Tool
modlens
liustack/modlens
What it is
A vision bridge for coding agents running text-only models, turning pasted images into structured JSON covering text, layout and meaning. It installs as an agent skill or as a DeepSeek Harness plugin.
Who it might suit
People running text-only models inside a coding agent who want to paste screenshots and diagrams into the conversation.
What to check first
Images go to an external vision engine, such as a Gemini, OpenAI-compatible or Anthropic endpoint, or through a signed-in CLI whose login it may reuse. Its documentation lists image content as untrusted input. One optional install step pipes a script into bash.
How it runs
Agent skill, plugin and command-line tool
Written