fix: explicitly tag. I thought the action did this

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

.changeset/README.md

@@ -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)

.changeset/brave-moments-type.md

@@ -0,0 +1,5 @@
+---
+"llama-cloud-services-py": minor
+---
+
+Escaping dollar signs in markdown output in jupyter notebooks to prevent them being interpreted as equation delimiters

.changeset/config.json

@@ -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": []
+}

.github/workflows/release_py.yml

@@ -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

.github/workflows/release_ts.yml

@@ -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 }} - LlamaCloud Services TS
-          bodyFile: "ts/llama_cloud_services/CHANGELOG.md"
-          token: ${{ secrets.GITHUB_TOKEN }}

.github/workflows/version_bump_and_release.yml

@@ -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@v4
+        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 }}

.gitignore

@@ -9,3 +9,4 @@ __pycache__/
 node_modules/
 .turbo/
 dist/
+.npmrc

.pre-commit-config.yaml

@@ -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:

README.md

@@ -9,7 +9,6 @@ This repository contains the code for hand-written SDKs and clients for interact
 This includes:
 
 - [LlamaParse](./parse.md) - A GenAI-native document parser that can parse complex document data for any downstream LLM use case (Agents, RAG, data processing, etc.).
-- [LlamaReport (beta/invite-only)](./report.md) - A prebuilt agentic report builder that can be used to build reports from a variety of data sources.
 - [LlamaExtract](./extract.md) - A prebuilt agentic data extractor that can be used to transform data into a structured JSON representation.
 - [LlamaCloud Index](./index.md) - A widely customizable and fully automated document ingestion pipeline that also serves retrieval purposes.
 
@@ -28,13 +27,11 @@ Then, you can use the services in your code:
 ```python
 from llama_cloud_services import (
     LlamaParse,
-    LlamaReport,
     LlamaExtract,
     LlamaCloudIndex,
 )
 
 parser = LlamaParse(api_key="YOUR_API_KEY")
-report = LlamaReport(api_key="YOUR_API_KEY")
 extract = LlamaExtract(api_key="YOUR_API_KEY")
 index = LlamaCloudIndex(
     "my_first_index", project_name="default", api_key="YOUR_API_KEY"
@@ -44,7 +41,6 @@ index = LlamaCloudIndex(
 See the quickstart guides for each service for more information:
 
 - [LlamaParse](./parse.md)
-- [LlamaReport (beta/invite-only)](./report.md)
 - [LlamaExtract](./extract.md)
 - [LlamaCloud Index](./index.md)
 
@@ -57,13 +53,11 @@ You can also create your API key in the EU region [here](https://cloud.eu.llamai
 ```python
 from llama_cloud_services import (
     LlamaParse,
-    LlamaReport,
     LlamaExtract,
     EU_BASE_URL,
 )
 
 parser = LlamaParse(api_key="YOUR_API_KEY", base_url=EU_BASE_URL)
-report = LlamaReport(api_key="YOUR_API_KEY", base_url=EU_BASE_URL)
 extract = LlamaExtract(api_key="YOUR_API_KEY", base_url=EU_BASE_URL)
 index = LlamaCloudIndex(
     "my_first_index",

examples/README.md

@@ -4,7 +4,6 @@ In this folder you will find several python notebooks that contain examples rega
 
 - [LlamaParse](./parse/)
 - [LlamaExtract](./extract/)
-- [LlamaReport](./report/)
 - [LlamaCloudIndex](./index/)
 
 Follow the instructions in each notebook to get started!

examples/extract/asset_manager_fund_analysis.ipynb

@@ -7,7 +7,7 @@
    "source": [
     "# Extraction and Analysis over a Fidelity Multi-Fund Annual Report\n",
     "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services-demo/blob/main/examples/extract/asset_manager_fund_analysis.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/extract/asset_manager_fund_analysis.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
     "\n",
     "In this notebook we show you how to create an agentic document workflow over a complex document that contains annual reports for multiple funds - each fund reports financials in a standardized reporting structure, and it's all consolidated in the same document.\n",
     "\n",

examples/extract/automotive_sector_analysis.ipynb

@@ -7,7 +7,7 @@
    "source": [
     "# Automotive Equity Research: A Multi-Step Agentic Workflow\n",
     "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services-demo/blob/main/examples/extract/automotive_sector_analysis.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/extract/automotive_sector_analysis.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
     "\n",
     "This notebook demonstrates an end‑to‑end agentic workflow using LlamaExtract and the LlamaIndex event‑driven workflow framework for automotive sector analysis.\n",
     "\n",

examples/index/getting_started.ipynb

@@ -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
 }

examples/misc/parse_classify_extract_workflow.ipynb

@@ -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
+}

examples/parse/advanced_rag/dynamic_section_retrieval.ipynb

@@ -7,7 +7,7 @@
    "source": [
     "# Dynamic Section Retrieval with LlamaParse\n",
     "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services-demo/blob/main/examples/parse/advanced_rag/dynamic_section_retrieval.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/advanced_rag/dynamic_section_retrieval.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
     "\n",
     "This notebook showcases a concept called \"dynamic section retrieval\".\n",
     "\n",

examples/parse/demo_advanced.ipynb

@@ -6,7 +6,7 @@
    "source": [
     "# Advanced RAG with LlamaParse\n",
     "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_parse/blob/main/examples/demo_advanced.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_parse/blob/main/examples/parse/demo_advanced.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
     "\n",
     "This notebook is a complete walkthrough for using LlamaParse with advanced indexing/retrieval techniques in LlamaIndex over the Apple 10K Filing. \n",
     "\n",

examples/parse/demo_excel.ipynb

@@ -6,7 +6,7 @@
    "source": [
     "# RAG with Excel Spreadsheet using LlamaPrase\n",
     "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/demo_excel.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/demo_excel.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
     "\n",
     "This notebook shows you using LlamaParse with Excel Spreadsheet.\n",
     "\n",

examples/parse/demo_get_charts.ipynb

@@ -7,7 +7,7 @@
    "source": [
     "# Download Charts\n",
     "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/demo_get_charts.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/demo_get_charts.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
     "\n",
     "This notebook demonstrates how to download charts from a document using the result object.\n",
     "\n",

examples/parse/demo_insurance.ipynb

@@ -6,7 +6,7 @@
    "source": [
     "# LlamaParse - Fast checking Insurance Contract for Coverage\n",
     "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/demo_insurance.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/demo_insurance.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
     "\n",
     "In this notebook we will look at how LlamaParse can be used to extract structured coverage information from an insurance policy.\n",
     "\n",
@@ -36,7 +36,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## Download an insurance policy fron IRDAI\n",
+    "## Download an insurance policy from IRDAI\n",
     "\n",
     "The Insurance Regulatory and Development Authority of India (IRDAI) maintains a great resource: https://policyholder.gov.in/web/guest/non-life-insurance-products where all insurance policies available in India are publicly available for download! Let's download a complex health insurance policy as an example."
    ]
@@ -228,11 +228,11 @@
     "    result_type=\"markdown\",\n",
     "    system_prompt_append=\"\"\"\n",
     "This document is an insurance policy.\n",
-    "When a benefits/coverage/exlusion is describe in the document ammend to it add a text in the follwing benefits string format (where coverage could be an exclusion).\n",
+    "When a benefits/coverage/exlusion is describe in the document amend to it add a text in the following benefits string format (where coverage could be an exclusion).\n",
     "\n",
     "For {nameofrisk} and in this condition {whenDoesThecoverageApply} the coverage is {coverageDescription}. \n",
     "                                        \n",
-    "If the document contain a benefits TABLE that describe coverage amounts, do not ouput it as a table, but instead as a list of benefits string.\n",
+    "If the document contain a benefits TABLE that describe coverage amounts, do not output it as a table, but instead as a list of benefits string.\n",
     "                                       \n",
     "\"\"\",\n",
     ").aparse(\"./policy.pdf\")\n",

examples/parse/demo_json_tour.ipynb

@@ -7,7 +7,7 @@
    "source": [
     "# LlamaParse `JobResult` Tour\n",
     "\n",
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/demo_json.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/demo_json.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
     "\n",
     "The `JobResult` object is the main object returned by the LlamaParse API. It contains all the information about the job, including the parsed data, metadata, and any errors.\n",
     "\n",

examples/parse/demo_languages.ipynb

@@ -9,7 +9,7 @@
     "\n",
     "LlamaParse supports users to specify a `language` parameter before uploading documents, giving users better OCR capabilities over non-English PDFs, parsing images into more accurate representations.\n",
     "\n",
-    "You can specify 80+ different languages: see this file for a full list of supported languages: https://github.com/run-llama/llama_cloud_services/blob/main/llama_parse/base.py.\n",
+    "You can specify 80+ different languages: see this file for a full list of supported languages: https://github.com/run-llama/llama_cloud_services/blob/main/py/llama_cloud_services/parse/base.py.\n",
     "\n",
     "This notebook shows a demo of this in action. \n",
     "\n",

examples/parse/excel/o1_excel_rag.ipynb

@@ -4,7 +4,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/excel/o1_excel_rag.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
+    "<a href=\"https://colab.research.google.com/github/run-llama/llama_cloud_services/blob/main/examples/parse/excel/o1_excel_rag.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
    ]
   },
   {

examples/parse/parsing_modes/demo_auto_mode.ipynb

@@ -740,7 +740,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "In this example, these pages aren't going to be that different when parsed, but we can verify which pages triggered auto-made by looking at the [JSON output](https://github.com/run-llama/llama_cloud_services/blob/main/examples/demo_json_tour.ipynb) of LlamaParse:"
+    "In this example, these pages aren't going to be that different when parsed, but we can verify which pages triggered auto-made by looking at the [JSON output](https://github.com/run-llama/llama_cloud_services/blob/main/examples/parse/demo_json_tour.ipynb) of LlamaParse:"
    ]
   },
   {

