A Pi agent extension
Elohia/pi-mm-vision · model-adapters
A general coding device — allows pure text models to acquire pixel image recognition through structured spatial text: visual models look at original maps and output compact coordinates (paints/elements/ percentage coordinates/shares/values/relations) into the text model, which recreates images with words, delineates position relationships.
★ 2GitHub stars
0Forks
2026-08-14Last updated
TypeScriptLanguage
MITLicense
Key features
- Perceptive encodingcanvas / element / (x%, y%) coordinates / shape / value / relationship, structured output
- mm-vision drew out the drawings, ugly.
- Only send pictures to the visual model that you've configured.
- Selective array mode (dotMatrixtrue): Add pixel level ASCII dots (with grid ruler) to the curve-specific scene
Install command
dsh plugin --profile web add github:Elohia/pi-mm-vision