DSH Vision Tools

Key features

• View understand toolCall OpenAI compatible visual model API to understand local pictures (description images, identification text, answer questions), register as a global tool and all sessions are available.
• Pictures are sent via external visual API (base64 transmission) for sensitive data using local OCR (testeract/paddleocr), or self-built Intranet visual services pointing to VISION BASE URL
• POST /api/vision-paste router pasting picture
• Three-entry mappingCmd/Ctrl+V paste, drag to the button, dot selected file ⁇ Picture drops automatically into $DSH HOME/pasted-images/ ⁇ Enter box to identify this image: ⁇ The auto-moveration of the map tool after sending.
• View.env with API Key, do not submit to git (this repository .gitignore ignored)