examples/report/basic_report.ipynb

@@ -1,762 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "# Report Generation with LlamaReport\n",
-    "\n",
-    "In this notebook, we'll walk through the basic process of generating a report with LlamaReport, and highlight some of the key features of the library.\n",
-    "\n",
-    "TLDR:\n",
-    "1. Download source data to use as knowledge base for the report\n",
-    "2. Kick off report generation with a template\n",
-    "3. Get the plan and review/accept/reject suggestions\n",
-    "4. Get the final report\n",
-    "5. Review/accept/reject suggestions to edit the final report\n",
-    "6. Print the final report"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "%pip install llama-cloud-services"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "## 1. Download Source Data\n",
-    "\n",
-    "Here, we download the `Attention is All You Need` paper as a PDF.\n",
-    "\n",
-    "LlamaReport currently supports up to 5 files as input, and essentially any file type that can be parsed by LlamaParse.\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "!wget \"https://arxiv.org/pdf/1706.03762.pdf\" -O \"./attention.pdf\""
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "## 2. Kick off Report Generation\n",
-    "\n",
-    "Here, we kick off report generation with a template.\n",
-    "\n",
-    "The template can either be a string or a file path, but here we'll use a string.\n",
-    "\n",
-    "In our experiments, anything works as a template, but some general guidelines:\n",
-    "\n",
-    "- Use markdown formatting + instructions in each section to guide the report generation\n",
-    "- If using an existing file as a template, provide extra instructions to guide the report generation\n",
-    "\n",
-    "**NOTE:** Since we are in a notebook, we will use async functions and `await` throughout. Synchronous methods that work without `await` are available by just removing the `a` from the method name and removing the `await` keyword."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "from llama_cloud_services import LlamaReport\n",
-    "\n",
-    "llama_report = LlamaReport(\n",
-    "    api_key=\"llx-...\",\n",
-    ")\n",
-    "\n",
-    "report_client = await llama_report.acreate_report(\n",
-    "    name=\"my_cool_report_on_attention\",\n",
-    "    # can pass in file paths or bytes\n",
-    "    input_files=[\"./attention.pdf\"],\n",
-    "    template_text=\"\"\"\\\n",
-    "# [Some title]\\n\\n\n",
-    "## TLDR\\n\n",
-    "A quick summary of the paper.\\n\\n\n",
-    "## Details\\n\n",
-    "More details about the paper, possibly more than one section here.\\n\n",
-    "\"\"\",\n",
-    "    # optional additional instructions for the report generation\n",
-    "    # template_instructions=None,\n",
-    "    # optional file path to an existing template instead of template_text\n",
-    "    # template_file=None,\n",
-    ")"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "The returned `ReportClient` object is used to interact with the report generation process for this specific report."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Report(id=0a394b33-1a3e-463c-b5cb-7ff8ab827d0a, name=my_cool_report_on_attention)\n"
-     ]
-    }
-   ],
-   "source": [
-    "print(report_client)"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "## 3. Get the plan\n",
-    "\n",
-    "The first phases of report generation involve ingesting the source data and generating a plan.\n",
-    "\n",
-    "The plan is a list of instructions for the report generation, and can be reviewed/accepted/rejected by the user.\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "plan = await report_client.await_for_plan(\n",
-    "    timeout=10000,\n",
-    "    poll_interval=10,\n",
-    ")"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "# {title}\n",
-      "[ReportQuery(field='title', prompt='Generate a clear and concise title for this paper about the Transformer model and attention mechanisms', context='The paper discusses the Transformer architecture for sequence transduction using attention mechanisms, focusing on machine translation applications')]\n",
-      "==================\n",
-      "## TLDR\n",
-      "\n",
-      "{tldr_content}\n",
-      "[ReportQuery(field='tldr_content', prompt='Write a brief, clear summary of the key points about the Transformer model', context='Focus on the main innovations: attention mechanisms, efficiency improvements, and state-of-the-art results in machine translation')]\n",
-      "==================\n",
-      "## Details\n",
-      "\n",
-      "{details_content}\n",
-      "[ReportQuery(field='details_content', prompt='Provide detailed information about the Transformer model architecture and its applications', context='Include information about:\\n- The attention mechanism implementation\\n- Advantages over recurrent and convolutional models\\n- Performance in machine translation tasks\\n- Training efficiency improvements')]\n",
-      "==================\n"
-     ]
-    }
-   ],
-   "source": [
-    "for plan_block in plan.blocks:\n",
-    "    print(plan_block.block.template)\n",
-    "    print(plan_block.queries)\n",
-    "    print(\"==================\")"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "With the plan, we can either use it to kick off generation of the final report, or we can edit the plan and adjust it as needed.\n",
-    "\n",
-    "While we could manually edit the objects here and use `await report_client.aupdate_plan(action=\"edit\", updated_plan=plan)`, we can also use `LlamaReport` to agentically edit the plan."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "suggestions = await report_client.asuggest_edits(\n",
-    "    \"Can you split the details section into two sections?\"\n",
-    ")"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Justification for change: \n",
-      "I'll help you break down the details section into two distinct parts - one focusing on the architecture and another on the practical applications and performance. This will make the content more organized and easier to follow. The original block at index 2 will be replaced with these two new sections.\n",
-      "\n",
-      "Proposed changes:\n",
-      "\n",
-      "## Architecture Details\n",
-      "\n",
-      "{architecture_content}\n",
-      "\n",
-      "[ReportQuery(field='architecture_content', prompt='Describe the technical details of the Transformer model architecture', context='Focus on:\\n- Core components of the Transformer architecture\\n- Self-attention mechanism implementation\\n- Multi-head attention details\\n- Position encoding approach\\n- Feed-forward network structure')]\n",
-      "==================\n",
-      "\n",
-      "## Performance and Applications\n",
-      "\n",
-      "{applications_content}\n",
-      "\n",
-      "[ReportQuery(field='applications_content', prompt='Explain the practical applications and performance advantages of the Transformer model', context='Cover:\\n- Comparison with RNN and CNN models\\n- Machine translation results and benchmarks\\n- Training efficiency improvements\\n- Real-world applications and use cases\\n- Scalability benefits')]\n",
-      "==================\n"
-     ]
-    }
-   ],
-   "source": [
-    "for suggestion in suggestions:\n",
-    "    print(\"Justification for change:\", suggestion.justification)\n",
-    "    print(\"Proposed changes:\")\n",
-    "    for plan_block in suggestion.blocks:\n",
-    "        print(plan_block.block.template)\n",
-    "        print(plan_block.queries)\n",
-    "        print(\"==================\")"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "This looks pretty good! We can also use the client to automatically accept and apply, or reject, these suggestions.\n",
-    "\n",
-    "This will (locally) keep track of the history of changes, so that future suggestions can be based on the previous changes."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "for suggestion in suggestions:\n",
-    "    await report_client.aaccept_edit(suggestion)"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "What effect did that have on the tracked local history? Let's see!"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "[EditAction(block_idx=2, old_content='## Details\\n\\n{details_content}\\n\\nField: details_content, Prompt: Provide detailed information about the Transformer model architecture and its applications, Context: Include information about:\\n- The attention mechanism implementation\\n- Advantages over recurrent and convolutional models\\n- Performance in machine translation tasks\\n- Training efficiency improvements\\nDepends on: none', new_content='\\n## Architecture Details\\n\\n{architecture_content}\\n\\n\\nField: architecture_content, Prompt: Describe the technical details of the Transformer model architecture, Context: Focus on:\\n- Core components of the Transformer architecture\\n- Self-attention mechanism implementation\\n- Multi-head attention details\\n- Position encoding approach\\n- Feed-forward network structure\\nDepends on: none', action='approved', timestamp=datetime.datetime(2025, 2, 4, 20, 59, 55, 773558)),\n",
-       " EditAction(block_idx=3, old_content='[No old content]', new_content='\\n## Performance and Applications\\n\\n{applications_content}\\n\\n\\nField: applications_content, Prompt: Explain the practical applications and performance advantages of the Transformer model, Context: Cover:\\n- Comparison with RNN and CNN models\\n- Machine translation results and benchmarks\\n- Training efficiency improvements\\n- Real-world applications and use cases\\n- Scalability benefits\\nDepends on: previous', action='approved', timestamp=datetime.datetime(2025, 2, 4, 20, 59, 55, 773687))]"
-      ]
-     },
-     "execution_count": null,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "report_client.edit_history"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "[Message(role=<MessageRole.USER: 'user'>, content='Can you split the details section into two sections?', timestamp=datetime.datetime(2025, 2, 4, 20, 59, 47, 754848)),\n",
-       " Message(role=<MessageRole.ASSISTANT: 'assistant'>, content=\"\\nI'll help you break down the details section into two distinct parts - one focusing on the architecture and another on the practical applications and performance. This will make the content more organized and easier to follow. The original block at index 2 will be replaced with these two new sections.\\n\", timestamp=datetime.datetime(2025, 2, 4, 20, 59, 55, 482070))]"
-      ]
-     },
-     "execution_count": null,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "report_client.chat_history"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "These two items are used to provide context for future suggestions! You can always clear this, or provide your own history."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# report_client.suggest_edits(\"....\", chat_history=[{\"role\": \"user\", \"content\": \"...\"}, ...])"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "## 4. Get the final report\n",
-    "\n",
-    "Now that we have a plan, we can kick off generation of the final report."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# kicks off report generation\n",
-    "await report_client.aupdate_plan(action=\"approve\")\n",
-    "\n",
-    "# waits for report generation to complete\n",
-    "report = await report_client.await_completion(\n",
-    "    timeout=10000,\n",
-    "    poll_interval=10,\n",
-    ")"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "# Attention Is All You Need: A Pure Attention-Based Architecture for Neural Machine Translation\n",
-      "\n",
-      "## TLDR\n",
-      "\n",
-      "The Transformer introduced a revolutionary architecture that relies entirely on attention mechanisms, eliminating the need for recurrence or convolution in sequence processing. Its key innovations include multi-head self-attention for parallel processing of input sequences, scaled dot-product attention for efficient computation, and positional encodings for sequence order awareness. The model achieved breakthrough results in machine translation (28.4 BLEU on English-to-German, 41.8 BLEU on English-to-French) while requiring significantly less training time than previous approaches, training in 3.5 days on 8 GPUs. This architecture demonstrated that attention mechanisms alone are sufficient for state-of-the-art sequence modeling, setting a new direction for natural language processing.\n",
-      "\n",
-      "\n",
-      "## Architecture Details\n",
-      "\n",
-      "The Transformer architecture represents a groundbreaking approach to sequence processing, built entirely on attention mechanisms without recurrence or convolution. Here are its key technical details:\n",
-      "\n",
-      "Core Components:\n",
-      "- Encoder-decoder architecture with stacked self-attention and point-wise feed-forward layers\n",
-      "- Each layer contains two main sub-layers: multi-head self-attention mechanism and position-wise feed-forward network\n",
-      "- Layer normalization and residual connections between sub-layers\n",
-      "- No recurrent or convolutional elements, enabling parallel processing\n",
-      "\n",
-      "Self-Attention Mechanism:\n",
-      "- Processes relationships between all positions in a sequence simultaneously\n",
-      "- Computes attention weights using queries, keys, and values derived from input representations\n",
-      "- Implements scaled dot-product attention to prevent gradient issues with large input dimensions\n",
-      "- Allows direct modeling of dependencies regardless of positional distance\n",
-      "- Uses masking in decoder to prevent leftward information flow and maintain auto-regressive property\n",
-      "\n",
-      "Multi-Head Attention:\n",
-      "- Employs multiple attention heads operating in parallel\n",
-      "- Each head processes information in different representation subspaces\n",
-      "- Three types of attention applications:\n",
-      "  1. Encoder self-attention (all positions attend to each other)\n",
-      "  2. Decoder self-attention (each position attends to previous positions)\n",
-      "  3. Encoder-decoder attention (decoder queries attend to encoder outputs)\n",
-      "- Counteracts reduced resolution from attention averaging through parallel processing\n",
-      "\n",
-      "Position-wise Feed-Forward Network:\n",
-      "- Applied identically to each position separately\n",
-      "- Consists of two linear transformations with ReLU activation\n",
-      "- Structure: FFN(x) = max(0, xW1 + b1)W2 + b2\n",
-      "- Input and output dimensionality: dmodel = 512\n",
-      "- Inner-layer dimensionality: dff = 2048\n",
-      "- Parameters vary between layers but remain constant across positions\n",
-      "\n",
-      "Position Encoding:\n",
-      "- Adds positional information to input embeddings\n",
-      "- Enables the model to consider sequential order without recurrence\n",
-      "- Implements sinusoidal position encodings to allow model to attend to relative positions\n",
-      "- Maintains constant number of operations between any two positions, unlike convolutional approaches\n",
-      "- Allows effective modeling of both local and long-range dependencies\n",
-      "\n",
-      "\n",
-      "\n",
-      "## Performance and Applications\n",
-      "\n",
-      "The Transformer model demonstrates significant performance advantages and practical applications across multiple domains:\n",
-      "\n",
-      "Performance Advantages over RNN/CNN Models:\n",
-      "- Eliminates sequential computation constraints present in RNNs, enabling superior parallelization\n",
-      "- Reduces operations needed for relating distant positions to a constant number, compared to linear/logarithmic scaling in CNNs\n",
-      "- Processes all input and output positions simultaneously through self-attention mechanisms\n",
-      "- Achieves state-of-the-art results while requiring significantly less computational resources\n",
-      "\n",
-      "Machine Translation Benchmarks:\n",
-      "- WMT 2014 English-to-German: 28.4 BLEU score, exceeding previous best results by over 2 BLEU points\n",
-      "- WMT 2014 English-to-French: 41.8 BLEU score (single-model state-of-the-art)\n",
-      "- Surpasses performance of existing model ensembles in translation tasks\n",
-      "\n",
-      "Training Efficiency:\n",
-      "- Requires only 3.5 days of training on eight GPUs for state-of-the-art performance\n",
-      "- Achieves superior results at \"a small fraction of the training costs\" compared to previous models\n",
-      "- Enables significantly faster training through parallel processing of input/output sequences\n",
-      "- Can reach production-quality performance in as little as twelve hours on modern GPU hardware\n",
-      "\n",
-      "Real-world Applications:\n",
-      "- Machine translation systems\n",
-      "- Natural language understanding tasks\n",
-      "- Reading comprehension\n",
-      "- Abstractive summarization\n",
-      "- Text entailment analysis\n",
-      "- Constituency parsing (achieving 92.7 F1 score in semi-supervised settings)\n",
-      "- Adaptable to both large and limited training data scenarios\n",
-      "\n",
-      "Scalability Benefits:\n",
-      "- Highly parallelizable architecture enables efficient scaling across multiple GPUs\n",
-      "- Constant computational complexity for relating any input/output positions\n",
-      "- Effective handling of long-range dependencies in sequences\n",
-      "- Maintains performance quality while scaling to larger datasets and model sizes\n",
-      "- Generalizes well across different tasks and domains without architectural changes\n",
-      "- Supports efficient inference and deployment in production environments\n",
-      "\n"
-     ]
-    }
-   ],
-   "source": [
-    "report_text = \"\\n\\n\".join([block.template for block in report.blocks])\n",
-    "print(report_text)"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "## 5. Edit the final report\n",
-    "\n",
-    "Now that we have a report, we can edit it.\n",
-    "\n",
-    "We can use the `asuggest_edits` method to get suggestions for edits, and then use the `aaccept_edit`/`areject_edit` methods to apply them.\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "Justification for change: \n",
-      "I'd suggest changing \"TLDR\" to \"Executive Summary\" which is more appropriate for a professional or academic report. This term is widely used in formal documents and better reflects the nature of this concise overview section while maintaining the same function of providing a quick summary of the key points.\n",
-      "\n",
-      "Proposed changes:\n",
-      "## Executive Summary\n",
-      "\n",
-      "The Transformer introduced a revolutionary architecture that relies entirely on attention mechanisms, eliminating the need for recurrence or convolution in sequence processing. Its key innovations include multi-head self-attention for parallel processing of input sequences, scaled dot-product attention for efficient computation, and positional encodings for sequence order awareness. The model achieved breakthrough results in machine translation (28.4 BLEU on English-to-German, 41.8 BLEU on English-to-French) while requiring significantly less training time than previous approaches, training in 3.5 days on 8 GPUs. This architecture demonstrated that attention mechanisms alone are sufficient for state-of-the-art sequence modeling, setting a new direction for natural language processing.\n",
-      "==================\n"
-     ]
-    }
-   ],
-   "source": [
-    "suggestions = await report_client.asuggest_edits(\n",
-    "    \"Can you change the TLDR header to something more professional?\"\n",
-    ")\n",
-    "for suggestion in suggestions:\n",
-    "    print(\"Justification for change:\", suggestion.justification)\n",
-    "    print(\"Proposed changes:\")\n",
-    "    for block in suggestion.blocks:\n",
-    "        print(block.template)\n",
-    "        print(\"==================\")"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "Changing to \"Executive Summary\" sounds reasonable, lets accept that!\n"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "for suggestion in suggestions:\n",
-    "    await report_client.aaccept_edit(suggestion)"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "## 7. Print the final report\n",
-    "\n",
-    "Now that we have a report, we can print it."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "# Attention Is All You Need: A Pure Attention-Based Architecture for Neural Machine Translation\n",
-      "\n",
-      "## Executive Summary\n",
-      "\n",
-      "The Transformer introduced a revolutionary architecture that relies entirely on attention mechanisms, eliminating the need for recurrence or convolution in sequence processing. Its key innovations include multi-head self-attention for parallel processing of input sequences, scaled dot-product attention for efficient computation, and positional encodings for sequence order awareness. The model achieved breakthrough results in machine translation (28.4 BLEU on English-to-German, 41.8 BLEU on English-to-French) while requiring significantly less training time than previous approaches, training in 3.5 days on 8 GPUs. This architecture demonstrated that attention mechanisms alone are sufficient for state-of-the-art sequence modeling, setting a new direction for natural language processing.\n",
-      "\n",
-      "\n",
-      "## Architecture Details\n",
-      "\n",
-      "The Transformer architecture represents a groundbreaking approach to sequence processing, built entirely on attention mechanisms without recurrence or convolution. Here are its key technical details:\n",
-      "\n",
-      "Core Components:\n",
-      "- Encoder-decoder architecture with stacked self-attention and point-wise feed-forward layers\n",
-      "- Each layer contains two main sub-layers: multi-head self-attention mechanism and position-wise feed-forward network\n",
-      "- Layer normalization and residual connections between sub-layers\n",
-      "- No recurrent or convolutional elements, enabling parallel processing\n",
-      "\n",
-      "Self-Attention Mechanism:\n",
-      "- Processes relationships between all positions in a sequence simultaneously\n",
-      "- Computes attention weights using queries, keys, and values derived from input representations\n",
-      "- Implements scaled dot-product attention to prevent gradient issues with large input dimensions\n",
-      "- Allows direct modeling of dependencies regardless of positional distance\n",
-      "- Uses masking in decoder to prevent leftward information flow and maintain auto-regressive property\n",
-      "\n",
-      "Multi-Head Attention:\n",
-      "- Employs multiple attention heads operating in parallel\n",
-      "- Each head processes information in different representation subspaces\n",
-      "- Three types of attention applications:\n",
-      "  1. Encoder self-attention (all positions attend to each other)\n",
-      "  2. Decoder self-attention (each position attends to previous positions)\n",
-      "  3. Encoder-decoder attention (decoder queries attend to encoder outputs)\n",
-      "- Counteracts reduced resolution from attention averaging through parallel processing\n",
-      "\n",
-      "Position-wise Feed-Forward Network:\n",
-      "- Applied identically to each position separately\n",
-      "- Consists of two linear transformations with ReLU activation\n",
-      "- Structure: FFN(x) = max(0, xW1 + b1)W2 + b2\n",
-      "- Input and output dimensionality: dmodel = 512\n",
-      "- Inner-layer dimensionality: dff = 2048\n",
-      "- Parameters vary between layers but remain constant across positions\n",
-      "\n",
-      "Position Encoding:\n",
-      "- Adds positional information to input embeddings\n",
-      "- Enables the model to consider sequential order without recurrence\n",
-      "- Implements sinusoidal position encodings to allow model to attend to relative positions\n",
-      "- Maintains constant number of operations between any two positions, unlike convolutional approaches\n",
-      "- Allows effective modeling of both local and long-range dependencies\n",
-      "\n",
-      "\n",
-      "\n",
-      "## Performance and Applications\n",
-      "\n",
-      "The Transformer model demonstrates significant performance advantages and practical applications across multiple domains:\n",
-      "\n",
-      "Performance Advantages over RNN/CNN Models:\n",
-      "- Eliminates sequential computation constraints present in RNNs, enabling superior parallelization\n",
-      "- Reduces operations needed for relating distant positions to a constant number, compared to linear/logarithmic scaling in CNNs\n",
-      "- Processes all input and output positions simultaneously through self-attention mechanisms\n",
-      "- Achieves state-of-the-art results while requiring significantly less computational resources\n",
-      "\n",
-      "Machine Translation Benchmarks:\n",
-      "- WMT 2014 English-to-German: 28.4 BLEU score, exceeding previous best results by over 2 BLEU points\n",
-      "- WMT 2014 English-to-French: 41.8 BLEU score (single-model state-of-the-art)\n",
-      "- Surpasses performance of existing model ensembles in translation tasks\n",
-      "\n",
-      "Training Efficiency:\n",
-      "- Requires only 3.5 days of training on eight GPUs for state-of-the-art performance\n",
-      "- Achieves superior results at \"a small fraction of the training costs\" compared to previous models\n",
-      "- Enables significantly faster training through parallel processing of input/output sequences\n",
-      "- Can reach production-quality performance in as little as twelve hours on modern GPU hardware\n",
-      "\n",
-      "Real-world Applications:\n",
-      "- Machine translation systems\n",
-      "- Natural language understanding tasks\n",
-      "- Reading comprehension\n",
-      "- Abstractive summarization\n",
-      "- Text entailment analysis\n",
-      "- Constituency parsing (achieving 92.7 F1 score in semi-supervised settings)\n",
-      "- Adaptable to both large and limited training data scenarios\n",
-      "\n",
-      "Scalability Benefits:\n",
-      "- Highly parallelizable architecture enables efficient scaling across multiple GPUs\n",
-      "- Constant computational complexity for relating any input/output positions\n",
-      "- Effective handling of long-range dependencies in sequences\n",
-      "- Maintains performance quality while scaling to larger datasets and model sizes\n",
-      "- Generalizes well across different tasks and domains without architectural changes\n",
-      "- Supports efficient inference and deployment in production environments\n",
-      "\n"
-     ]
-    }
-   ],
-   "source": [
-    "report_response = await report_client.aget()\n",
-    "report_text = \"\\n\\n\".join([block.template for block in report_response.report.blocks])\n",
-    "print(report_text)"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "We can also see the sources for each block!"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "0.99687636\n",
-      "# Abstract\n",
-      "\n",
-      "The dominant sequence transduction models are based on complex recurrent or convolutiona\n",
-      "==================\n",
-      "0.99591404\n",
-      "# 2 Background\n",
-      "\n",
-      "The goal of reducing sequential computation also forms the foundation of the Extende\n",
-      "==================\n",
-      "0.9951325\n",
-      "# 1 Introduction\n",
-      "\n",
-      "Recurrent neural networks, long short-term memory [13] and gated recurrent [7] neu\n",
-      "==================\n",
-      "0.99442345\n",
-      "# 7 Conclusion\n",
-      "\n",
-      "In this work, we presented the Transformer, the first sequence transduction model ba\n",
-      "==================\n",
-      "0.9967649\n",
-      "# 3.2.3 Applications of Attention in our Model\n",
-      "\n",
-      "The Transformer uses multi-head attention in three d\n",
-      "==================\n",
-      "0.99533635\n",
-      "# 2 Background\n",
-      "\n",
-      "The goal of reducing sequential computation also forms the foundation of the Extende\n",
-      "==================\n",
-      "0.9935868\n",
-      "# Abstract\n",
-      "\n",
-      "The dominant sequence transduction models are based on complex recurrent or convolutiona\n",
-      "==================\n",
-      "0.98780584\n",
-      "# Outputs\n",
-      "\n",
-      "(shifted right)\n",
-      "\n",
-      "Figure 1: The Transformer - model architecture.\n",
-      "\n",
-      "The Transformer follows\n",
-      "==================\n",
-      "0.9205043\n",
-      "# 3.3 Position-wise Feed-Forward Networks\n",
-      "\n",
-      "In addition to attention sub-layers, each of the layers i\n",
-      "==================\n",
-      "0.79581684\n",
-      "# 1 Introduction\n",
-      "\n",
-      "Recurrent neural networks, long short-term memory [13] and gated recurrent [7] neu\n",
-      "==================\n",
-      "0.9946774\n",
-      "# Abstract\n",
-      "\n",
-      "The dominant sequence transduction models are based on complex recurrent or convolutiona\n",
-      "==================\n",
-      "0.97079873\n",
-      "# 7 Conclusion\n",
-      "\n",
-      "In this work, we presented the Transformer, the first sequence transduction model ba\n",
-      "==================\n",
-      "0.9535353\n",
-      "# 6.3 English Constituency Parsing\n",
-      "\n",
-      "To evaluate if the Transformer can generalize to other tasks we \n",
-      "==================\n",
-      "0.9514138\n",
-      "# 2 Background\n",
-      "\n",
-      "The goal of reducing sequential computation also forms the foundation of the Extende\n",
-      "==================\n",
-      "0.9790758\n",
-      "# 1 Introduction\n",
-      "\n",
-      "Recurrent neural networks, long short-term memory [13] and gated recurrent [7] neu\n",
-      "==================\n",
-      "0.92262185\n",
-      "# Outputs\n",
-      "\n",
-      "(shifted right)\n",
-      "\n",
-      "Figure 1: The Transformer - model architecture.\n",
-      "\n",
-      "The Transformer follows\n",
-      "==================\n"
-     ]
-    }
-   ],
-   "source": [
-    "for block in report_response.report.blocks:\n",
-    "    # Each block has a list of sources, which are the nodes that were used to generate the block\n",
-    "    for source in block.sources:\n",
-    "        print(source.score)\n",
-    "        print(source.node.text[:100])\n",
-    "        print(\"==================\")"
-   ]
-  }
- ],
- "metadata": {
-  "kernelspec": {
-   "display_name": "llama-parse-aNC435Vv-py3.10",
-   "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": 2
-}

