PaperclipOrg
Claude Agent Skill · by Run Llama

Liteparse

Install Liteparse skill for Claude Code from run-llama/llamaparse-agent-skills.

Install
Terminal · npx
$npx skills add https://github.com/vercel-labs/agent-skills --skill vercel-react-best-practices
Works with Paperclip

How Liteparse fits into a Paperclip company.

Liteparse drops into any Paperclip agent that handles this kind of work. Assign it to a specialist inside a pre-configured PaperclipOrg company and the skill becomes available on every heartbeat — no prompt engineering, no tool wiring.

S
SaaS FactoryPaired

Pre-configured AI company — 18 agents, 18 skills, one-time purchase.

$27$59
Explore pack
Source file
SKILL.md222 lines
Expand
---name: liteparsedescription: Use this skill when the user asks to parse, perform multi-format document conversion or spatially extract text from an unstructured file (PDF, DOCX, PPTX, XLSX, images, etc.) locally without cloud dependencies.compatibility: Requires Node 18+ and `@llamaindex/liteparse` installed globally via npm (`npm i -g @llamaindex/liteparse`)license: MITmetadata:  author: LlamaIndex  version: "0.1.0"--- # LiteParse Skill Parse unstructured documents (PDF, DOCX, PPTX, XLSX, images, and more) locally with LiteParse: fast, lightweight, no cloud dependencies or LLM required. ## Initial Setup When this skill is invoked, respond with: ```I'm ready to use LiteParse to parse files locally. Before we begin, please confirm that: - `@llamaindex/liteparse` is installed globally (`npm i -g @llamaindex/liteparse`)- The `lit` CLI command is available in your terminal If both are set, please provide: 1. One or more files to parse (PDF, DOCX, PPTX, XLSX, images, etc.)2. Any specific options: output format (json/text), page ranges, OCR preferences, DPI, etc.3. What you'd like to do with the parsed content. I will produce the appropriate `lit` CLI command or TypeScript script, and once approved, report the results.``` Then wait for the user's input. --- ## Step 0 — Install LiteParse (if needed) If `liteparse` is not yet installed, install it globally: ```bashnpm i -g @llamaindex/liteparse``` Verify installation: ```bashlit --version``` For Office document support (DOCX, PPTX, XLSX), LibreOffice is required: ```bash# macOSbrew install --cask libreoffice # Ubuntu/Debianapt-get install libreoffice``` For image parsing, ImageMagick is required:```bash# macOSbrew install imagemagick # Ubuntu/Debianapt-get install imagemagick``` --- ## Step 1 — Produce the CLI Command or Script ### Parse a Single File ```bash# Basic text extractionlit parse document.pdf # JSON output saved to a filelit parse document.pdf --format json -o output.json # Specific page rangelit parse document.pdf --target-pages "1-5,10,15-20" # Disable OCR (faster, text-only PDFs)lit parse document.pdf --no-ocr # Use an external HTTP OCR server for higher accuracylit parse document.pdf --ocr-server-url http://localhost:8828/ocr # Higher DPI for better qualitylit parse document.pdf --dpi 300``` ### Batch Parse a Directory ```bashlit batch-parse ./input-directory ./output-directory # Only process PDFs, recursivelylit batch-parse ./input ./output --extension .pdf --recursive``` ### Generate Page Screenshots Screenshots are useful for LLM agents that need to see visual layout. ```bash# All pageslit screenshot document.pdf -o ./screenshots # Specific pageslit screenshot document.pdf --pages "1,3,5" -o ./screenshots # High-DPI PNGlit screenshot document.pdf --dpi 300 --format png -o ./screenshots # Page rangelit screenshot document.pdf --pages "1-10" -o ./screenshots``` --- ## Step 3 — Key Options Reference ### OCR Options | Option | Description ||--------|-------------|| (default) | Tesseract.js — zero setup, built-in || `--ocr-language fra` | Set OCR language (ISO code) || `--ocr-server-url <url>` | Use external HTTP OCR server (EasyOCR, PaddleOCR, custom) || `--no-ocr` | Disable OCR entirely | ### Output Options | Option | Description ||--------|-------------|| `--format json` | Structured JSON with bounding boxes || `--format text` | Plain text (default) || `-o <file>` | Save output to file | ### Performance / Quality Options | Option | Description ||--------|-------------|| `--dpi <n>` | Rendering DPI (default: 150; use 300 for high quality) || `--max-pages <n>` | Limit pages parsed || `--target-pages <pages>` | Parse specific pages (e.g. `"1-5,10"`) || `--no-precise-bbox` | Disable precise bounding boxes (faster) || `--skip-diagonal-text` | Ignore rotated/diagonal text || `--preserve-small-text` | Keep very small text that would otherwise be dropped | --- ## Step 4 — Using a Config File For repeated use with consistent options, generate a `liteparse.config.json`: ```json{  "ocrLanguage": "en",  "ocrEnabled": true,  "maxPages": 1000,  "dpi": 150,  "outputFormat": "json",  "preciseBoundingBox": true,  "skipDiagonalText": false,  "preserveVerySmallText": false}``` For an HTTP OCR server: ```json{  "ocrServerUrl": "http://localhost:8828/ocr",  "ocrLanguage": "en",  "outputFormat": "json"}``` Use with: ```bashlit parse document.pdf --config liteparse.config.json``` --- ## Step 5 — HTTP OCR Server API (Advanced) If the user wants to plug in a custom OCR backend, the server must implement: - **Endpoint**: `POST /ocr`- **Accepts**: `file` (multipart) and `language` (string) parameters- **Returns**:```json{  "results": [    { "text": "Hello", "bbox": [x1, y1, x2, y2], "confidence": 0.98 }  ]}``` Ready-to-use wrappers exist for EasyOCR and PaddleOCR in the LiteParse repo. --- ## Supported Input Formats | Category | Formats ||----------|---------|| PDF | `.pdf` || Word | `.doc`, `.docx`, `.docm`, `.odt`, `.rtf` || PowerPoint | `.ppt`, `.pptx`, `.pptm`, `.odp` || Spreadsheets | `.xls`, `.xlsx`, `.xlsm`, `.ods`, `.csv`, `.tsv` || Images | `.jpg`, `.jpeg`, `.png`, `.gif`, `.bmp`, `.tiff`, `.webp`, `.svg` | Office documents require LibreOffice; images require ImageMagick. LiteParse auto-converts these formats to PDF before parsing.