration_Schema = {
  title: "PaginatedResponse[QuotaConfiguration]",
 } as const;

+export const PaginatedResponse_SpreadsheetJob_Schema = {
+  properties: {
+    items: {
+      items: {
+        $ref: "#/components/schemas/SpreadsheetJob",
+      },
+      type: "array",
+      title: "Items",
+      description: "The list of items.",
+    },
+    next_page_token: {
+      anyOf: [
+        {
+          type: "string",
+        },
+        {
+          type: "null",
+        },
+      ],
+      title: "Next Page Token",
+      description:
+        "A token, which can be sent as page_token to retrieve the next page. If this field is omitted, there are no subsequent pages.",
+    },
+    total_size: {
+      anyOf: [
+        {
+          type: "integer",
+        },
+        {
+          type: "null",
+        },
+      ],
+      title: "Total Size",
+      description:
+        "The total number of items available. This is only populated when specifically requested. The value may be an estimate and can be used for display purposes only.",
+    },
+  },
+  type: "object",
+  required: ["items"],
+  title: "PaginatedResponse[SpreadsheetJob]",
+} as const;
+
 export const ParseConfigurationSchema = {
  properties: {
    id: {
@@ -17841,69 +17808,6 @@ export const ProjectUpdateSchema = {
  description: "Schema for updating a project.",
 } as const;

-export const PromptConfSchema = {
-  properties: {
-    system_prompt: {
-      type: "string",
-      title: "System Prompt",
-      description: "The system prompt to use for the extraction.",
-      default:
-        "Given a JSON schema, extract the data from the provided SOURCE TEXT according to the schema. Only output information that is explicitly stated or can be inferred from the SOURCE TEXT.",
-    },
-    extraction_prompt: {
-      type: "string",
-      title: "Extraction Prompt",
-      description: "The prompt to use for the extraction.",
-      default: "The extracted data using the given JSON schema.",
-    },
-    error_handling_prompt: {
-      type: "string",
-      title: "Error Handling Prompt",
-      description: "The prompt to use for error handling.",
-      default:
-        "If the source text does not contain enough information to extract the value, explain the reason very briefly. Else, output null and fill out the value__ field.",
-    },
-    reasoning_prompt: {
-      type: "string",
-      title: "Reasoning Prompt",
-      description: "The prompt to use for reasoning.",
-      default: `
-Provide a brief explanation for how you arrived at the extracted value based on the source text provided.
- For inferred values, explain the reasoning behind the extraction briefly.
- For simple verbatim extraction, output 'VERBATIM EXTRACTION'.
- When supporting data is not present in the source text, output 'INSUFFICIENT DATA' and emit blank or null values for the value__ field.
-`,
-    },
-    cite_sources_prompt: {
-      additionalProperties: {
-        type: "string",
-      },
-      type: "object",
-      title: "Cite Sources Prompt",
-      description: "The prompt to use for citing sources.",
-      default: {
-        description: `
-### Citation Rules (read carefully):
- You must ANNOTATE every value with the MOST RELEVANT short EXACT substring from the source text that supports it.
- For inferred values, cite the text used to infer it in the matching_text field or output 'INFERRED FROM TEXT'
- If no support exists, output 'INSUFFICIENT DATA' and leave value__ null or '', 0.0, False etc depending on the type of the field.
-`,
-        page: "Cite the page number of the source text that the extracted value is from. The page number is the integer that appears right after <<<PAGE:. If no page number is present in this format, use the default value of 1.",
-        matching_text:
-          'Cite the **MOST RELEVANT EXACT TEXT from the SOURCE TEXT** that supports the extracted value within 80 characters. If the exact substring is >80 chars, truncate with ellipsis "...". Provide only the single most relevant citation.',
-      },
-    },
-    scratchpad_prompt: {
-      type: "string",
-      title: "Scratchpad Prompt",
-      description: "The prompt to use for scratchpad.",
-      default: "Use for intermediate step-by-step reasoning. Be concise.",
-    },
-  },
-  type: "object",
-  title: "PromptConf",
-} as const;
-
 export const PublicModelNameSchema = {
  type: "string",
  enum: [
@@ -17926,6 +17830,7 @@ export const PublicModelNameSchema = {
    "gemini-2.5-pro",
    "gemini-2.0-flash",
    "gemini-2.0-flash-lite",
+    "gemini-2.5-flash-lite",
    "gemini-1.5-flash",
    "gemini-1.5-pro",
  ],
@@ -18752,12 +18657,6 @@ export const RoleSchema = {
  description: "Schema for a role.",
 } as const;