package.json

@@ -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}": [

parse.md

@@ -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
 

pnpm-workspace.yaml

@@ -1,2 +1,3 @@
 packages:
-  - "ts/**"
+  - "ts/*"
+  - "py"

py/CHANGELOG.md

@@ -0,0 +1,7 @@
+# llama-cloud-services-py
+
+## 0.6.70
+
+### Patch Changes
+
+- d028397: Update llama-cloud api version, and integrate with agent data deletion

py/README.md

@@ -9,7 +9,6 @@ This repository contains the code for hand-written SDKs and clients for interact
 This includes:
 
 - [LlamaParse](../parse.md) - A GenAI-native document parser that can parse complex document data for any downstream LLM use case (Agents, RAG, data processing, etc.).
-- [LlamaReport (beta/invite-only)](../report.md) - A prebuilt agentic report builder that can be used to build reports from a variety of data sources.
 - [LlamaExtract](../extract.md) - A prebuilt agentic data extractor that can be used to transform data into a structured JSON representation.
 - [LlamaCloud Index](../index.md) - A widely customizable and fully automated document ingestion pipeline that also serves retrieval purposes.
 
@@ -28,14 +27,12 @@ Then, you can use the services in your code:
 ```python
 from llama_cloud_services import (
     LlamaParse,
-    LlamaReport,
     LlamaExtract,
     LlamaCloudIndex,
 )
-from llama_cloud_services import LlamaParse, LlamaReport, LlamaExtract
+from llama_cloud_services import LlamaParse, LlamaExtract
 
 parser = LlamaParse(api_key="YOUR_API_KEY")
-report = LlamaReport(api_key="YOUR_API_KEY")
 extract = LlamaExtract(api_key="YOUR_API_KEY")
 index = LlamaCloudIndex(
     "my_first_index", project_name="default", api_key="YOUR_API_KEY"
@@ -45,7 +42,6 @@ index = LlamaCloudIndex(
 See the quickstart guides for each service for more information:
 
 - [LlamaParse](../parse.md)
-- [LlamaReport (beta/invite-only)](../report.md)
 - [LlamaExtract](../extract.md)
 - [LlamaCloud Index](../index.md)
 
@@ -58,13 +54,11 @@ You can also create your API key in the EU region [here](https://cloud.eu.llamai
 ```python
 from llama_cloud_services import (
     LlamaParse,
-    LlamaReport,
     LlamaExtract,
     EU_BASE_URL,
 )
 
 parser = LlamaParse(api_key="YOUR_API_KEY", base_url=EU_BASE_URL)
-report = LlamaReport(api_key="YOUR_API_KEY", base_url=EU_BASE_URL)
 extract = LlamaExtract(api_key="YOUR_API_KEY", base_url=EU_BASE_URL)
 index = LlamaCloudIndex(
     "my_first_index",

py/llama_cloud_services/__init__.py

@@ -1,5 +1,4 @@
 from llama_cloud_services.parse import LlamaParse
-from llama_cloud_services.report import ReportClient, LlamaReport
 from llama_cloud_services.extract import LlamaExtract, ExtractionAgent, SourceText
 from llama_cloud_services.constants import EU_BASE_URL
 from llama_cloud_services.index import (
@@ -10,8 +9,6 @@ from llama_cloud_services.index import (
 
 __all__ = [
     "LlamaParse",
-    "ReportClient",
-    "LlamaReport",
     "LlamaExtract",
     "ExtractionAgent",
     "SourceText",

py/llama_cloud_services/beta/agent_data/client.py

@@ -1,6 +1,11 @@
 import os
 from typing import Any, Dict, Generic, List, Optional, Type
 
+from llama_cloud import (
+    AgentData,
+    PaginatedResponseAgentData,
+    PaginatedResponseAggregateGroup,
+)
 from llama_cloud.client import AsyncLlamaCloud
 from tenacity import (
     WrappedFn,
@@ -86,7 +91,7 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
             client=llama_client,
             type=ExtractedPerson,
             collection="extracted_people",
-            agent_url_id="person-extraction-agent"
+            deployment_name="person-extraction-agent"
         )
 
