VisionAgent MCP Server

What is VisionAgent MCP Server?

VisionAgent MCP Server is a lightweight, side-car MCP server that runs locally on STDIN/STDOUT, translating tool calls from MCP‑compatible clients (Claude Desktop, Cursor, Cline) into authenticated HTTPS requests to Landing AI’s VisionAgent REST APIs. It enables natural‑language computer‑vision and document‑analysis commands without writing custom REST code or loading an extra SDK.

How to use VisionAgent MCP Server?

Install globally via npm install -g vision-tools-mcp, set the VISION_AGENT_API_KEY environment variable, then configure your MCP client with npx vision-tools-mcp as the command and the required environment variables. Prerequisites are Node.js 20 LTS, a VisionAgent account with an API key, and an MCP‑compatible client.

Key features of VisionAgent MCP Server

• Translates MCP tool calls into authenticated VisionAgent API requests.
• Supports agentic document analysis, text‑to‑object detection, text‑to‑instance segmentation, activity recognition, and depth estimation.
• Auto‑generates tool definitions from a live OpenAPI spec via npm run generate-tools.
• Renders masks, bounding boxes, and depth maps to files or inline previews.
• Validates arguments with Zod schemas derived from the OpenAPI spec.
• Outputs JSON and media results streamed back to the MCP client.

Use cases of VisionAgent MCP Server

• Extract vendor, invoice date, and total from a PDF using agentic‑document‑analysis.
• Locate every pedestrian in an image with text‑to‑object‑detection.
• Segment all tomatoes in a kitchen scene with text‑to‑instance‑segmentation.
• Identify activities occurring in a video file via activity‑recognition.
• Produce a depth map for a selfie image using depth‑pro.

FAQ from VisionAgent MCP Server

How do I get a VisionAgent API key?

Create an account at va.landing.ai and obtain your API key from the settings page.

What are the minimum requirements?

Node.js 20 LTS, a VisionAgent account (any tier) with an API key, and an MCP‑compatible client such as Claude Desktop, Cursor, or Cline.

How do I configure image display?

Set the environment variable IMAGE_DISPLAY_ENABLED to "true" for clients that support resources (e.g., Claude Desktop) or "false" for clients without image display (e.g., Cursor).

What does IMAGE_DISPLAY_ENABLED do?

When enabled, the server renders masks, boxes, and depth maps to files and provides inline previews; when disabled, only text results are returned.