-export const SchemaRelaxModeSchema = {
-  type: "string",
-  enum: ["FULL", "TOP_LEVEL", "LEAF"],
-  title: "SchemaRelaxMode",
-} as const;
-
 export const SearchRequestSchema = {
  properties: {
    page_size: {
@@ -18950,6 +18849,135 @@ BM25: Uses Qdrant's FastEmbed BM25 model for sparse embeddings
 AUTO: Automatically selects based on deployment mode (BYOC uses term frequency, Cloud uses Splade)`,
 } as const;

+export const SpreadsheetJobSchema = {
+  properties: {
+    id: {
+      type: "string",
+      title: "Id",
+      description: "The ID of the job",
+    },
+    user_id: {
+      type: "string",
+      title: "User Id",
+      description: "The ID of the user",
+    },
+    project_id: {
+      type: "string",
+      format: "uuid",
+      title: "Project Id",
+      description: "The ID of the project",
+    },
+    file_id: {
+      type: "string",
+      format: "uuid",
+      title: "File Id",
+      description: "The ID of the file to parse",
+    },
+    config: {
+      $ref: "#/components/schemas/SpreadsheetParsingConfig",
+      description: "Configuration for the parsing job",
+    },
+    status: {
+      $ref: "#/components/schemas/StatusEnum",
+      description: "The status of the parsing job",
+    },
+    created_at: {
+      type: "string",
+      title: "Created At",
+      description: "When the job was created",
+    },
+    updated_at: {
+      type: "string",
+      title: "Updated At",
+      description: "When the job was last updated",
+    },
+    success: {
+      anyOf: [
+        {
+          type: "boolean",
+        },
+        {
+          type: "null",
+        },
+      ],
+      title: "Success",
+      description: "Whether the job completed successfully",
+    },
+    tables: {
+      items: {
+        $ref: "#/components/schemas/ExtractedTable",
+      },
+      type: "array",
+      title: "Tables",
+      description: "All extracted tables (populated when job is complete)",
+    },
+    errors: {
+      items: {
+        type: "string",
+      },
+      type: "array",
+      title: "Errors",
+      description: "Any errors encountered",
+    },
+  },
+  type: "object",
+  required: [
+    "id",
+    "user_id",
+    "project_id",
+    "file_id",
+    "config",
+    "status",
+    "created_at",
+    "updated_at",
+  ],
+  title: "SpreadsheetJob",
+  description: "A spreadsheet parsing job",
+} as const;
+
+export const SpreadsheetJobCreateSchema = {
+  properties: {
+    file_id: {
+      type: "string",
+      format: "uuid",
+      title: "File Id",
+      description: "The ID of the file to parse",
+    },
+    config: {
+      $ref: "#/components/schemas/SpreadsheetParsingConfig",
+      description: "Configuration for the parsing job",
+    },
+  },
+  type: "object",
+  required: ["file_id"],
+  title: "SpreadsheetJobCreate",
+  description: "Request to create a spreadsheet parsing job",
+} as const;
+
+export const SpreadsheetParsingConfigSchema = {
+  properties: {
+    sheet_names: {
+      anyOf: [
+        {
+          items: {
+            type: "string",
+          },
+          type: "array",
+        },
+        {
+          type: "null",
+        },
+      ],
+      title: "Sheet Names",
+      description:
+        "The names of the sheets to parse. If empty, all sheets will be parsed.",
+    },
+  },
+  type: "object",
+  title: "SpreadsheetParsingConfig",
+  description: "Configuration for spreadsheet parsing",
+} as const;
+
 export const StatusEnumSchema = {
  type: "string",
  enum: ["PENDING", "SUCCESS", "ERROR", "PARTIAL_SUCCESS", "CANCELLED"],
@@ -18957,101 +18985,6 @@ export const StatusEnumSchema = {
  description: "Enum for representing the status of a job",
 } as const;