         # Create data
@@ -109,10 +114,12 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
         self,
         type: Type[AgentDataT],
         collection: str = "default",
-        agent_url_id: Optional[str] = None,
+        deployment_name: Optional[str] = None,
         client: Optional[AsyncLlamaCloud] = None,
         token: Optional[str] = None,
         base_url: Optional[str] = None,
+        # deprecated, use deployment_name instead
+        agent_url_id: Optional[str] = None,
     ):
         """
         Initialize the AsyncAgentDataClient.
@@ -123,11 +130,11 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
             collection: Named collection within the agent for organizing data.
                 Defaults to "default". Collections allow logical separation of
                 different data types or workflows within the same agent.
-            agent_url_id: Unique identifier for the agent. This normally appears in the
-                url of an agent within the llama cloud platform. If not provided,
-                will attempt to use the LLAMA_DEPLOY_DEPLOYMENT_NAME environment
-                variable. Data can only be added to an already existing agent in the
-                platform.
+            deployment_name: Unique identifier for the agent deployment. This normally
+                appears in the URL of an agent within the Llama Cloud platform. If not
+                provided, will attempt to use the LLAMA_DEPLOY_DEPLOYMENT_NAME
+                environment variable. Data can only be added to an already existing
+                agent in the platform.
             client: AsyncLlamaCloud client instance for API communication. If not provided, will
                 construct one from the provided api token and base url
             token: Llama Cloud API token. Reads from LLAMA_CLOUD_API_KEY if not provided
@@ -135,15 +142,14 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
                 defaults to https://api.cloud.llamaindex.ai
 
         Raises:
-            ValueError: If agent_url_id is not provided and the
+            ValueError: If deployment_name is not provided and the
                 LLAMA_DEPLOY_DEPLOYMENT_NAME environment variable is not set
 
         Note:
             The client automatically applies retry logic to all API calls with
             exponential backoff for timeout, connection, and HTTP status errors.
         """
-
-        self.agent_url_id = agent_url_id or get_default_agent_id()
+        self.deployment_name = deployment_name or agent_url_id or get_default_agent_id()
 
         self.collection = collection
         if not client:
@@ -156,15 +162,19 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
 
     @agent_data_retry
     async def get_item(self, item_id: str) -> TypedAgentData[AgentDataT]:
-        raw_data = await self.client.beta.get_agent_data(
+        raw_data = await self.untyped_get_item(item_id)
+        return TypedAgentData.from_raw(raw_data, self.type)
+
+    @agent_data_retry
+    async def untyped_get_item(self, item_id: str) -> AgentData:
+        return await self.client.beta.get_agent_data(
             item_id=item_id,
         )
-        return TypedAgentData.from_raw(raw_data, validator=self.type)
 
     @agent_data_retry
     async def create_item(self, data: AgentDataT) -> TypedAgentData[AgentDataT]:
         raw_data = await self.client.beta.create_agent_data(
-            agent_slug=self.agent_url_id,
+            deployment_name=self.deployment_name,
             collection=self.collection,
             data=data.model_dump(),
         )
@@ -184,6 +194,21 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
     async def delete_item(self, item_id: str) -> None:
         await self.client.beta.delete_agent_data(item_id=item_id)
 
+    @agent_data_retry
+    async def delete(
+        self, filter: Optional[Dict[str, Dict[ComparisonOperator, Any]]] = None
+    ) -> int:
+        """
+        Delete agent data by query, similar to search.
+        Returns the number of deleted items.
+        """
+        response = await self.client.beta.delete_agent_data_by_query_api_v_1_beta_agent_data_delete_post(
+            deployment_name=self.deployment_name,
+            collection=self.collection,
+            filter=filter,
+        )
+        return response.deleted_count
+
     @agent_data_retry
     async def search(
         self,
@@ -210,9 +235,7 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
             offset: Number of items to skip from the beginning. Defaults to 0.
             include_total: Whether to include the total count in the response. Defaults to False to improve performance. It's recommended to only request on the first page.
         """
-        raw = await self.client.beta.search_agent_data_api_v_1_beta_agent_data_search_post(
-            agent_slug=self.agent_url_id,
-            collection=self.collection,
+        raw = await self.untyped_search(
             filter=filter,
             order_by=order_by,
             offset=offset,
@@ -227,6 +250,25 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
             total=raw.total_size,
         )
 
+    @agent_data_retry
+    async def untyped_search(
+        self,
+        filter: Optional[Dict[str, Dict[ComparisonOperator, Any]]] = None,
+        order_by: Optional[str] = None,
+        offset: Optional[int] = None,
+        page_size: Optional[int] = None,
+        include_total: bool = False,
+    ) -> PaginatedResponseAgentData:
+        return await self.client.beta.search_agent_data_api_v_1_beta_agent_data_search_post(
+            deployment_name=self.deployment_name,
+            collection=self.collection,
+            filter=filter,
+            order_by=order_by,
+            offset=offset,
+            page_size=page_size,
+            include_total=include_total,
+        )
+
     @agent_data_retry
     async def aggregate(
         self,
@@ -253,8 +295,38 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
             offset: Number of groups to skip from the beginning. Defaults to 0.
             page_size: Maximum number of groups to return per page.
         """
-        raw = await self.client.beta.aggregate_agent_data_api_v_1_beta_agent_data_aggregate_post(
-            agent_slug=self.agent_url_id,
+        raw = await self.untyped_aggregate(
+            filter=filter,
+            group_by=group_by,
+            count=count,
+            first=first,
+            order_by=order_by,
+            offset=offset,
+            page_size=page_size,
+        )
+
+        return TypedAggregateGroupItems(
+            items=[
+                TypedAggregateGroup.from_raw(grp, validator=self.type)
+                for grp in raw.items
+            ],
+            has_more=raw.next_page_token is not None,
+            total=raw.total_size,
+        )
+
+    @agent_data_retry
+    async def untyped_aggregate(
+        self,
+        filter: Optional[Dict[str, Dict[ComparisonOperator, Any]]] = None,
+        group_by: Optional[List[str]] = None,
+        count: Optional[bool] = None,
+        first: Optional[bool] = None,
+        order_by: Optional[str] = None,
+        offset: Optional[int] = None,
+        page_size: Optional[int] = None,
+    ) -> PaginatedResponseAggregateGroup:
+        return await self.client.beta.aggregate_agent_data_api_v_1_beta_agent_data_aggregate_post(
+            deployment_name=self.deployment_name,
             collection=self.collection,
             page_size=page_size,
             filter=filter,
@@ -264,11 +336,3 @@ class AsyncAgentDataClient(Generic[AgentDataT]):
             first=first,
             offset=offset,
         )
-        return TypedAggregateGroupItems(
-            items=[
-                TypedAggregateGroup.from_raw(item, validator=self.type)
-                for item in raw.items
-            ],
-            has_more=raw.next_page_token is not None,
-            total=raw.total_size,
-        )

py/llama_cloud_services/beta/agent_data/schema.py

@@ -10,7 +10,7 @@ CRUD operations, search capabilities, filtering, and aggregation functionality
 for managing agent-generated data at scale.
 
 Key Concepts:
-- Agent Slug: Unique identifier for an agent instance
+- Deployment Name: Unique identifier for an agent deployment
 - Collection: Named grouping of data within an agent (defaults to "default"). Data within a collection should be of the same type.
 - Agent Data: Individual structured data records with metadata and timestamps
 
@@ -26,7 +26,7 @@ Example Usage:
         client=async_llama_cloud,
         type=Person,
         collection="people",
-        agent_url_id="my-extraction-agent-xyz"
+        deployment_name="my-extraction-agent-xyz"
     )
 
     # Create typed data
@@ -56,7 +56,6 @@ from typing import (
 
 # Type variable for user-defined data models
 AgentDataT = TypeVar("AgentDataT", bound=BaseModel)
-
 # Type variable for extracted data (can be dict or Pydantic model)
 ExtractedT = TypeVar("ExtractedT", bound=Union[BaseModel, dict])
 
@@ -78,7 +77,7 @@ class TypedAgentData(BaseModel, Generic[AgentDataT]):
 
     Attributes:
         id: Unique identifier for this data record
-        agent_url_id: Identifier of the agent that created this data
+        deployment_name: Identifier of the agent deployment that created this data
         collection: Named collection within the agent (used for organization)
         data: The actual structured data payload (typed as AgentDataT)
         created_at: Timestamp when the record was first created
@@ -94,8 +93,8 @@ class TypedAgentData(BaseModel, Generic[AgentDataT]):
     """
 
     id: Optional[str] = Field(description="Unique identifier for this data record")
-    agent_url_id: str = Field(
-        description="Identifier of the agent that created this data"
+    deployment_name: str = Field(
+        description="Identifier of the agent deployment that created this data"
     )
     collection: Optional[str] = Field(
         description="Named collection within the agent for data organization"
@@ -116,15 +115,15 @@ class TypedAgentData(BaseModel, Generic[AgentDataT]):
         Args:
             raw_data: Raw agent data from the API
             validator: Pydantic model class to validate the data field
-
         Returns:
             TypedAgentData instance with validated data
         """
+
         data: AgentDataT = validator.model_validate(raw_data.data)
 
         return cls(
             id=raw_data.id,
-            agent_url_id=raw_data.agent_slug,
+            deployment_name=raw_data.deployment_name,
             collection=raw_data.collection,
             data=data,
             created_at=raw_data.created_at,
@@ -222,12 +221,16 @@ def parse_extracted_field_metadata(
     return {
         k: _parse_extracted_field_metadata_recursive(v)
         for k, v in field_metadata.items()
-        if k not in _METADATA_FIELDS_SIBLING_TO_LEAF
-        and k not in _ADDITIONAL_ROOT_METADATA_FIELDS
+        if not _is_reasoning_field(k, v) and k not in _ADDITIONAL_ROOT_METADATA_FIELDS
     }
 
 
-_METADATA_FIELDS_SIBLING_TO_LEAF = {"reasoning"}
+def _is_reasoning_field(field_name: str, field_value: Any) -> bool:
+    # There can either be a user specified reasoning field (from the schema), or a reasoning metadata field for the
+    # dict of values
+    return field_name == "reasoning" and isinstance(field_value, str)
+
+
 _ADDITIONAL_ROOT_METADATA_FIELDS = {"error"}
 
 
@@ -257,14 +260,12 @@ def _parse_extracted_field_metadata_recursive(
             except ValidationError:
                 pass
         additional_fields = {
-            k: v
-            for k, v in field_value.items()
-            if k in _METADATA_FIELDS_SIBLING_TO_LEAF
+            k: v for k, v in field_value.items() if _is_reasoning_field(k, v)
         }
         return {
             k: _parse_extracted_field_metadata_recursive(v, additional_fields)
             for k, v in field_value.items()
-            if k not in _METADATA_FIELDS_SIBLING_TO_LEAF
+            if not _is_reasoning_field(k, v)
         }
     elif isinstance(field_value, list):
         return [_parse_extracted_field_metadata_recursive(item) for item in field_value]

py/llama_cloud_services/beta/classifier/client.py

@@ -9,7 +9,6 @@ from llama_cloud.types import (
     ClassifyJobResults,
     ClassifyParsingConfiguration,
     StatusEnum,
-    ClassifyJobWithStatus,
     File,
 )
 from llama_cloud.resources.classifier.client import OMIT
@@ -229,7 +228,7 @@ class ClassifyClient:
                 )
             )
 
-    async def wait_for_job_completion(self, job_id: str) -> ClassifyJobWithStatus:
+    async def wait_for_job_completion(self, job_id: str) -> ClassifyJob:
         """
         Wait for a classify job to complete.
         Meant to expose lower level access to classifier jobs for advanced use cases.

py/llama_cloud_services/extract/extract.py

@@ -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
         ]

py/llama_cloud_services/index/base.py

@@ -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)

py/llama_cloud_services/parse/base.py

@@ -396,6 +396,10 @@ class LlamaParse(BasePydanticReader):
         default=False,
         description="If set, the parser will try to preserve very small text lines. This can be useful for documents containing vector graphics with very small text lines that may not be recognized by OCR or a vision model (such as in CAD drawings).",
     )
+    precise_bounding_box: Optional[bool] = Field(
+        default=False,
+        description="If set to true, the parser will use a more precise bounding box to extract text from documents. This will increase the accuracy of the parsing job, but reduce the speed.",
+    )
     replace_failed_page_mode: Optional[FailedPageMode] = Field(
         default=None,
         description="The mode to use to replace the failed page, see FailedPageMode enum for possible value. If set, the parser will replace the failed page with the specified mode. If not set, the default mode (raw_text) will be used.",
@@ -416,7 +420,22 @@ class LlamaParse(BasePydanticReader):
         default=False,
         description="If set to true, the parser will extract sub-tables from the spreadsheet when possible (more than one table per sheet).",
     )
-
+    spreadsheet_force_formula_computation: Optional[bool] = Field(
+        default=False,
+        description="If set to true, the parser will re-compute values for all spreadsheet cells containing formulas.",
+    )
+    specialized_chart_parsing_agentic: Optional[bool] = Field(
+        default=False,
+        description="If set to true, the parser will use a specialized agentic chart parsing model to extract data from charts. This model is able to understand the chart type and extract the data accordingly.",
+    )
+    specialized_chart_parsing_efficient: Optional[bool] = Field(
+        default=False,
+        description="If set to true, the parser will use a specialized efficient chart parsing model to extract data from charts. This model is faster and cheaper than the agentic model, but may be less accurate.",
+    )
+    specialized_chart_parsing_plus: Optional[bool] = Field(
+        default=False,
+        description="If set to true, the parser will use a specialized one-shot chart parsing model to extract data from charts. This model is able to understand the chart type and extract the data accordingly. It is more accurate than the efficient model, but also more expensive.",
+    )
     strict_mode_buggy_font: Optional[bool] = Field(
         default=False,
         description="If set to true, the parser will fail if it can't extract text from a document because of a buggy font.",
@@ -928,6 +947,9 @@ class LlamaParse(BasePydanticReader):
         if self.preset is not None:
             data["preset"] = self.preset
 
