AI Agent Hub
Back to skills
🎨

OCR Text Recognition

Design & Media Updated 2026.08.30

Paste the following prompt into your AI chat to install this skill:

Please install @user_628f4cb3/ocr according to https://skillhub.cn/install/skillhub.md.

About this skill

Problem

Text often appears in screenshots, photographed documents, or signage photos. The practical issue is not only model inference, but also input format, authentication, missing parameters, and result normalization.

How It Works

OCR Text Recognition takes an image as input, detects text, and returns recognized content. It is suitable for documents, signage, and screenshots. The workflow usually has three parts:
- Tool selection: use scripts.tools.ocr for an image URL, or scripts.tools.ocr_for_data_base64 for Base64 image data.
- Parameter handling: ocr requires dataUrl; ocr_for_data_base64 requires dataBase64. Missing parameters should be clarified before invocation.
- Result handling: the response includes success, message, and raw. The raw field contains the API payload and can be formatted for display or downstream processing.

Boundaries

This skill requires XBY_APIKEY. If the key is missing, it should be requested from the user rather than fabricating results. It works best with image URLs or Base64 payloads; local private files must first become accessible URLs or Base64 data. If success is false, inspect message and raw to diagnose the failure.

Use Cases

  • Extract error codes and stack text from a screenshot URL before pasting it into a support ticket.
  • Convert a customer's Base64 contract image into editable text for checking clause fields.
  • Recognize a store signage photo to capture the shop name and phone number for a data entry sheet.
  • Pull parameter names and values from a console screenshot to reproduce an issue.

Best For

  • Support engineers who need tickets: extract error codes, parameters, and log text from screenshots.
  • Operations staff handling customer uploads: convert contract or proof images into editable text for field entry.
  • Marketing staff collecting offline store data: recognize shop names, phone numbers, and addresses from signage photos.
  • Test engineers reproducing frontend issues: pull console parameters, errors, and request values from screenshots.