-export const StructModeSchema = {
-  type: "string",
-  enum: [
-    "STRUCT_PARSE",
-    "JSON_MODE",
-    "FUNC_CALL",
-    "STRUCT_RELAXED",
-    "UNSTRUCTURED",
-  ],
-  title: "StructMode",
-} as const;
-
-export const StructParseConfSchema = {
-  properties: {
-    model: {
-      $ref: "#/components/schemas/ExtractModels",
-      description: "The model to use for the structured parsing.",
-      default: "openai-gpt-4-1",
-    },
-    temperature: {
-      type: "number",
-      title: "Temperature",
-      description: "The temperature to use for the structured parsing.",
-      default: 0,
-    },
-    relaxation_mode: {
-      $ref: "#/components/schemas/SchemaRelaxMode",
-      description: "The relaxation mode to use for the structured parsing.",
-      default: "LEAF",
-    },
-    struct_mode: {
-      $ref: "#/components/schemas/StructMode",
-      description: "The struct mode to use for the structured parsing.",
-      default: "STRUCT_PARSE",
-    },
-    fetch_logprobs: {
-      type: "boolean",
-      title: "Fetch Logprobs",
-      description: "Whether to fetch logprobs for the structured parsing.",
-      default: false,
-    },
-    handle_missing: {
-      type: "boolean",
-      title: "Handle Missing",
-      description: "Whether to handle missing fields in the schema.",
-      default: false,
-    },
-    use_reasoning: {
-      type: "boolean",
-      title: "Use Reasoning",
-      description: "Whether to use reasoning for the structured extraction.",
-      default: false,
-    },
-    cite_sources: {
-      type: "boolean",
-      title: "Cite Sources",
-      description: "Whether to cite sources for the structured extraction.",
-      default: false,
-    },
-    prompt_conf: {
-      $ref: "#/components/schemas/PromptConf",
-      description: "The prompt configuration for the structured parsing.",
-      default: {
-        system_prompt:
-          "Given a JSON schema, extract the data from the provided SOURCE TEXT according to the schema. Only output information that is explicitly stated or can be inferred from the SOURCE TEXT.",
-        extraction_prompt: "The extracted data using the given JSON schema.",
-        error_handling_prompt:
-          "If the source text does not contain enough information to extract the value, explain the reason very briefly. Else, output null and fill out the value__ field.",
-        reasoning_prompt: `
-Provide a brief explanation for how you arrived at the extracted value based on the source text provided.
- For inferred values, explain the reasoning behind the extraction briefly.
- For simple verbatim extraction, output 'VERBATIM EXTRACTION'.
- When supporting data is not present in the source text, output 'INSUFFICIENT DATA' and emit blank or null values for the value__ field.
-`,
-        cite_sources_prompt: {
-          description: `
-### Citation Rules (read carefully):
- You must ANNOTATE every value with the MOST RELEVANT short EXACT substring from the source text that supports it.
- For inferred values, cite the text used to infer it in the matching_text field or output 'INFERRED FROM TEXT'
- If no support exists, output 'INSUFFICIENT DATA' and leave value__ null or '', 0.0, False etc depending on the type of the field.
-`,
-          matching_text:
-            'Cite the **MOST RELEVANT EXACT TEXT from the SOURCE TEXT** that supports the extracted value within 80 characters. If the exact substring is >80 chars, truncate with ellipsis "...". Provide only the single most relevant citation.',
-          page: "Cite the page number of the source text that the extracted value is from. The page number is the integer that appears right after <<<PAGE:. If no page number is present in this format, use the default value of 1.",
-        },
-        scratchpad_prompt:
-          "Use for intermediate step-by-step reasoning. Be concise.",
-      },
-    },
-  },
-  type: "object",
-  title: "StructParseConf",
-  description: "Configuration for the structured parsing agent.",
-} as const;
-
 export const SupportedLLMModelSchema = {
  properties: {
    name: {
@@ -691,115 +691,6 @@ export const zBodyRunJobOnFileApiV1ExtractionJobsFilePost = z.object({
  config_override: z.union([z.string(), z.null()]).optional(),
 });

