OCR and Documents

Recover text from scanned or image-heavy documents before attempting structured analysis or downstream writing tasks.

agentic-in Updated 586 repo stars
Elephant Agent follows a personal path with people, places, risks, rhythms, decisions, and a Personal Model

File contents

OCR and Documents

Use this skill when a document is not directly machine-readable.

Preferred Flow

  1. Determine whether the file is text-native or scanned.
  2. Run OCR first when the text layer is missing or broken.
  3. Preserve page order, tables, and obvious headings where possible.
  4. Only after extraction should you summarize or transform the content.

Guardrails

  • State clearly when OCR confidence is low.
  • Do not fabricate unreadable text.

agentic-in/elephant-agent commit d7f964d804

Frequently asked questions

npx get-skill add agentic-in/ocr-and-documents