Read images even with a text-only AI model.
Provides evidence-based image answers for models that cannot see images.
When to use it
Use when an image's content is unavailable to the model or the user asks to configure modlens; skip images the model can already see.
Give it an image the model cannot see; it returns an answer grounded in structured image evidence and identifies uncertainty.
What you provide
No setup requirements listed.
This skill
modlens
Submits images for provider analysis
image URLs
Reads images from URLs
@liustack/modlens
The pinned modlens package
~/.claude/image-cache/<session-id>/
Reads Claude Code cached images