-export const zExtractTarget = z.enum(["PER_DOC", "PER_PAGE"]);
-
-export const zExtractMode = z.enum([
-  "FAST",
-  "BALANCED",
-  "PREMIUM",
-  "MULTIMODAL",
-]);
-
-export const zPublicModelName = z.enum([
-  "openai-gpt-4o",
-  "openai-gpt-4o-mini",
-  "openai-gpt-4-1",
-  "openai-gpt-4-1-mini",
-  "openai-gpt-4-1-nano",
-  "openai-gpt-5",
-  "openai-gpt-5-mini",
-  "openai-gpt-5-nano",
-  "openai-text-embedding-3-small",
-  "openai-text-embedding-3-large",
-  "openai-whisper-1",
-  "anthropic-sonnet-3.5",
-  "anthropic-sonnet-3.5-v2",
-  "anthropic-sonnet-3.7",
-  "anthropic-sonnet-4.0",
-  "gemini-2.5-flash",
-  "gemini-2.5-pro",
-  "gemini-2.0-flash",
-  "gemini-2.0-flash-lite",
-  "gemini-1.5-flash",
-  "gemini-1.5-pro",
-]);
-
-export const zExtractModels = z.enum([
-  "openai-gpt-4-1",
-  "openai-gpt-4-1-mini",
-  "openai-gpt-4-1-nano",
-  "openai-gpt-5",
-  "openai-gpt-5-mini",
-  "gemini-2.0-flash",
-  "gemini-2.5-flash",
-  "gemini-2.5-pro",
-  "openai-gpt-4o",
-  "openai-gpt-4o-mini",
-]);
-
-export const zDocumentChunkMode = z.enum(["PAGE", "SECTION"]);
-
-export const zExtractConfig = z.object({
-  priority: z
-    .union([z.enum(["low", "medium", "high", "critical"]), z.null()])
-    .optional(),
-  extraction_target: zExtractTarget.optional(),
-  extraction_mode: zExtractMode.optional(),
-  parse_model: z.union([zPublicModelName, z.null()]).optional(),
-  extract_model: z.union([zExtractModels, z.null()]).optional(),
-  multimodal_fast_mode: z.boolean().optional().default(false),
-  system_prompt: z.union([z.string(), z.null()]).optional(),
-  use_reasoning: z.boolean().optional().default(false),
-  cite_sources: z.boolean().optional().default(false),
-  confidence_scores: z.boolean().optional().default(false),
-  chunk_mode: zDocumentChunkMode.optional(),
-  high_resolution_mode: z.boolean().optional().default(false),
-  invalidate_cache: z.boolean().optional().default(false),
-  page_range: z.union([z.string(), z.null()]).optional(),
-});
-
-export const zExtractJobCreate = z.object({
-  priority: z
-    .union([z.enum(["low", "medium", "high", "critical"]), z.null()])
-    .optional(),
-  webhook_configurations: z
-    .union([z.array(zWebhookConfiguration), z.null()])
-    .optional(),
-  extraction_agent_id: z.string().uuid(),
-  file_id: z.string().uuid(),
-  data_schema_override: z
-    .union([z.object({}), z.string(), z.null()])
-    .optional(),
-  config_override: z.union([zExtractConfig, z.null()]).optional(),
-});
-
-export const zChunkMode = z.enum([
-  "PAGE",
-  "DOCUMENT",
-  "SECTION",
-  "GROUPED_PAGES",
-]);
-
-export const zMultimodalParseResolution = z.enum(["medium", "high"]);
-
-export const zLlamaExtractSettings = z.object({
-  max_file_size: z.number().int().optional().default(104857600),
-  max_file_size_ui: z.number().int().optional().default(31457280),
-  max_pages: z.number().int().optional().default(500),
-  chunk_mode: zChunkMode.optional(),
-  max_chunk_size: z.number().int().optional().default(10000),
-  extraction_agent_config: z.object({}).optional(),
-  use_multimodal_parsing: z.boolean().optional().default(false),
-  use_pixel_extraction: z.boolean().optional().default(false),
-  llama_parse_params: zLlamaParseParameters.optional(),
-  multimodal_parse_resolution: zMultimodalParseResolution.optional(),
-});
-
-export const zBodyRunJobTestUserApiV1ExtractionJobsTestPost = z.object({
-  job_create: zExtractJobCreate,
-  extract_settings: z.union([zLlamaExtractSettings, z.null()]).optional(),
-});
-
 export const zBodyScreenshotApiParsingScreenshotPost = z.object({
  file: z.union([z.string(), z.null()]).optional(),
  do_not_cache: z.boolean().optional().default(false),
@@ -1072,11 +963,6 @@ export const zBodyUploadFileApiV1ParsingUploadPost = z.object({
  page_footer_suffix: z.string().optional(),
 });