+        if self.precise_bounding_box:
+            data["precise_bounding_box"] = self.precise_bounding_box
+
         if self.replace_failed_page_mode is not None:
             data["replace_failed_page_mode"] = self.replace_failed_page_mode.value
 
@@ -947,6 +969,24 @@ class LlamaParse(BasePydanticReader):
         if self.spreadsheet_extract_sub_tables:
             data["spreadsheet_extract_sub_tables"] = self.spreadsheet_extract_sub_tables
 
+        if self.spreadsheet_force_formula_computation:
+            data[
+                "spreadsheet_force_formula_computation"
+            ] = self.spreadsheet_force_formula_computation
+
+        if self.specialized_chart_parsing_agentic:
+            data[
+                "specialized_chart_parsing_agentic"
+            ] = self.specialized_chart_parsing_agentic
+
+        if self.specialized_chart_parsing_efficient:
+            data[
+                "specialized_chart_parsing_efficient"
+            ] = self.specialized_chart_parsing_efficient
+
+        if self.specialized_chart_parsing_plus:
+            data["specialized_chart_parsing_plus"] = self.specialized_chart_parsing_plus
+
         if self.strict_mode_buggy_font:
             data["strict_mode_buggy_font"] = self.strict_mode_buggy_font
 

py/llama_cloud_services/parse/types.py

@@ -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
 
@@ -159,6 +162,9 @@ class Page(BaseModel):
     durationInSeconds: Optional[float] = Field(
         default=None, description="The duration of the audio transcript in seconds."
     )
+    slideSpeakerNotes: Optional[str] = Field(
+        default=None, description="The speaker notes for the slide."
+    )
 
 
 class JobResult(BaseModel):
@@ -255,6 +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.
@@ -265,17 +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},
                 )
             ]
@@ -325,7 +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:
         """

py/llama_cloud_services/parse/utils.py

@@ -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]:

py/llama_cloud_services/report/__init__.py

@@ -1,4 +0,0 @@
-from llama_cloud_services.report.report import ReportClient
-from llama_cloud_services.report.base import LlamaReport
-
-__all__ = ["ReportClient", "LlamaReport"]

py/llama_cloud_services/report/base.py

@@ -1,269 +0,0 @@
-import asyncio
-import httpx
-import os
-import io
-from concurrent.futures import ThreadPoolExecutor
-from typing import Optional, List, Union, Any, Coroutine, TypeVar
-from urllib.parse import urljoin
-
-from llama_cloud.types import ReportMetadata
-from llama_cloud_services.report.report import ReportClient
-
-T = TypeVar("T")
-
-
-class LlamaReport:
-    """Client for managing reports and general report operations."""
-
-    def __init__(
-        self,
-        api_key: Optional[str] = None,
-        project_id: Optional[str] = None,
-        organization_id: Optional[str] = None,
-        base_url: Optional[str] = None,
-        timeout: Optional[int] = None,
-        async_httpx_client: Optional[httpx.AsyncClient] = None,
-    ):
-        self.api_key = api_key or os.getenv("LLAMA_CLOUD_API_KEY", None)
-        if not self.api_key:
-            raise ValueError("No API key provided.")
-
-        self.base_url = base_url or os.getenv(
-            "LLAMA_CLOUD_BASE_URL", "https://api.cloud.llamaindex.ai"
-        )
-        self.timeout = timeout or 60
-
-        # Initialize HTTP clients
-        self._aclient = async_httpx_client or httpx.AsyncClient(timeout=self.timeout)
-
-        # Set auth headers
-        self.headers = {
-            "Authorization": f"Bearer {self.api_key}",
-        }
-
-        self.organization_id = organization_id
-        self.project_id = project_id
-        self._client_params = {
-            "timeout": self._aclient.timeout,
-            "headers": self._aclient.headers,
-            "base_url": self._aclient.base_url,
-            "auth": self._aclient.auth,
-            "event_hooks": self._aclient.event_hooks,
-            "cookies": self._aclient.cookies,
-            "max_redirects": self._aclient.max_redirects,
-            "params": self._aclient.params,
-            "trust_env": self._aclient.trust_env,
-        }
-        self._thread_pool = ThreadPoolExecutor(
-            max_workers=min(10, (os.cpu_count() or 1) + 4)
-        )
-
-    @property
-    def aclient(self) -> httpx.AsyncClient:
-        if self._aclient is None:
-            self._aclient = httpx.AsyncClient(**self._client_params)
-        return self._aclient
-
-    def _run_sync(self, coro: Coroutine[Any, Any, T]) -> T:
-        """Run coroutine in a separate thread to avoid event loop issues"""
-
-        # force a new client for this thread/event loop
-        original_client = self._aclient
-        self._aclient = None
-
-        def run_coro() -> T:
-            async def wrapped_coro() -> T:
-                return await coro
-
-            return asyncio.run(wrapped_coro())
-
-        result = self._thread_pool.submit(run_coro).result()
-
-        # restore the original client
-        self._aclient = original_client
-
-        return result
-
-    async def _get_default_project(self) -> str:
-        response = await self.aclient.get(
-            urljoin(str(self.base_url), "/api/v1/projects"), headers=self.headers
-        )
-        response.raise_for_status()
-        projects = response.json()
-        default_project = [p for p in projects if p.get("is_default")]
-        return default_project[0]["id"]
-
-    async def _build_url(
-        self, endpoint: str, extra_params: Optional[List[str]] = None
-    ) -> str:
-        """Helper method to build URLs with common query parameters."""
-        url = urljoin(str(self.base_url), endpoint)
-
-        if not self.project_id:
-            self.project_id = await self._get_default_project()
-
-        query_params = []
-        if self.organization_id:
-            query_params.append(f"organization_id={self.organization_id}")
-        if self.project_id:
-            query_params.append(f"project_id={self.project_id}")
-        if extra_params:
-            query_params.extend([p for p in extra_params if p is not None])
-
-        if query_params:
-            url += "?" + "&".join(query_params)
-
-        return url
-
-    async def acreate_report(
-        self,
-        name: str,
-        template_instructions: Optional[str] = None,
-        template_text: Optional[str] = None,
-        template_file: Optional[Union[str, tuple[str, bytes]]] = None,
-        input_files: Optional[List[Union[str, tuple[str, bytes]]]] = None,
-        existing_retriever_id: Optional[str] = None,
-    ) -> ReportClient:
-        """Create a new report asynchronously."""
-        url = await self._build_url("/api/v1/reports/")
-        open_files: List[io.BufferedReader] = []
-
-        data = {"name": name}
-        if template_instructions:
-            data["template_instructions"] = template_instructions
-        if template_text:
-            data["template_text"] = template_text
-        if existing_retriever_id:
-            data["existing_retriever_id"] = str(existing_retriever_id)
-
-        files: List[tuple[str, io.BufferedReader | bytes]] = []
-        if template_file:
-            if isinstance(template_file, str):
-                open_files.append(open(template_file, "rb"))
-                files.append(("template_file", open_files[-1]))
-            else:
-                files.append(("template_file", template_file[1]))
-
-        if input_files:
-            for f in input_files:
-                if isinstance(f, str):
-                    open_files.append(open(f, "rb"))
-                    files.append(("files", open_files[-1]))
-                else:
-                    files.append(("files", f[1]))
-
-        response = await self.aclient.post(
-            url, headers=self.headers, data=data, files=files
-        )
-        try:
-            response.raise_for_status()
-            report_id = response.json()["id"]
-            return ReportClient(report_id, name, self)
-        except httpx.HTTPStatusError as e:
-            raise ValueError(
-                f"Failed to create report: {e.response.text}\nError Code: {e.response.status_code}"
-            )
-        finally:
-            for open_file in open_files:
-                open_file.close()
-
-    def create_report(
-        self,
-        name: str,
-        template_instructions: Optional[str] = None,
-        template_text: Optional[str] = None,
-        template_file: Optional[Union[str, tuple[str, bytes]]] = None,
-        input_files: Optional[List[Union[str, tuple[str, bytes]]]] = None,
-        existing_retriever_id: Optional[str] = None,
-    ) -> ReportClient:
-        """Create a new report."""
-        return self._run_sync(
-            self.acreate_report(
-                name=name,
-                template_instructions=template_instructions,
-                template_text=template_text,
-                template_file=template_file,
-                input_files=input_files,
-                existing_retriever_id=existing_retriever_id,
-            )
-        )
-
-    async def alist_reports(
-        self, state: Optional[str] = None, limit: int = 100, offset: int = 0
-    ) -> List[ReportClient]:
-        """List all reports asynchronously."""
-        params = []
-        if state:
-            params.append(f"state={state}")
-        if limit:
-            params.append(f"limit={limit}")
-        if offset:
-            params.append(f"offset={offset}")
-
-        url = await self._build_url(
-            "/api/v1/reports/list",
-            extra_params=params,
-        )
-
-        response = await self.aclient.get(url, headers=self.headers)
-        response.raise_for_status()
-        data = response.json()
-
-        return [
-            ReportClient(r["report_id"], r["name"], self)
-            for r in data["report_responses"]
-        ]
-
-    def list_reports(
-        self, state: Optional[str] = None, limit: int = 100, offset: int = 0
-    ) -> List[ReportClient]:
-        """Synchronous wrapper for listing reports."""
-        return self._run_sync(self.alist_reports(state, limit, offset))
-
-    async def aget_report(self, report_id: str) -> ReportClient:
-        """Get a Report instance for working with a specific report."""
-        url = await self._build_url(f"/api/v1/reports/{report_id}")
-
-        response = await self.aclient.get(url, headers=self.headers)
-        response.raise_for_status()
-        data = response.json()
-
-        return ReportClient(data["report_id"], data["name"], self)
-
-    def get_report(self, report_id: str) -> ReportClient:
-        """Synchronous wrapper for getting a report."""
-        return self._run_sync(self.aget_report(report_id))
-
-    async def aget_report_metadata(self, report_id: str) -> ReportMetadata:
-        """Get metadata for a specific report asynchronously.
-
-        Returns:
-            dict containing:
-            - id: Report ID
-            - name: Report name
-            - state: Current report state
-            - report_metadata: Additional metadata
-            - template_file: Name of template file if used
-            - template_instructions: Template instructions if provided
-            - input_files: List of input file names
-        """
-        url = await self._build_url(f"/api/v1/reports/{report_id}/metadata")
-
-        response = await self.aclient.get(url, headers=self.headers)
-        response.raise_for_status()
-        return ReportMetadata(**response.json())
-
-    def get_report_metadata(self, report_id: str) -> ReportMetadata:
-        """Synchronous wrapper for getting report metadata."""
-        return self._run_sync(self.aget_report_metadata(report_id))
-
-    async def adelete_report(self, report_id: str) -> None:
-        """Delete a specific report asynchronously."""
-        url = await self._build_url(f"/api/v1/reports/{report_id}")
-
-        response = await self.aclient.delete(url, headers=self.headers)
-        response.raise_for_status()
-
-    def delete_report(self, report_id: str) -> None:
-        """Synchronous wrapper for deleting a report."""
-        return self._run_sync(self.adelete_report(report_id))

