VisionAgent MCP Server is a lightweight side-car MCP server that runs locally on STDIN/STDOUT, translating each tool call from an MCP-compatible client (Claude Desktop, Cursor, Cline, etc.) into authenticated HTTPS requests to Landing AI’s VisionAgent REST APIs. The response JSON, plus any images or masks, is streamed back to the model so that you can issue natural-language computer-vision and document-analysis commands from your editor without writing custom REST code or loading an extra SDK. v0.1 adds support for agentic-document-analysis (PDFs/images text extraction), text-to-object-detection (bounding boxes), text-to-instance-segmentation (pixel masks), activity-recognition (video actions with timestamps), and depth-pro (monocular depth maps). The server validates inputs with Zod schemas derived from the live OpenAPI, auto-generates tool definitions via generate-tools, reads file-based inputs, and forwards authenticated requests to VisionAgent. Outputs can be saved and visualized to OUTPUT_DIRECTORY when IMAGE_DISPLAY_ENABLED is true. It runs locally, with no telemetry, and is configurable via a minimal JSON config and sample MCP client entries.