-export const zBodyUploadFileV2ApiV2Alpha1ParseUploadPost = z.object({
-  configuration: z.string(),
-  file: z.union([z.string(), z.null()]).optional(),
-});
-
 export const zBoxAuthMechanism = z.enum(["developer_token", "ccg"]);

 export const zSupportedLlmModelNames = z.enum([
@@ -1700,6 +1586,7 @@ export const zCustomClaims = z.object({
  allowed_classify: z.boolean().optional().default(true),
  api_datasource_access: z.boolean().optional().default(false),
  allow_org_deletion: z.boolean().optional().default(false),
+  allowed_spreadsheet: z.boolean().optional().default(false),
 });

 export const zCustomerPortalSessionCreatePayload = z.object({
@@ -1855,6 +1742,16 @@ export const zDefaultOrganizationUpdate = z.object({
  organization_id: z.string().uuid(),
 });

+export const zDeleteRequest = z.object({
+  deployment_name: z.string(),
+  collection: z.string().optional().default("default"),
+  filter: z.union([z.object({}), z.null()]).optional(),
+});
+
+export const zDeleteResponse = z.object({
+  deleted_count: z.number().int(),
+});
+
 export const zRetrieverPipeline = z.object({
  name: z.union([z.string().min(1).max(3000), z.null()]),
  description: z.union([z.string().max(15000), z.null()]),
@@ -1870,6 +1767,8 @@ export const zDirectRetrievalParams = z.object({
  pipelines: z.array(zRetrieverPipeline).optional(),
 });

+export const zDocumentChunkMode = z.enum(["PAGE", "SECTION"]);
+
 export const zDocumentIngestionJobParams = z.object({
  custom_metadata: z.union([z.object({}), z.null()]).optional(),
  resource_info: z.union([z.object({}), z.null()]).optional(),
@@ -2122,6 +2021,74 @@ Query: {query_str}
 Answer: `),
 });

+export const zExtractTarget = z.enum(["PER_DOC", "PER_PAGE"]);
+
+export const zExtractMode = z.enum([
+  "FAST",
+  "BALANCED",
+  "PREMIUM",
+  "MULTIMODAL",
+]);
+
+export const zPublicModelName = z.enum([
+  "openai-gpt-4o",
+  "openai-gpt-4o-mini",
+  "openai-gpt-4-1",
+  "openai-gpt-4-1-mini",
+  "openai-gpt-4-1-nano",
+  "openai-gpt-5",
+  "openai-gpt-5-mini",
+  "openai-gpt-5-nano",
+  "openai-text-embedding-3-small",
+  "openai-text-embedding-3-large",
+  "openai-whisper-1",
+  "anthropic-sonnet-3.5",
+  "anthropic-sonnet-3.5-v2",
+  "anthropic-sonnet-3.7",
+  "anthropic-sonnet-4.0",
+  "gemini-2.5-flash",
+  "gemini-2.5-pro",
+  "gemini-2.0-flash",
+  "gemini-2.0-flash-lite",
+  "gemini-2.5-flash-lite",
+  "gemini-1.5-flash",
+  "gemini-1.5-pro",
+]);
+
+export const zExtractModels = z.enum([
+  "openai-gpt-4-1",
+  "openai-gpt-4-1-mini",
+  "openai-gpt-4-1-nano",
+  "openai-gpt-5",
+  "openai-gpt-5-mini",
+  "gemini-2.0-flash",
+  "gemini-2.5-flash",
+  "gemini-2.5-flash-lite",
+  "gemini-2.5-pro",
+  "openai-gpt-4o",
+  "openai-gpt-4o-mini",
+]);
+
+export const zExtractConfig = z.object({
+  priority: z
+    .union([z.enum(["low", "medium", "high", "critical"]), z.null()])
+    .optional(),
+  extraction_target: zExtractTarget.optional(),
+  extraction_mode: zExtractMode.optional(),
+  parse_model: z.union([zPublicModelName, z.null()]).optional(),
+  extract_model: z.union([zExtractModels, z.null()]).optional(),
+  multimodal_fast_mode: z.boolean().optional().default(false),
+  system_prompt: z.union([z.string(), z.null()]).optional(),
+  use_reasoning: z.boolean().optional().default(false),
+  cite_sources: z.boolean().optional().default(false),
+  confidence_scores: z.boolean().optional().default(false),
+  chunk_mode: zDocumentChunkMode.optional(),
+  high_resolution_mode: z.boolean().optional().default(false),
+  invalidate_cache: z.boolean().optional().default(false),
+  num_pages_context: z.union([z.number().int().gte(1), z.null()]).optional(),
+  page_range: z.union([z.string(), z.null()]).optional(),
+});
+
 export const zExtractAgent = z.object({
  id: z.string().uuid(),
  name: z.string(),
@@ -2167,6 +2134,21 @@ export const zExtractJob = z.object({
  file: zFile,
 });

