关于
Vision Relay wraps locally installed multimodal CLIs, specifically Google Gemini CLI and Qwen CLI, into a Model Context Protocol (MCP) server. This innovative approach allows MCP-compatible tools, such as Claude and Codex, to directly leverage the powerful image analysis features of these underlying models. It offers an efficient and cost-effective solution for enriching the functionality of text-only AI deployments with visual understanding, providing multimodal capabilities without the need to switch primary AI providers.
主要功能
- Extensible provider relay architecture to switch or expand different CLIs while retaining configuration.
- Actionable outputs that regularize stdout and append metadata for client-side display or further processing.
- Robust input handling with automatic image size/extension validation, temporary file management, and cleanup.
- Unified image analysis tools (gemini_analyze_image, qwen_analyze_image) supporting local paths, URLs, and base64 strings.
- Configurable execution options including sandbox mode, timeouts, extra flags, and model overrides via .env or environment variables.
- 2 GitHub stars
使用案例
- Enable vision capabilities for text-only AI models (e.g., Claude, Codex) integrated via MCP.
- Integrate Google Gemini or Qwen image analysis into existing Model Context Protocol workflows.
- Provide an inexpensive pathway for AI deployments to gain multimodal features without changing core providers.
Sponsored