M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
D

DINO-X MCP Server

Official
Official MCP server powered by DINO-X for fine-grained object detection and image understanding.
How to install ↓ Compare GitHub source repository ↗
Category
Other
Stars
★ 111 Popular
Transport
stdio (local process) · Streamable HTTP
Runtime
Node.js · Hosted (no local runtime)
Credentials
API key / credential required
License
Apache-2.0
Last commit
⚠
Tools
4
54FMRS · D

DINO-X MCP Server is a feature-rich, officially maintained object detection server offering multiple tools and flexible deployment options. It is suitable for developers to quickly integrate vision capabilities, but attention must be paid to API key security and data privacy.

Strongest · Documentation 13/20 Weakest · Reliability 8/20

Reliability
8/20
Security and permissions
10/20
Maintenance
11/20
Documentation
13/20
Setup experience
12/20
Why each score
Reliability 8/20
Source code shows complete TypeScript implementation with tool definitions and HTTP/STDIO transports, plus build scripts. Tests exist but are limited (unit only, no integration) and CI workflows are not configured, so reliability score is capped.
Security and permissions 10/20
API keys are provided via environment variables or command-line arguments, with optional auth token. No evidence of malicious code or data exfiltration, but lack of detailed key handling documentation and explicit least-privilege statements.
Maintenance 11/20
Repository has recent commits, clear license (Apache-2.0), and version releases. However, no security response channel is evident, and issue tracking and dependency update details are limited.
Documentation 13/20
README provides detailed installation, configuration, tool descriptions, and use cases, but lacks parameter specifications, limitations, and troubleshooting sections. Documentation is usable but not deep.
Setup experience 12/20
Clear instructions for installation via npm and from source, with config examples for mainstream clients. However, manual API key acquisition is required, and no automated setup scripts or detailed platform compatibility notes.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessWrites / deletes local filesUses the network

Best for

  • Developers building multimodal AI applications that require object detection and image understanding.
  • Vision QA tasks requiring precise object localization and attribute reasoning.
  • Scenarios needing rapid deployment of local or cloud vision perception services.

Not for

  • Scenarios processing video streams (currently only static images are supported).
  • Offline operation (requires API key and network connectivity).
  • Users with strict privacy requirements who cannot send images to external APIs.

Required permissions

  • Requires access to remote APIs (sending images to the DINO-X cloud platform).
  • In STDIO mode, may require local filesystem write permissions to save annotated images.

Risks and side effects

  • API key leakage: keep your API key secure and avoid exposing it in public repositories.
  • Data privacy: images are sent to the DINO-X platform for processing; ensure compliance with your data policies.
  • Dependency on external service: the server relies on the DINO-X cloud platform; network outages or service disruptions may affect availability.

Setup

Before you start

Runtime：Node.js · Hosted (no local runtime)

DINOX_API_KEY requiredsecret API key for the DINO-X platform, requested at https://cloud.deepdataspace.com/request_api (new users get free quota).
AUTH_TOKEN optionalsecret Optional, HTTP mode only; when set, clients must send Authorization: Bearer <token> to connect.
Other optional settings (1)
IMAGE_STORAGE_DIRECTORY optional Optional, STDIO only; local directory where annotated images are saved.
  1. Ensure Node.js is installed (LTS recommended).
  2. Get an API key from the DINO-X platform.
  3. Choose a configuration method:
  • Official hosted Streamable HTTP: Add the URL to your MCP client config, e.g., https://mcp.deepdataspace.com/mcp?key=your-api-key.
  • Local npm package (STDIO): Configure your MCP client with the npx command and set the DINOX_API_KEY environment variable.
  • Run from source: Clone the repository, install dependencies, build, and run with node.

Check that it works

Check that the tools detect-all-objects, detect-objects-by-text, detect-human-pose-keypoints, and visualize-detection-result appear in your MCP client's tool list; sending a request like "detect all objects in this image" that returns categories and bounding boxes confirms the setup works.

Troubleshooting

  1. If connection fails, check the API key and network access to the DINO-X platform.
  2. If images cannot be processed, ensure the image format is jpg, jpeg, webp, or png, and that the URL is valid.
  3. If annotated images cannot be saved in STDIO mode, check the `IMAGE_STORAGE_DIRECTORY` environment variable and write permissions.
  4. In HTTP mode, verify the port is not in use and check firewall settings.

Things to try

Once connected, you can ask your AI assistant things like:

  • Detect all objects in this image and give me their locations
  • Find all person.car in the image with bounding boxes
  • Analyze the human pose keypoints in this image
  • Visualize the detection results and save the annotated image

Tools 4

detect-all-objects read-only
Detect all objects in the image, outputting categories, bounding boxes, and optional captions.
detect-objects-by-text read-only
Detect objects based on English noun prompts, supporting dot-separated multiple categories, outputting bounding boxes and optional captions.
detect-human-pose-keypoints read-only
Estimate human pose keypoints in the image, outputting 17 keypoints and bounding box, with optional captions.
visualize-detection-result writes
Visualize detection results by drawing on the image and saving the annotated image locally (STDIO only).

Use cases

Detect and visualize fire areas in a forest image.
Count the number of cardboard boxes in a warehouse image.
Find all red cars in an image.
Identify the tallest person and describe their clothing.
Full-scene detection, such as finding the fruit with the highest vitamin C content.
Analyze yoga poses from images.

Overview

DINO-X MCP Server is the official Model Context Protocol (MCP) server that empowers LLMs with real-world visual perception through image object detection, localization, and captioning APIs. It leverages the DINO-X and Grounding DINO models to provide fine-grained object detection and image understanding for multimodal applications. The server supports STDIO and Streamable HTTP transport modes and offers tools for full-scene object detection, text-prompted object detection, human pose estimation, and visualization.

Similar servers

Telegram MCP Server 70 · B

Telegram MCP server powered by Telethon to let MCP clients read chats, manage groups, and send/modify messages, media, contacts, and settings.

★ 1.8k · Tools 76 Compare with this →
Compare with other MCP servers

Source revision 83be2e9f02b1 Data synced 2026-10-12 Read the FMRS scoring method