+export const zExtractJobCreate = z.object({
+  priority: z
+    .union([z.enum(["low", "medium", "high", "critical"]), z.null()])
+    .optional(),
+  webhook_configurations: z
+    .union([z.array(zWebhookConfiguration), z.null()])
+    .optional(),
+  extraction_agent_id: z.string().uuid(),
+  file_id: z.string().uuid(),
+  data_schema_override: z
+    .union([z.object({}), z.string(), z.null()])
+    .optional(),
+  config_override: z.union([zExtractConfig, z.null()]).optional(),
+});
+
 export const zExtractJobCreateBatch = z.object({
  extraction_agent_id: z.string().uuid(),
  file_ids: z.array(z.string().uuid()).min(1),
@@ -2234,6 +2216,15 @@ export const zExtractStatelessRequest = z.object({
  file: z.union([zFileData, z.null()]).optional(),
 });

+export const zExtractedTable = z.object({
+  table_id: z.number().int(),
+  sheet_name: z.string(),
+  row_span: z.number().int(),
+  col_span: z.number().int(),
+  has_headers: z.boolean(),
+  metadata_json: z.union([z.string(), z.null()]).optional(),
+});
+
 export const zFileCountByStatusResponse = z.object({
  counts: z.object({}),
  total_count: z.number().int(),
@@ -2987,6 +2978,30 @@ export const zPaginatedResponseQuotaConfiguration = z.object({
  items: z.array(zQuotaConfiguration),
 });

+export const zSpreadsheetParsingConfig = z.object({
+  sheet_names: z.union([z.array(z.string()), z.null()]).optional(),
+});
+
+export const zSpreadsheetJob = z.object({
+  id: z.string(),
+  user_id: z.string(),
+  project_id: z.string().uuid(),
+  file_id: z.string().uuid(),
+  config: zSpreadsheetParsingConfig,
+  status: zStatusEnum,
+  created_at: z.string(),
+  updated_at: z.string(),
+  success: z.union([z.boolean(), z.null()]).optional(),
+  tables: z.array(zExtractedTable).optional(),
+  errors: z.array(z.string()).optional(),
+});
+
+export const zPaginatedResponseSpreadsheetJob = z.object({
+  items: z.array(zSpreadsheetJob),
+  next_page_token: z.union([z.string(), z.null()]).optional(),
+  total_size: z.union([z.number().int(), z.null()]).optional(),
+});
+
 export const zParseConfiguration = z.object({
  id: z.string(),
  name: z.string(),
@@ -3400,49 +3415,6 @@ export const zProjectUpdate = z.object({
  name: z.string().min(1).max(3000),
 });

