README Excerpt
A Model Context Protocol (MCP) server providing vision & video analysis tools, **configurable with any model provider**. This is a reverse-engineered and extended reimplementation of `@z_ai/mcp-server` (Apache-2.0, credit to Chao Gong, Lei Yuan / Z.AI). It introduces a **provider abstraction layer** so the same set of tools can run against any of three API families:
Tools (20)
VLM_ANTHROPIC_VERSIONVLM_API_KEYVLM_BASE_URLVLM_ENABLE_THINKINGVLM_LOG_PATHVLM_PROVIDERVLM_RETRY_COUNTVLM_TIMEOUTVLM_VISION_MODELVLM_VISION_MODEL_MAX_TOKENSVLM_VISION_MODEL_TEMPERATUREVLM_VISION_MODEL_TOP_Panalyze_data_visualizationanalyze_imageanalyze_videoanthropicautodiagnose_error_screenshotextract_text_from_screenshotui_diff_check