Gimmick Vision

Bridges Claude Code to local OpenAI-compatible vision models for image analysis and OCR.

Provides three vision analysis tools that route requests to locally-hosted vision models like Qwen2.5-VL through llama.cpp, Ollama, or vLLM. Supports single image analysis, multi-image comparison, and OCR text extraction without cloud API calls. Available as a Docker image with configurable endpoints and optional integration with gimmick-search control panel.