-export const zPromptConf = z.object({
-  system_prompt: z
-    .string()
-    .optional()
-    .default(
-      "Given a JSON schema, extract the data from the provided SOURCE TEXT according to the schema. Only output information that is explicitly stated or can be inferred from the SOURCE TEXT.",
-    ),
-  extraction_prompt: z
-    .string()
-    .optional()
-    .default("The extracted data using the given JSON schema."),
-  error_handling_prompt: z
-    .string()
-    .optional()
-    .default(
-      "If the source text does not contain enough information to extract the value, explain the reason very briefly. Else, output null and fill out the value__ field.",
-    ),
-  reasoning_prompt: z.string().optional().default(`
-Provide a brief explanation for how you arrived at the extracted value based on the source text provided.
- For inferred values, explain the reasoning behind the extraction briefly.
- For simple verbatim extraction, output 'VERBATIM EXTRACTION'.
- When supporting data is not present in the source text, output 'INSUFFICIENT DATA' and emit blank or null values for the value__ field.
-`),
-  cite_sources_prompt: z
-    .object({})
-    .optional()
-    .default({
-      description: `
-### Citation Rules (read carefully):
- You must ANNOTATE every value with the MOST RELEVANT short EXACT substring from the source text that supports it.
- For inferred values, cite the text used to infer it in the matching_text field or output 'INFERRED FROM TEXT'
- If no support exists, output 'INSUFFICIENT DATA' and leave value__ null or '', 0.0, False etc depending on the type of the field.
-`,
-      page: "Cite the page number of the source text that the extracted value is from. The page number is the integer that appears right after <<<PAGE:. If no page number is present in this format, use the default value of 1.",
-      matching_text:
-        'Cite the **MOST RELEVANT EXACT TEXT from the SOURCE TEXT** that supports the extracted value within 80 characters. If the exact substring is >80 chars, truncate with ellipsis "...". Provide only the single most relevant citation.',
-    }),
-  scratchpad_prompt: z
-    .string()
-    .optional()
-    .default("Use for intermediate step-by-step reasoning. Be concise."),
-});
-
 export const zRelatedNodeInfo = z.object({
  node_id: z.string(),
  node_type: z.union([zObjectType, z.string(), z.null()]).optional(),
@@ -3545,8 +3517,6 @@ export const zRole = z.object({
  permissions: z.array(zPermission),
 });

-export const zSchemaRelaxMode = z.enum(["FULL", "TOP_LEVEL", "LEAF"]);
-
 export const zSearchRequest = z.object({
  page_size: z.union([z.number().int(), z.null()]).optional(),
  page_token: z.union([z.string(), z.null()]).optional(),
@@ -3558,24 +3528,9 @@ export const zSearchRequest = z.object({
  offset: z.union([z.number().int().gte(0).lte(1000), z.null()]).optional(),
 });

-export const zStructMode = z.enum([
-  "STRUCT_PARSE",
-  "JSON_MODE",
-  "FUNC_CALL",
-  "STRUCT_RELAXED",
-  "UNSTRUCTURED",
-]);
-
-export const zStructParseConf = z.object({
-  model: zExtractModels.optional(),
-  temperature: z.number().optional().default(0),
-  relaxation_mode: zSchemaRelaxMode.optional(),
-  struct_mode: zStructMode.optional(),
-  fetch_logprobs: z.boolean().optional().default(false),
-  handle_missing: z.boolean().optional().default(false),
-  use_reasoning: z.boolean().optional().default(false),
-  cite_sources: z.boolean().optional().default(false),
-  prompt_conf: zPromptConf.optional(),
+export const zSpreadsheetJobCreate = z.object({
+  file_id: z.string().uuid(),
+  config: zSpreadsheetParsingConfig.optional(),
 });

 export const zSupportedLlmModel = z.object({
@@ -4017,6 +3972,33 @@ export const zCreateIntentAndCustomerSessionApiV1BillingCreateIntentAndCustomerS
 export const zGetMetronomeDashboardApiV1BillingMetronomeDashboardGetResponse =
  zMetronomeDashboardResponse;

+export const zListJobsApiV1ExtractionJobsGetResponse = z.array(zExtractJob);
+
+export const zRunJobApiV1ExtractionJobsPostResponse = zExtractJob;
+
+export const zGetJobApiV1ExtractionJobsJobIdGetResponse = zExtractJob;
+
+export const zRunJobOnFileApiV1ExtractionJobsFilePostResponse = zExtractJob;
+
+export const zRunBatchJobsApiV1ExtractionJobsBatchPostResponse =
+  z.array(zExtractJob);
+
+export const zGetJobResultApiV1ExtractionJobsJobIdResultGetResponse =
+  zExtractResultset;
+
+export const zListExtractRunsApiV1ExtractionRunsGetResponse =
+  zPaginatedExtractRunsResponse;
+
+export const zGetLatestRunFromUiApiV1ExtractionRunsLatestFromUiGetResponse =
+  z.union([zExtractRun, z.null()]);
+
+export const zGetRunByJobIdApiV1ExtractionRunsByJobJobIdGetResponse =
+  zExtractRun;
+
+export const zGetRunApiV1ExtractionRunsRunIdGetResponse = zExtractRun;
+
+export const zExtractStatelessApiV1ExtractionRunPostResponse = zExtractJob;
+
 export const zListExtractionAgentsApiV1ExtractionExtractionAgentsGetResponse =
  z.array(zExtractAgent);

