| name | image-ocr |
| description | Use this skill whenever the user sends an image and you cannot see it directly (e.g., you are running on a non-multimodal model like deepseek-v4-pro). This skill runs Tesseract OCR locally to extract text from images. Use when: the user pastes/sends an image, asks to "read" or "recognize" an image, or asks what's in a picture/screenshot. Also use when the user asks for OCR or text extraction from any image file. |
Image OCR
When
The current model cannot process images directly. When the user sends an image, this skill extracts text from it using local Tesseract OCR.
How
Run scripts/ocr_image.py <image_path> with appropriate language and PSM settings:
- Default:
chi_sim+eng with PSM 3 (auto page segmentation)
- For screenshots with dense text: PSM 6
- For sparse text (book covers, signs): PSM 11
- English only:
--lang eng
- Chinese only:
--lang chi_sim
PSM modes
| PSM | Description | Best for |
|---|
| 3 | Fully automatic | General purpose, most screenshots |
| 4 | Single column of text | Book pages with variable text sizes |
| 6 | Uniform block of text | Dense paragraphs |
| 11 | Sparse text | Covers, posters, UI elements |
Script location
scripts/ocr_image.py — takes image path, outputs UTF-8 text to stdout.
Notes
- Chinese output may display as garbled text in the bash terminal. Write to a file (
stdout -> /tmp/ocr_result.txt) and read it instead.
- Tesseract is installed at
C:\Program Files\Tesseract-OCR\tesseract.exe
- Available languages:
chi_sim, eng, osd
