Skip to content
Get packs

Universal Document Reader

Lets an agent read PDFs, Word, Excel, PowerPoint, images (OCR), and text or code files via one tool.

Source Download skill
Repository
HKUDS/Vibe-Trading
Stars
37,433
Licence
MIT
Files
1 file

What it does

Once installed, the agent can open most common document and data files (PDF, Word, Excel, PowerPoint, images, CSV, plain text, config and source code) and pull out their text through a single read_document tool. For scanned pages or images it can run OCR (optical character recognition) to read text from pictures, either locally or by sending the pages to a cloud vision model. Long content is truncated to about 15,000 characters, so large PDFs can be read in page ranges.

Great for

  • Anyone who wants an agent to summarise a report, paper or contract from a PDF or Word file
  • Analysts who want a quick look at the sheets in an Excel workbook
  • Developers who want an agent to read config files or source code without reformatting them

Skip it if

  • Full analysis of large spreadsheets: previews are capped at 100 rows per sheet, so use the companion analyze_trade_journal tool or another method for complete data
  • Sensitive documents when OCR is set to llm-vision: page images are sent to a cloud LLM provider, which may conflict with confidentiality rules
  • Archives such as zip or tar files and executables, which are blocked and must be unpacked locally first

Try asking

  • “Summarise the methodology and conclusions of uploads/paper.pdf”
  • “Review uploads/contract.docx and flag termination, liability and payment clauses”
  • “Read the first 10 pages of uploads/annual_report.pdf”
  • “Extract the text from uploads/scan.png and tell me if OCR returns nothing”
Step-by-step guide: how to use Universal Document ReaderInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    doc-reader.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip doc-reader.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip doc-reader.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Set VIBE_TRADING_OCR_ENGINE=auto (the default) if you want document pages to stay on your machine; choose llm-vision only when cloud processing is acceptable.
  • For long PDFs, pass a pages range such as pages="1-10" to read in slices within the 15,000-character limit.
  • If OCR returns empty text, install rapidocr_onnxruntime or configure a vision model rather than relying on guessed content.

What's inside

  • SKILL.md

Redistributed unmodified from HKUDS/Vibe-Trading under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.