py/llama_cloud_services/report/report.py

@@ -1,527 +0,0 @@
-import asyncio
-import httpx
-import time
-from typing import Optional, List, Literal, Union, TYPE_CHECKING
-from dataclasses import dataclass
-from datetime import datetime
-from enum import Enum
-
-from llama_cloud.types import (
-    ReportEventItemEventData_Progress,
-    ReportMetadata,
-    EditSuggestion,
-    ReportResponse,
-    ReportPlan,
-    ReportBlock,
-    ReportPlanBlock,
-    Report,
-)
-
-if TYPE_CHECKING:
-    from llama_cloud_services.report.base import LlamaReport
-
-
-class MessageRole(str, Enum):
-    USER = "user"
-    ASSISTANT = "assistant"
-
-
-@dataclass
-class Message:
-    role: MessageRole
-    content: str
-    timestamp: datetime
-
-
-@dataclass
-class EditAction:
-    block_idx: int
-    old_content: str
-    new_content: Optional[str]
-    action: Literal["approved", "rejected"]
-    timestamp: datetime
-
-
-DEFAULT_POLL_INTERVAL = 5
-DEFAULT_TIMEOUT = 600
-
-
-class ReportClient:
-    """Client for operations on a specific report."""
-
-    def __init__(self, report_id: str, name: str, parent_client: "LlamaReport"):
-        self.report_id = report_id
-        self.name = name
-        self._client = parent_client
-        self._headers = parent_client.headers
-        self._run_sync = parent_client._run_sync
-        self._build_url = parent_client._build_url
-        self.chat_history: List[Message] = []
-        self.edit_history: List[EditAction] = []
-
-    @property
-    def aclient(self) -> httpx.AsyncClient:
-        return self._client.aclient
-
-    def __str__(self) -> str:
-        return f"Report(id={self.report_id}, name={self.name})"
-
-    def __repr__(self) -> str:
-        return f"Report(id={self.report_id}, name={self.name})"
-
-    def _get_block_content(self, block: Union[ReportBlock, ReportPlanBlock]) -> str:
-        if isinstance(block, ReportBlock):
-            return block.template
-        elif isinstance(block, ReportPlanBlock):
-            return block.block.template
-        else:
-            raise ValueError(f"Invalid block type: {type(block)}")
-
-    def _get_block_idx(self, block: Union[ReportBlock, ReportPlanBlock]) -> int:
-        if isinstance(block, ReportBlock):
-            return block.idx
-        elif isinstance(block, ReportPlanBlock):
-            return block.block.idx
-        else:
-            raise ValueError(f"Invalid block type: {type(block)}")
-
-    async def aget(self, version: Optional[int] = None) -> ReportResponse:
-        """Get this report's details asynchronously."""
-        extra_params = []
-        if version is not None:
-            extra_params.append(f"version={version}")
-
-        url = await self._build_url(f"/api/v1/reports/{self.report_id}", extra_params)
-
-        response = await self.aclient.get(url, headers=self._headers)
-        response.raise_for_status()
-        return ReportResponse(**response.json())
-
-    def get(self, version: Optional[int] = None) -> ReportResponse:
-        """Synchronous wrapper for getting this report's details."""
-        return self._run_sync(self.aget(version))
-
-    async def aupdate_report(self, updated_report: Report) -> ReportResponse:
-        """Update this report's content asynchronously."""
-        url = await self._build_url(f"/api/v1/reports/{self.report_id}")
-        response = await self.aclient.patch(
-            url, headers=self._headers, json={"content": updated_report.dict()}
-        )
-        response.raise_for_status()
-        return ReportResponse(**response.json())
-
-    def update_report(self, updated_report: Report) -> ReportResponse:
-        """Synchronous wrapper for updating this report's content."""
-        return self._run_sync(self.aupdate_report(updated_report))
-
-    async def aupdate_plan(
-        self,
-        action: Literal["approve", "reject", "edit"],
-        updated_plan: Optional[ReportPlan] = None,
-    ) -> ReportResponse:
-        """Update this report's plan asynchronously."""
-        if action == "edit" and not updated_plan:
-            raise ValueError("updated_plan is required when action is 'edit'")
-
-        url = await self._build_url(
-            f"/api/v1/reports/{self.report_id}/plan", [f"action={action}"]
-        )
-
-        data = None
-        if updated_plan is not None:
-            plan_dict = updated_plan.dict()
-            plan_dict.pop("generated_at", None)
-            data = plan_dict
-
-        if updated_plan is None and action == "edit":
-            raise ValueError("updated_plan is required when action is 'edit'")
-
-        response = await self.aclient.patch(url, headers=self._headers, json=data)
-        response.raise_for_status()
-        return ReportResponse(**response.json())
-
-    def update_plan(
-        self,
-        action: Literal["approve", "reject", "edit"],
-        updated_plan: Optional[ReportPlan] = None,
-    ) -> ReportResponse:
-        """Synchronous wrapper for updating this report's plan."""
-        return self._run_sync(self.aupdate_plan(action, updated_plan))
-
-    async def asuggest_edits(
-        self,
-        user_query: str,
-        auto_history: bool = True,
-        chat_history: Optional[List[dict]] = None,
-    ) -> List[EditSuggestion]:
-        """Get AI suggestions for edits to this report asynchronously.
-
-        Args:
-            user_query: The user's request/question about what to edit
-            auto_history: Whether to automatically add the user's message to the chat history
-            chat_history:
-                A list of chat messages to include in the chat history.
-                The format being a list of dictionaries with "role" and "content" keys.
-        """
-        # Add user message to history
-        self.chat_history.append(
-            Message(role=MessageRole.USER, content=user_query, timestamp=datetime.now())
-        )
-
-        # Format chat history with edit summaries
-        chat_history_dicts = []
-        for msg in self.chat_history[:-1]:  # Exclude current message
-            content = msg.content
-            if msg.role == MessageRole.USER:
-                # Add edit summary for user messages
-                edit_summary = self._get_edit_summary_after_message(msg.timestamp)
-                if edit_summary:
-                    content = f"{content}\n\nActions taken:\n{edit_summary}"
-
-            chat_history_dicts.append({"role": msg.role.value, "content": content})
-
-        # decide whether to include chat history or not
-        if chat_history:
-            chat_history_dicts = chat_history
-        elif auto_history:
-            chat_history_dicts = chat_history_dicts
-        else:
-            chat_history_dicts = []
-
-        # Make the API call
-        url = await self._build_url(f"/api/v1/reports/{self.report_id}/suggest_edits")
-        data = {"user_query": user_query, "chat_history": chat_history_dicts}
-
-        response = await self.aclient.post(url, headers=self._headers, json=data)
-        response.raise_for_status()
-        suggestions = response.json()
-        suggestions = [EditSuggestion(**suggestion) for suggestion in suggestions]
-
-        # Add assistant response to history
-        if suggestions:
-            for suggestion in suggestions:
-                self.chat_history.append(
-                    Message(
-                        role=MessageRole.ASSISTANT,
-                        content=suggestion.justification,
-                        timestamp=datetime.now(),
-                    )
-                )
-
-        return suggestions
-
-    def suggest_edits(
-        self,
-        user_query: str,
-        auto_history: bool = True,
-        chat_history: Optional[List[dict]] = None,
-    ) -> List[EditSuggestion]:
-        """Synchronous wrapper for getting edit suggestions."""
-        return self._run_sync(
-            self.asuggest_edits(user_query, auto_history, chat_history)
-        )
-
-    async def await_completion(
-        self, timeout: int = DEFAULT_TIMEOUT, poll_interval: int = DEFAULT_POLL_INTERVAL
-    ) -> Report:
-        """Wait for this report to complete processing."""
-        start_time = time.time()
-        while True:
-            report_response = await self.aget()
-            status = report_response.status
-
-            if status == "completed":
-                return report_response.report
-            elif status == "error":
-                events = await self.aget_events()
-                raise ValueError(f"Report entered error state: {events[-1].msg}")
-            elif time.time() - start_time > timeout:
-                raise TimeoutError(f"Report did not complete within {timeout} seconds")
-
-            await asyncio.sleep(poll_interval)
-
-    def wait_for_completion(
-        self, timeout: int = DEFAULT_TIMEOUT, poll_interval: int = DEFAULT_POLL_INTERVAL
-    ) -> Report:
-        """Synchronous wrapper for awaiting report completion."""
-        return self._run_sync(self.await_completion(timeout, poll_interval))
-
-    async def await_for_plan(
-        self, timeout: int = DEFAULT_TIMEOUT, poll_interval: int = DEFAULT_POLL_INTERVAL
-    ) -> ReportPlan:
-        """Wait for this report's plan to be ready for review."""
-        start_time = time.time()
-        while True:
-            report_metadata = await self.aget_metadata()
-            state = report_metadata.state
-
-            if state == "waiting_approval":
-                report_response = await self.aget()
-                return report_response.plan
-            elif state == "error":
-                events = await self.aget_events()
-                raise ValueError(f"Report entered error state: {events[-1].msg}")
-            elif time.time() - start_time > timeout:
-                raise TimeoutError(f"Plan was not ready within {timeout} seconds")
-
-            await asyncio.sleep(poll_interval)
-
-    def wait_for_plan(
-        self, timeout: int = DEFAULT_TIMEOUT, poll_interval: int = DEFAULT_POLL_INTERVAL
-    ) -> ReportPlan:
-        """Synchronous wrapper for awaiting plan readiness."""
-        return self._run_sync(self.await_for_plan(timeout, poll_interval))
-
-    async def aget_metadata(self) -> ReportMetadata:
-        """Get this report's metadata asynchronously."""
-        return await self._client.aget_report_metadata(self.report_id)
-
-    def get_metadata(self) -> ReportMetadata:
-        """Synchronous wrapper for getting this report's metadata."""
-        return self._run_sync(self.aget_metadata())
-
-    async def adelete(self) -> None:
-        """Delete this report asynchronously."""
-        return await self._client.adelete_report(self.report_id)
-
-    def delete(self) -> None:
-        """Synchronous wrapper for deleting this report."""
-        return self._run_sync(self.adelete())
-
-    async def aaccept_edit(self, suggestion: EditSuggestion) -> None:
-        """Accept a suggested edit.
-
-        Args:
-            suggestion: The EditSuggestion to accept, typically from suggest_edits()
-        """
-        if len(suggestion.blocks) == 0:
-            return
-
-        # Determine if we're editing a plan or report based on first block type
-        is_plan_edit = isinstance(suggestion.blocks[0], ReportPlanBlock)
-
-        # Get current content
-        report_response = await self.aget()
-        current_blocks = (
-            report_response.plan.blocks
-            if is_plan_edit
-            else report_response.report.blocks
-        )
-
-        # Track the edit
-        new_blocks = []
-        for edit_block in suggestion.blocks:
-            # Find matching block in current content
-            old_block = next(
-                (
-                    b
-                    for b in current_blocks
-                    if self._get_block_idx(b) == self._get_block_idx(edit_block)
-                ),
-                None,
-            )
-
-            old_content = (
-                self._get_block_content(old_block) if old_block else "[No old content]"
-            )
-            new_content = self._get_block_content(edit_block)
-
-            if is_plan_edit:
-                new_queries_str = "\n".join(
-                    [
-                        f"Field: {q.field}, Prompt: {q.prompt}, Context: {q.context}"
-                        for q in edit_block.queries
-                    ]
-                )
-                new_dependency_str = (
-                    f"Depends on: {edit_block.dependency}"
-                    if edit_block.dependency
-                    else ""
-                )
-                new_content += f"\n\n{new_queries_str}\n{new_dependency_str}"
-
-                if old_block:
-                    old_queries_str = "\n".join(
-                        [
-                            f"Field: {q.field}, Prompt: {q.prompt}, Context: {q.context}"
-                            for q in old_block.queries
-                        ]
-                    )
-                    old_dependency_str = (
-                        f"Depends on: {old_block.dependency}"
-                        if old_block.dependency
-                        else ""
-                    )
-                    old_content += f"\n\n{old_queries_str}\n{old_dependency_str}"
-
-            self.edit_history.append(
-                EditAction(
-                    block_idx=self._get_block_idx(edit_block),
-                    old_content=old_content,
-                    new_content=new_content,
-                    action="approved",
-                    timestamp=datetime.now(),
-                )
-            )
-
-            # Create updated block
-            if is_plan_edit:
-                new_blocks.append(
-                    ReportPlanBlock(
-                        block=ReportBlock(
-                            idx=edit_block.block.idx,
-                            template=self._get_block_content(edit_block),
-                            sources=edit_block.block.sources,
-                        ),
-                        queries=edit_block.queries,
-                        dependency=edit_block.dependency,
-                    )
-                )
-            else:
-                new_blocks.append(
-                    ReportBlock(
-                        idx=edit_block.idx,
-                        template=self._get_block_content(edit_block),
-                        sources=edit_block.sources,
-                    )
-                )
-
-        if new_blocks:
-            if is_plan_edit:
-                # Update plan in place
-                plan = report_response.plan
-
-                # Replace edited blocks and add new ones
-                for new_block in new_blocks:
-                    block_idx = self._get_block_idx(new_block)
-                    existing_block_idx = next(
-                        (
-                            i
-                            for i, b in enumerate(plan.blocks)
-                            if b.block.idx == block_idx
-                        ),
-                        None,
-                    )
-
-                    if existing_block_idx is not None:
-                        # Replace existing block
-                        plan.blocks[existing_block_idx] = new_block
-                    else:
-                        # Add new block to end
-                        plan.blocks.append(new_block)
-
-                await self.aupdate_plan("edit", plan)
-            else:
-                # Update report in place
-                report = report_response.report
-
-                # Replace edited blocks and add new ones
-                for new_block in new_blocks:
-                    block_idx = self._get_block_idx(new_block)
-                    existing_block_idx = next(
-                        (i for i, b in enumerate(report.blocks) if b.idx == block_idx),
-                        None,
-                    )
-
-                    if existing_block_idx is not None:
-                        # Replace existing block
-                        report.blocks[existing_block_idx] = new_block
-                    else:
-                        # Add new block to end
-                        report.blocks.append(new_block)
-
-                await self.aupdate_report(report)
-
-    def accept_edit(self, suggestion: EditSuggestion) -> None:
-        """Synchronous wrapper for accepting an edit."""
-        return self._run_sync(self.aaccept_edit(suggestion))
-
-    async def areject_edit(self, suggestion: EditSuggestion) -> None:
-        """Reject a suggested edit.
-
-        Args:
-            suggestion: The EditSuggestion to reject, typically from suggest_edits()
-        """
-        # Track the rejections
-        for edit_block in suggestion.blocks:
-            self.edit_history.append(
-                EditAction(
-                    block_idx=self._get_block_idx(edit_block),
-                    old_content=self._get_block_content(edit_block),
-                    new_content=None,
-                    action="rejected",
-                    timestamp=datetime.now(),
-                )
-            )
-
-    def reject_edit(self, suggestion: EditSuggestion) -> None:
-        """Synchronous wrapper for rejecting an edit."""
-        return self._run_sync(self.areject_edit(suggestion))
-
-    def _get_edit_summary_after_message(
-        self, message_timestamp: datetime
-    ) -> Optional[str]:
-        """Get a summary of edits that occurred after a specific message."""
-        relevant_edits = [
-            edit for edit in self.edit_history if edit.timestamp > message_timestamp
-        ]
-
-        if not relevant_edits:
-            return None
-
-        approved = [edit for edit in relevant_edits if edit.action == "approved"]
-        rejected = [edit for edit in relevant_edits if edit.action == "rejected"]
-
-        summary = []
-
-        if approved:
-            summary.append("Approved edits:")
-            for edit in approved:
-                summary.append(
-                    f'Block {edit.block_idx}: "{edit.old_content}" -> "{edit.new_content}"'
-                )
-
-        if rejected:
-            if approved:  # Add spacing if we had approved edits
-                summary.append("")
-            summary.append("Rejected edits:")
-            for edit in rejected:
-                summary.append(f'Block {edit.block_idx}: "{edit.old_content}"')
-
-        return "\n".join(summary)
-
-    async def aget_events(
-        self, last_sequence: Optional[int] = None
-    ) -> List[ReportEventItemEventData_Progress]:
-        """Get all events for this report asynchronously.
-
-        Args:
-            last_sequence: If provided, only get events after this sequence number
-
-        Returns:
-            List of ReportEvent objects
-        """
-        extra_params = []
-        if last_sequence is not None:
-            extra_params.append(f"last_sequence={last_sequence}")
-
-        url = await self._build_url(
-            f"/api/v1/reports/{self.report_id}/events", extra_params
-        )
-
-        response = await self.aclient.get(url, headers=self._headers)
-        response.raise_for_status()
-        progress_events = []
-        for event in response.json():
-            if event["event_type"] == "progress":
-                progress_events.append(
-                    ReportEventItemEventData_Progress(**event["event_data"])
-                )
-
-        return progress_events
-
-    def get_events(
-        self, last_sequence: Optional[int] = None
-    ) -> List[ReportEventItemEventData_Progress]:
-        """Synchronous wrapper for getting report events."""
-        return self._run_sync(self.aget_events(last_sequence))