@@ -4041,35 +4023,6 @@ export const zGetExtractionAgentApiV1ExtractionExtractionAgentsExtractionAgentId
 export const zUpdateExtractionAgentApiV1ExtractionExtractionAgentsExtractionAgentIdPutResponse =
  zExtractAgent;

-export const zListJobsApiV1ExtractionJobsGetResponse = z.array(zExtractJob);
-
-export const zRunJobApiV1ExtractionJobsPostResponse = zExtractJob;
-
-export const zGetJobApiV1ExtractionJobsJobIdGetResponse = zExtractJob;
-
-export const zRunJobTestUserApiV1ExtractionJobsTestPostResponse = zExtractJob;
-
-export const zRunJobOnFileApiV1ExtractionJobsFilePostResponse = zExtractJob;
-
-export const zRunBatchJobsApiV1ExtractionJobsBatchPostResponse =
-  z.array(zExtractJob);
-
-export const zGetJobResultApiV1ExtractionJobsJobIdResultGetResponse =
-  zExtractResultset;
-
-export const zListExtractRunsApiV1ExtractionRunsGetResponse =
-  zPaginatedExtractRunsResponse;
-
-export const zGetLatestRunFromUiApiV1ExtractionRunsLatestFromUiGetResponse =
-  z.union([zExtractRun, z.null()]);
-
-export const zGetRunByJobIdApiV1ExtractionRunsByJobJobIdGetResponse =
-  zExtractRun;
-
-export const zGetRunApiV1ExtractionRunsRunIdGetResponse = zExtractRun;
-
-export const zExtractStatelessApiV1ExtractionRunPostResponse = zExtractJob;
-
 export const zListApiKeysApiV1BetaApiKeysGetResponse = zApiKeyQueryResponse;

 export const zCreateApiKeyApiV1BetaApiKeysPostResponse = zApiKey;
@@ -4100,6 +4053,9 @@ export const zSearchAgentDataApiV1BetaAgentDataSearchPostResponse =
 export const zAggregateAgentDataApiV1BetaAgentDataAggregatePostResponse =
  zPaginatedResponseAggregateGroup;

+export const zDeleteAgentDataByQueryApiV1BetaAgentDataDeletePostResponse =
+  zDeleteResponse;
+
 export const zListQuotaConfigurationsApiV1BetaQuotaManagementGetResponse =
  zPaginatedResponseQuotaConfiguration;

@@ -4135,6 +4091,18 @@ export const zQueryParseConfigurationsApiV1BetaParseConfigurationsQueryPostRespo
 export const zGetLatestParseConfigurationApiV1BetaParseConfigurationsLatestGetResponse =
  z.union([zParseConfiguration, z.null()]);

+export const zListSpreadsheetJobsApiV1BetaSpreadsheetJobsGetResponse =
+  zPaginatedResponseSpreadsheetJob;
+
+export const zCreateSpreadsheetJobApiV1BetaSpreadsheetJobsPostResponse =
+  zSpreadsheetJob;
+
+export const zGetSpreadsheetJobApiV1BetaSpreadsheetJobsSpreadsheetJobIdGetResponse =
+  zSpreadsheetJob;
+
+export const zGetTableDownloadPresignedUrlApiV1BetaSpreadsheetJobsSpreadsheetJobIdTablesTableIdResultGetResponse =
+  zPresignedUrl;
+
 export const zUploadFileV2ApiV2Alpha1ParseUploadPostResponse = zParsingJob;

 export const zGetSupportedFileExtensionsApiParsingSupportedFileExtensionsGetResponse =
@@ -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
Logan Markewich	40a83ebd15	loosen packaging requirements	2025-10-06 09:05:34 -06:00
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