# Shunwang Tech MCP for AI Agents AI Agent Connect

> Shunwang Tech is an Connector that gives your AI agent direct control over edge computing and PC Bang infrastructure. It lets you monitor GPU clusters, schedule inference tasks, and manage hardware nodes in China without jumping between different management dashboards.

## Overview
- **Category:** cloud-infrastructure
- **Price:** Free
- **Endpoint:** https://edge.vinkius.com/vk_preview_bgiyuAWhRzWrIFnn1swCL5IcEpCA8FYUUwp4FNb7/ai-agent-connect
- **Tags:** edge-computing, gpu-resource-management, cluster-monitoring, task-scheduling, distributed-systems

## Description

Managing a distributed network of edge nodes or a massive PC Bang setup usually means constantly switching tabs and checking complex metrics. You have to hunt for specific GPU specs, check if a node is hanging, and manually push tasks to different clusters. This Connector changes that by turning your infrastructure into a conversational interface. Instead of digging through a technical dashboard, you just ask your agent to tell you which nodes are under heavy load or to deploy a new inference job to a specific cluster. It handles the heavy lifting of auditing resource usage and sending remote commands like reboots. Since it's hosted on the Vinkius catalog, you can get these capabilities into your favorite AI client in minutes. You move from being a dashboard operator to a high-level orchestrator, letting your agent handle the routine checks while you focus on scaling your compute.

## Tools

### list_images
View all available system images for your nodes. This helps you see what software environments you can deploy.

### get_gpu_specs
Check the hardware specifications for your GPUs. This is useful for planning your inference capacity.

### create_task
Launch a new computing task on your infrastructure. This lets you deploy new jobs to specific clusters quickly.

### get_node
Pull detailed information for a specific node. Use this to deep dive into the status of a single machine.

### get_resource_usage
See a summary of current resource consumption. It helps you monitor how hard your clusters are working.

### list_clusters
Browse all available computing clusters in your network. Use this to organize your distributed infrastructure.

### list_nodes
Get a full list of all computing nodes in your network. Use this to see which hardware is currently online.

### list_tasks
See every computing task currently running across your clusters. It helps you track active jobs at a glance.

### send_node_command
Send a direct management command to a specific node. You can use this to reboot or shut down hardware remotely.

### stop_task
Halt a running computing task immediately. Use this to free up resources from jobs that are no longer needed.

## Prompt Examples

**Prompt:** 
```
Are there any nodes in my cluster that are currently offline or showing high GPU usage?
```

**Response:** 
```
I've checked your Shunwang network. Here is the current status of your nodes:

| Node ID | Status | GPU Load | RAM Usage |
| :--- | :--- | :--- | :--- |
| SW-001 | Online | 12% | 4GB |
| SW-002 | **Offline** | 0% | 0GB |
| SW-003 | Online | **94%** | 12GB |

Node **SW-002** is currently down, and **SW-003** is under heavy load. Would you like me to send a reboot command to SW-002?
```

**Prompt:** 
```
I need to start a new inference job on the 'East-Cluster' using the latest v2 image. Can you do that?
```

**Response:** 
```
I'm starting that deployment for you now.

**Deployment Details:**
*   **Cluster:** East-Cluster
*   **Image:** ai-inference-v2
*   **Task ID:** TSK-9921

I've successfully created the task and it is now initializing on the cluster. I'll keep an eye on it and let you know when it's live.
```

**Prompt:** 
```
Give me a summary of how much of our GPU capacity is actually being used right now.
```

**Response:** 
```
Here is the current resource summary across your infrastructure:

*   **Total GPU Capacity:** 120 units
*   **Current Utilization:** 68%
*   **Active Inference Jobs:** 45
*   **Available Capacity:** 36 units

Most of your load is concentrated in the 'Huzhou' cluster. Do you want a breakdown of usage by specific cluster?
```

## Capabilities

### Check real-time node health
Get instant status updates on CPU, GPU, and RAM for all your computing nodes.

### Deploy new computing tasks
Launch new jobs across your clusters using specific system images.

### Audit GPU resource usage
See a clear summary of how your GPU resources are being utilized across the network.

### List available system images
Browse all system images available for deployment on your infrastructure.

### Send remote reboot commands
Send direct management commands to specific nodes to clear hangs or perform maintenance.

### Browse distributed clusters
View and organize your various computing clusters in one place.

## Use Cases

### Scaling AI Inference
An AI dev asks the agent to deploy a model to the 'Huzhou' cluster using a specific image. The agent uses create_task to handle the deployment.

### Emergency Node Recovery
An admin notices a hung node and asks the agent to send a reboot command. The agent uses send_node_command to fix the hardware issue instantly.

### Capacity Planning
A manager asks for a summary of GPU usage across all clusters to see if they need more hardware. The agent pulls data via get_resource_usage.

### Cluster Audit
An engineer wants to see all active tasks and node specs to prepare for a maintenance window. The agent uses list_tasks and get_gpu_specs.

## Benefits

- Stop manually checking node health by using list_nodes to get a real-time status of your entire network.
- Deploy tasks faster by using create_task to push new jobs to specific clusters in seconds.
- Keep your hardware running smoothly by using send_node_command to reboot or shut down unresponsive nodes.
- Audit your hardware capacity with get_gpu_specs to ensure you have the right hardware for your inference needs.
- Monitor your spend and load with get_resource_usage to see exactly how your clusters are being utilized.
- Manage your software environment easily by using list_images to see what's available for deployment.

## How It Works

The bottom line is you get a natural language interface for managing complex edge computing hardware.

1. Subscribe to the Shunwang Tech MCP on Vinkius.
2. Enter your Shunwang App Key and App Secret in the configuration.
3. Start managing your nodes and tasks through your AI client.

## Frequently Asked Questions

**How can I use Shunwang Tech to manage my PC Bang?**
You can use this Connector to let your AI agent oversee your entire PC Bang network. It can check if PCs are online, monitor their health, and send remote commands to fix issues without you needing to visit the site.

**Can I use an AI agent to reboot my remote computing nodes?**
Yes, the Connector allows your agent to send direct management commands like reboot or shutdown to specific nodes. This is great for clearing up hung processes or performing remote maintenance.

**How do I deploy AI inference tasks on Shunwang Tech?**
You can simply describe the task and the image you want to use to your AI agent. The Connector will handle the creation and deployment of that task on your designated clusters automatically.

**Can Shunwang Tech help me monitor GPU usage across multiple clusters?**
Absolutely. The Connector can pull resource usage summaries across your entire network, letting you see exactly how much of your GPU capacity is being used at any given moment.

**Is there a way to manage edge computing nodes without a dashboard?**
This Connector is designed for exactly that. It moves the management from a technical dashboard into a natural conversation with your AI client, making it much faster to get things done.

**How do I get my Shunwang App Key into my AI client?**
Once you subscribe via Vinkius, you'll just need to provide your Shunwang App Key and Secret in the setup. From there, your AI client can start managing your infrastructure directly.

**How do I check the real-time health of a specific computing node?**
Use the `get_node` tool with the corresponding `node_id`. It will return real-time metrics including CPU, GPU, and Memory usage, as well as the current online status.

**Can I deploy a new computing task using the agent?**
Yes. Use the `create_task` tool. You will need to provide the `image_id` and the `cluster_id` where you want to deploy the task. You can also optionaly give the task a name.

**Is it possible to remotely reboot a node through the chat?**
Yes, using the `send_node_command` tool. You just need the `node_id` and specify 'reboot' as the command. This allows for rapid remote maintenance without manual intervention.