py/llama_parse/pyproject.toml

@@ -11,13 +11,13 @@ dev = [
 
 [project]
 name = "llama-parse"
-version = "0.6.63"
+version = "0.6.70"
 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>=0.6.63"]
+dependencies = ["llama-cloud-services>=0.6.70"]
 
 [project.scripts]
 llama-parse = "llama_parse.cli.main:parse"

py/package.json

@@ -0,0 +1,7 @@
+{
+  "name": "llama-cloud-services-py",
+  "version": "0.6.70",
+  "private": "true",
+  "license": "MIT",
+  "scripts": {}
+}

py/pyproject.toml

@@ -19,7 +19,7 @@ dev = [
 
 [project]
 name = "llama-cloud-services"
-version = "0.6.63"
+version = "0.6.70"
 description = "Tailored SDK clients for LlamaCloud services."
 authors = [{name = "Logan Markewich", email = "logan@runllama.ai"}]
 requires-python = ">=3.9,<4.0"
@@ -27,7 +27,7 @@ readme = "README.md"
 license = "MIT"
 dependencies = [
   "llama-index-core>=0.12.0",
-  "llama-cloud==0.1.37",
+  "llama-cloud==0.1.43",
   "pydantic>=2.8,!=2.10",
   "click>=8.1.7,<9",
   "python-dotenv>=1.0.1,<2",

py/tests/beta/agent_data/test_agent_data_client.py

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

py/tests/extract/test_benchmark.py

@@ -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)

py/tests/extract/util.py

@@ -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:

py/tests/index/test_index.py

@@ -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,

py/tests/report/test_llama_report.py

@@ -1,129 +0,0 @@
-import os
-import pytest
-import uuid
-from typing import AsyncGenerator
-from pytest_asyncio import fixture as async_fixture
-from llama_cloud_services.report import LlamaReport, ReportClient
-
-# Skip tests if no API key is set
-pytestmark = pytest.mark.skipif(
-    not os.getenv("LLAMA_CLOUD_API_KEY") or os.getenv("CI") == "true",
-    reason="No API key provided",
-)
-
-
-@async_fixture(scope="function")
-async def client() -> AsyncGenerator[LlamaReport, None]:
-    """Create a LlamaReport client."""
-    client = LlamaReport()
-    reports_before = await client.alist_reports()
-    reports_before_ids = [r.report_id for r in reports_before]
-    try:
-        yield client
-    finally:
-        # clean up reports
-        try:
-            reports_after = await client.alist_reports()
-            reports_after_ids = [r.report_id for r in reports_after]
-            for report_id in reports_before_ids:
-                if report_id not in reports_after_ids:
-                    await client.adelete_report(report_id)
-        except Exception:
-            pass
-        finally:
-            await client.aclient.aclose()
-
-
-@pytest.fixture(scope="function")
-def unique_name() -> str:
-    """Generate a unique report name."""
-    return f"test-report-{uuid.uuid4()}"
-
-
-@async_fixture(scope="function")
-async def report(
-    client: LlamaReport, unique_name: str
-) -> AsyncGenerator[ReportClient, None]:
-    """Create a report."""
-    report = await client.acreate_report(
-        name=unique_name,
-        template_text=(
-            "# [Some title]\n\n"
-            " ## TLDR\n"
-            "A quick summary of the paper.\n\n"
-            "## Details\n"
-            "More details about the paper, possible more than one section here.\n"
-        ),
-        input_files=["tests/test_files/paper.md"],
-    )
-    try:
-        yield report
-    finally:
-        await report.adelete()
-
-
-@pytest.mark.asyncio
-@pytest.mark.xfail(
-    condition=lambda: os.getenv("CI"),
-    reason="Backend db issues; needs to be fixed.",
-)
-async def test_create_and_delete_report(
-    client: LlamaReport, report: ReportClient
-) -> None:
-    """Test basic report creation and deletion."""
-    # Verify the report exists
-    metadata = await report.aget_metadata()
-    assert metadata.name == report.name
-
-    # Test listing reports
-    reports = await client.alist_reports()
-    assert any(r.report_id == report.report_id for r in reports)
-
-    # Test getting report by ID
-    fetched_report = await client.aget_report(report.report_id)
-    assert fetched_report.report_id == report.report_id
-    assert fetched_report.name == report.name
-
-
-@pytest.mark.asyncio
-@pytest.mark.xfail(
-    condition=lambda: os.getenv("CI"),
-    reason="Report plan sometimes times out",
-    raises=TimeoutError,
-)
-async def test_report_plan_workflow(report: ReportClient) -> None:
-    """Test the report planning workflow."""
-    # Wait for the plan
-    plan = await report.await_for_plan()
-    assert plan is not None
-
-    # Approve the plan
-    response = await report.aupdate_plan(action="approve")
-    assert response is not None
-
-    # Wait for completion
-    completed_report = await report.await_completion()
-    assert len(completed_report.blocks) > 0
-
-    # Get edit suggestions
-    suggestions = await report.asuggest_edits(
-        "TLDR section header more formal.", auto_history=True
-    )
-    assert len(suggestions) > 0
-
-    # Test accepting an edit
-    await report.aaccept_edit(suggestions[0])
-
-    # Get more suggestions and test rejecting
-    more_suggestions = await report.asuggest_edits(
-        "Add a section about machine learning.", auto_history=True
-    )
-    assert len(more_suggestions) > 0
-    await report.areject_edit(more_suggestions[0])
-
-    # Verify chat history is maintained
-    assert len(report.chat_history) >= 4  # 2 user messages + 2 assistant responses
-
-    # get events
-    events = await report.aget_events()
-    assert len(events) > 0

py/unit_tests/beta/agent/test_agent_data_client_with_mocks.py

@@ -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"}

py/unit_tests/beta/agent/test_agent_data_schema.py

@@ -38,7 +38,7 @@ def test_typed_agent_data_from_raw():
     """Test TypedAgentData.from_raw class method."""
     raw_data = AgentData(
         id="456",
-        agent_slug="extraction-agent",
+        deployment_name="extraction-agent",
         collection="employees",
         data={"name": "Jane Smith", "age": 25, "email": "jane@company.com"},
         created_at=datetime.now(),
@@ -48,7 +48,7 @@ def test_typed_agent_data_from_raw():
     typed_data = TypedAgentData.from_raw(raw_data, Person)
 
     assert typed_data.id == "456"
-    assert typed_data.agent_url_id == "extraction-agent"
+    assert typed_data.deployment_name == "extraction-agent"
     assert typed_data.collection == "employees"
     assert typed_data.data.name == "Jane Smith"
     assert typed_data.data.age == 25
@@ -56,10 +56,10 @@ def test_typed_agent_data_from_raw():
 
 
 def test_typed_agent_data_from_raw_validation_error():
-    """Test TypedAgentData.from_raw with invalid data."""
+    """Test TypedAgentData.from_raw with invalid data now raises InvalidTypedAgentData."""
     raw_data = AgentData(
         id="789",
-        agent_slug="test-agent",
+        deployment_name="test-agent",
         collection="people",
         data={"name": "Invalid Person", "age": "not_a_number"},  # Invalid age
         created_at=datetime.now(),
@@ -613,3 +613,51 @@ def test_parses_field_metadata_with_error_field():
     }
     assert parsed.metadata.get("field_errors") == "This is an error"
     assert parsed.metadata.get("job_id") == "job-123"
+
+
+REASONING_IN_SCHEMA = {
+    "majority_opinion": {
+        "type": {
+            "citation": [
+                {
+                    "page": 4,
+                    "matching_text": "BARRETT, J., delivered the opinion for a unanimous Court.",
+                },
+                {"page": 11, "matching_text": "Opinion of the Court"},
+            ],
+            "parsing_confidence": 1.0,
+            "extraction_confidence": 0.9999998919950147,
+            "confidence": 0.9999998919950147,
+        },
+        "reasoning": {
+            "citation": [
+                {
+                    "page": 15,
+                    "matching_text": "We hold that §5110(b)(1) is not subject to equitable tolling and affirm the judg...",
+                }
+            ],
+            "parsing_confidence": 1.0,
+            "extraction_confidence": 0.414292785946868,
+            "confidence": 0.414292785946868,
+        },
+    },
+    "reasoning": {
+        "citation": [
+            {
+                "page": 15,
+                "matching_text": "We hold that §5110(b)(1) is not subject to equitable tolling and affirm the judg...",
+            }
+        ],
+        "parsing_confidence": 1.0,
+        "extraction_confidence": 0.414292785946868,
+        "confidence": 0.414292785946868,
+    },
+}
+
+
+def test_field_conflict_in_schema():
+    extracted = parse_extracted_field_metadata(REASONING_IN_SCHEMA)
+    assert isinstance(extracted["reasoning"], ExtractedFieldMetadata)
+    assert isinstance(
+        extracted["majority_opinion"]["reasoning"], ExtractedFieldMetadata
+    )

py/unit_tests/extract/test_extract.py

@@ -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

py/unit_tests/index/test_unit_index.py

@@ -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

report.md

@@ -1,101 +0,0 @@
-# LlamaReport (beta/invite-only)
-
-LlamaReport is a prebuilt agentic report builder that can be used to build reports from a variety of data sources.
-
-The python SDK for interacting with the LlamaReport API. The SDK provides two main classes:
-
-- `LlamaReport`: For managing reports (create, list, delete)
-- `ReportClient`: For working with a specific report (editing, approving, etc.)
-
-## Quickstart
-
-```bash
-pip install llama-cloud-services
-```
-
-```python
-from llama_cloud_services import LlamaReport
-
-# Initialize the client
-client = LlamaReport(
-    api_key="your-api-key",
-    # Optional: Specify project_id, organization_id, async_httpx_client
-)
-
-# Create a new report
-report = client.create_report(
-    "My Report",
-    # must have one of template_text or template_instructions
-    template_text="Your template text",
-    template_instructions="Instructions for the template",
-    # must have one of input_files or retriever_id
-    input_files=["data1.pdf", "data2.pdf"],
-    retriever_id="retriever-id",
-)
-```
-
-## Working with Reports
-
-The typical workflow for a report involves:
-
-1. Creating the report
-2. Waiting for and approving the plan
-3. Waiting for report generation
-4. Making edits to the report
-
-Here's a complete example:
-
-```python
-# Create a report
-report = client.create_report(
-    "Quarterly Analysis", input_files=["q1_data.pdf", "q2_data.pdf"]
-)
-
-# Wait for the plan to be ready
-plan = report.wait_for_plan()
-
-# Option 1: Directly approve the plan
-report.update_plan(action="approve")
-
-# Option 2: Suggest and review edits to the plan
-suggestions = report.suggest_edits(
-    "Can you add a section about market trends?"
-)
-for suggestion in suggestions:
-    print(suggestion)
-
-    # Accept or reject the suggestion
-    if input("Accept? (y/n): ").lower() == "y":
-        report.accept_edit(suggestion)
-    else:
-        report.reject_edit(suggestion)
-
-# Wait for the report to complete
-report = report.wait_for_completion()
-
-# Make edits to the final report
-suggestions = report.suggest_edits("Make the executive summary more concise")
-
-# Review and accept/reject suggestions as above
-...
-```
-
-### Getting the Final Report
-
-Once you are satisfied with the report, you can get the final report object and use the content as you see fit.
-
-Here's an example of printing out the final report:
-
-```python
-report = report.get()
-report_text = "\n\n".join([block.template for block in report.blocks])
-
-print(report_text)
-```
-
-## Additional Features
-
-- **Async Support**: All methods have async counterparts: `create_report` -> `acreate_report`, `wait_for_plan` -> `await_for_plan`, etc.
-- **Automatic Chat History**: The SDK automatically keeps track of chat history for each suggestion, unless you specify `auto_history=False` in `suggest_edits`.
-- **Custom HTTP Client**: You can provide your own `httpx.AsyncClient` to the `LlamaReport` class.
-- **Project and Organization IDs**: You can specify `project_id` and `organization_id` to use a specific project or organization.

scripts/changeset-version.py

@@ -0,0 +1,229 @@
+#!/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.
+
+"""
+
+import json
+import os
+import subprocess
+import sys
+from pathlib import Path
+from typing import List
+import urllib.request
+import urllib.error
+
+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 update_python_versions(version: str) -> None:
+    """llama-cloud-services and llama-parse share a version. llama-parse is just a silly sidecar that proxies to llama-cloud-services
+    for compatibility.
+
+    This function updates the version in both pyproject.toml files.
+    """
+    # Update main pyproject.toml
+    main_path = Path("py/pyproject.toml")
+    main_content = main_path.read_text()
+    main_doc = tomlkit.parse(main_content)
+    if main_doc["project"]["version"] != version:
+        click.echo(f"Updating llama-cloud-services version to {version}")
+        main_doc["project"]["version"] = version
+    main_path.write_text(tomlkit.dumps(main_doc))
+
+    # Update llama_parse/pyproject.toml
+    parse_path = Path("py/llama_parse/pyproject.toml")
+    parse_content = parse_path.read_text()
+    parse_doc = tomlkit.parse(parse_content)
+    if parse_doc["project"]["version"] != version:
+        click.echo(f"Updating llama-parse version to {version}")
+        parse_doc["project"]["version"] = version
+        parse_path.write_text(tomlkit.dumps(parse_doc))
+
+    # Update the dependency reference
+    dependencies = parse_doc["project"]["dependencies"]
+    for i, dep in enumerate(dependencies):
+        if isinstance(dep, str) and dep.startswith("llama-cloud-services"):
+            dependencies[i] = f"llama-cloud-services>={version}"
+            break
+
+    parse_path.write_text(tomlkit.dumps(parse_doc))
+
+    click.echo(f"Updated Python packages to version {version}")
+
+
+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, and propagate them to Python packages."""
+    # First, run changeset version to update all package.json files (including py/package.json)
+    _run_command(["npx", "@changesets/cli", "version"])
+
+    # Get the updated Python package version from py/package.json (updated by changesets)
+    py_package_path = Path("py/package.json")
+    if not py_package_path.exists():
+        click.echo("Python package.json not found", err=True)
+        sys.exit(1)
+
+    with open(py_package_path) as f:
+        py_package = json.load(f)
+
+    new_version = py_package["version"]
+    # Update Python pyproject.toml files based on the package.json version
+    update_python_versions(new_version)
+
+    click.echo(f"Successfully propagated version {new_version} to all Python packages")
+
+
+@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 not os.getenv("NPM_TOKEN"):
+        click.echo("NPM_TOKEN is not set, skipping publish", err=True)
+        raise click.Abort("No token set")
+    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")
+            return
+        else:
+            _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()

scripts/version-bump.py

@@ -1,227 +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) -> None:
-    """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(version):
-        # 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")
-
-
-def tag_exists(version: str) -> bool:
-    """Check if a git tag exists."""
-    tag_name = f"v{version}"
-    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(version: str) -> None:
-    """Push a git tag."""
-    tag_name = f"v{version}"
-    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
-
-    create_if_not_exists(version)
-    if push:
-        push_tag(version)
-
-
-if __name__ == "__main__":
-    cli()

ts/llama_cloud_services/CHANGELOG.md

@@ -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`.

ts/llama_cloud_services/package.json

@@ -1,9 +1,10 @@
 {
   "name": "llama-cloud-services",
-  "version": "0.3.4",
+  "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",

ts/llama_cloud_services/scripts/download.mjs

@@ -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();

ts/llama_cloud_services/src/beta/agent/client.ts

@@ -4,6 +4,7 @@ import {
   aggregateAgentDataApiV1BetaAgentDataAggregatePost,
   createAgentDataApiV1BetaAgentDataPost,
   deleteAgentDataApiV1BetaAgentDataItemIdDelete,
+  deleteAgentDataByQueryApiV1BetaAgentDataDeletePost,
   getAgentDataApiV1BetaAgentDataItemIdGet,
   searchAgentDataApiV1BetaAgentDataSearchPost,
   updateAgentDataApiV1BetaAgentDataItemIdPut,
@@ -12,6 +13,7 @@ import {
 } from "../../client";
 import type {
   AggregateAgentDataOptions,
+  DeleteAgentDataOptions,
   SearchAgentDataOptions,
   TypedAgentData,
   TypedAgentDataItems,
@@ -25,20 +27,23 @@ import type {
 export class AgentClient<T = unknown> {
   private client: ReturnType<typeof createClient>;
   private collection: string;
-  private agentUrlId: string;
+  private deploymentName: string;
 
   constructor({
     client = defaultClient,
     collection = "default",
-    agentUrlId = "_public",
+    deploymentName = "_public",
+    agentUrlId,
   }: {
     client?: ReturnType<typeof createClient>;
     collection?: string;
+    deploymentName?: string;
+    // deprecated, use deploymentName instead
     agentUrlId?: string;
   }) {
     this.client = client;
     this.collection = collection;
-    this.agentUrlId = agentUrlId;
+    this.deploymentName = agentUrlId || deploymentName;
   }
 
   /**
@@ -48,7 +53,7 @@ export class AgentClient<T = unknown> {
     const response = await createAgentDataApiV1BetaAgentDataPost({
       throwOnError: true,
       body: {
-        agent_slug: this.agentUrlId,
+        deployment_name: this.deploymentName,
         collection: this.collection,
         data: data as Record<string, unknown>,
       },
@@ -109,6 +114,24 @@ export class AgentClient<T = unknown> {
     });
   }
 
+  /**
+   * Delete all matching agent data, returns the total number of deleted items
+   */
+  async delete(options: DeleteAgentDataOptions): Promise<number> {
+    const response = await deleteAgentDataByQueryApiV1BetaAgentDataDeletePost({
+      throwOnError: true,
+      body: {
+        deployment_name: this.deploymentName,
+        ...(this.collection !== undefined && {
+          collection: this.collection,
+        }),
+        ...(options.filter !== undefined && { filter: options.filter }),
+      },
+      client: this.client,
+    });
+    return response.data.deleted_count;
+  }
+
   /**
    * Search agent data
    */
@@ -118,7 +141,7 @@ export class AgentClient<T = unknown> {
     const response = await searchAgentDataApiV1BetaAgentDataSearchPost({
       throwOnError: true,
       body: {
-        agent_slug: this.agentUrlId,
+        deployment_name: this.deploymentName,
         ...(this.collection !== undefined && {
           collection: this.collection,
         }),
@@ -165,7 +188,7 @@ export class AgentClient<T = unknown> {
     const response = await aggregateAgentDataApiV1BetaAgentDataAggregatePost({
       throwOnError: true,
       body: {
-        agent_slug: this.agentUrlId,
+        deployment_name: this.deploymentName,
         ...(this.collection !== undefined && {
           collection: this.collection,
         }),
@@ -209,7 +232,7 @@ export class AgentClient<T = unknown> {
   private transformResponse(data: AgentData): TypedAgentData<T> {
     const result: TypedAgentData<T> = {
       id: data.id!,
-      agentUrlId: data.agent_slug,
+      deploymentName: data.deployment_name,
       data: data.data as T,
       createdAt: new Date(data.created_at!),
       updatedAt: new Date(data.updated_at!),
@@ -250,10 +273,10 @@ export interface AgentDataClientOptions {
   /** Base URL for the client */
   /** Base URL of the llama cloud api */
   baseUrl?: string;
-  /** If running in an agent runtime, optionally provide the window url to infer the agent url id */
+  /** If running in an agent runtime, optionally provide the window url to infer the deployment name */
   windowUrl?: string;
-  /** Agent URL ID for the client, if not provided, it will be inferred from the window url, or fall back to "default" */
-  agentUrlId?: string;
+  /** Deployment name for the client, if not provided, it will be inferred from the window url, or fall back to "default" */
+  deploymentName?: string;
   /** Collection name for the client, defaults to "default" */
   collection?: string;
 }
@@ -267,22 +290,25 @@ export function createAgentDataClient<T = unknown>({
   client = defaultClient,
   windowUrl,
   env,
+  deploymentName,
   agentUrlId,
   collection = "default",
 }: {
   client?: ReturnType<typeof createClient>;
   windowUrl?: string;
   env?: Record<string, string>;
+  deploymentName?: string;
+  // deprecated, use deploymentName instead
   agentUrlId?: string;
   collection?: string;
 } = {}): AgentClient<T> {
-  if (env && !agentUrlId) {
-    agentUrlId =
+  if (env && !deploymentName) {
+    deploymentName =
       env.LLAMA_DEPLOY_DEPLOYMENT_NAME ||
       env.NEXT_PUBLIC_LLAMA_DEPLOY_DEPLOYMENT_NAME ||
       env.VITE_LLAMA_DEPLOY_DEPLOYMENT_NAME;
   }
-  if (windowUrl && !agentUrlId) {
+  if (windowUrl && !deploymentName) {
     try {
       const url = new URL(windowUrl);
       const path = url.pathname;
@@ -291,17 +317,18 @@ export function createAgentDataClient<T = unknown>({
         url.hostname.includes("127.0.0.1");
       if (path.startsWith("/deployments/") && !isLocalhost) {
         // /deployments/<agent-url-id>/ui/ -> ["", "deployments", "<agent-url-id>", "ui"]
-        agentUrlId = path.split("/")[2];
+        deploymentName = path.split("/")[2];
       }
     } catch (error) {
       console.warn(
-        "Failed to infer agent url id from window url, falling back to default",
+        "Failed to infer deployment name from window url, falling back to default",
         error,
       );
     }
   }
 
   return new AgentClient({
+    ...(deploymentName && { deploymentName }),
     ...(agentUrlId && { agentUrlId }),
     collection,
     client,

ts/llama_cloud_services/src/beta/agent/types.ts

@@ -87,8 +87,8 @@ export interface ExtractedData<T = unknown> {
 export interface TypedAgentData<T = unknown> {
   /** The unique ID of the agent data record. */
   id: string;
-  /** The ID of the agent that created the data. */
-  agentUrlId: string;
+  /** The deployment name of the agent that created the data. */
+  deploymentName: string;
   /** The collection of the agent data. */
   collection?: string;
   /** The data of the agent data. Usually an ExtractedData&lt;SomeOtherType&gt; */
@@ -127,6 +127,14 @@ export interface SearchAgentDataOptions {
   includeTotal?: boolean;
 }
 
+/**
+ * Options for deleting agent data
+ */
+export interface DeleteAgentDataOptions {
+  /** Filter options for the deletion. */
+  filter?: Record<string, FilterOperation>;
+}
+
 /**
  * Options for aggregating agent data
  */

ts/llama_cloud_services/src/reader.ts

@@ -147,6 +147,10 @@ export class LlamaParseReader extends FileReader {
   output_s3_region?: string | undefined;
   preserve_layout_alignment_across_pages?: boolean | undefined;
   spreadsheet_extract_sub_tables?: boolean | undefined;
+  specialized_chart_parsing_agentic?: boolean | undefined;
+  specialized_chart_parsing_efficient?: boolean | undefined;
+  specialized_chart_parsing_plus?: boolean | undefined;
+  precise_bounding_box?: boolean | undefined;
   formatting_instruction?: string | undefined;
   parse_mode?: ParsingMode | undefined;
   system_prompt?: string | undefined;
@@ -331,6 +335,11 @@ export class LlamaParseReader extends FileReader {
       preserve_layout_alignment_across_pages:
         this.preserve_layout_alignment_across_pages,
       spreadsheet_extract_sub_tables: this.spreadsheet_extract_sub_tables,
+      specialized_chart_parsing_agentic: this.specialized_chart_parsing_agentic,
+      specialized_chart_parsing_efficient:
+        this.specialized_chart_parsing_efficient,
+      specialized_chart_parsing_plus: this.specialized_chart_parsing_plus,
+      precise_bounding_box: this.precise_bounding_box,
       formatting_instruction: this.formatting_instruction,
       parse_mode: this.parse_mode,
       system_prompt: this.system_prompt,

ts/llama_cloud_services/src/schema.ts

@@ -87,6 +87,10 @@ export const parseFormSchema = z.object({
   preserve_layout_alignment_across_pages: z.boolean().optional(),
   skip_diagonal_text: z.boolean().optional(),
   spreadsheet_extract_sub_tables: z.boolean().optional(),
+  specialized_chart_parsing_agentic: z.boolean().optional(),
+  specialized_chart_parsing_efficient: z.boolean().optional(),
+  specialized_chart_parsing_plus: z.boolean().optional(),
+  precise_bounding_box: z.boolean().optional(),
   structured_output: z.boolean().optional(),
   structured_output_json_schema: z.string().optional(),
   structured_output_json_schema_name: z.string().optional(),

ts/llama_cloud_services/tests/agent_client.test.ts

@@ -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");
+  });
+});