# CoreWeave MCP for AI Agents AI Agent Connect > CoreWeave MCP lets you manage high-performance GPU infrastructure directly through your AI client. You can provision bare-metal Kubernetes clusters, configure VPCs for secure networking, and orchestrate inference gateways for your deployed models. It turns complex cloud operations into simple conversation. ## Overview - **Category:** cloud-infrastructure - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_a151JnxoNH7y2muaVwaqzr9Q5YKz8wmiZJSE1HZv/ai-agent-connect - **Tags:** gpu-cloud, kubernetes, inference, vpc, ai-infrastructure ## Description Managing high-performance GPU infrastructure usually involves juggling multiple tabs and complex cloud consoles. This Connector changes that by bringing the CoreWeave control plane directly into your chat interface. Instead of manually clicking through menus to spin up a new cluster or configure a network, you just tell your agent what you need. You can build out your entire infrastructure, from isolated VPCs to inference gateways, using only natural language. It's a big shift for anyone who needs to keep up with scaling AI workloads without the overhead of manual configuration. By connecting this to the Vinkius catalog, you get a streamlined way to handle your heavy-lifting compute resources while staying focused on building your models. You get a direct line to your hardware, making it much easier to see what's running and where your resources are going. ## Tools ### create_deployment Launch a new Inference Deployment for your AI models. This makes your model available for production use. ### create_gateway Set up a new Inference Gateway to manage model traffic. It handles the routing and authentication for your requests. ### delete_capacity_claim Remove an Inference Capacity Claim you no longer need. This helps you free up your reserved resources. ### delete_deployment Remove an Inference Deployment from your active list. Use this to clean up old model versions. ### delete_gateway Delete an Inference Gateway that is no longer in use. This simplifies your routing architecture. ### get_cluster Pull the specific details for a CKS cluster. You can use this to check the status of your active compute. ### list_deployments Review all your Inference Deployments. Use this to see which models are currently live. ### list_gateways Get a list of all your Inference Gateways. This helps you audit your traffic routing. ### list_vpcs See all your Virtual Private Clouds at a glance. You can quickly check your network inventory. ### query_metrics Pull Prometheus metrics to see performance data. Use this to monitor your system's health. ### update_capacity_claim Change the details of an Inference Capacity Claim. You can adjust your reserved capacity as needs change. ### update_cluster Modify an existing CKS cluster using an update mask. You can change your cluster specs without deleting it. ### update_deployment Adjust an Inference Deployment to change how your model runs. This is useful for swapping out model versions. ### create_vpc Build a new Virtual Private Cloud for your resources. This provides the network isolation your infrastructure needs. ### delete_cluster Tear down a CKS cluster when you're finished with a project. This stops billing for unused bare-metal compute. ### delete_vpc Remove a VPC from your account. This is the final step in decommissioning a network environment. ### list_capacity_claims View all your Inference Capacity Claims in one list. It's the easiest way to see what's reserved. ### list_clusters See every CoreWeave Kubernetes Service (CKS) cluster you have. This gives you a bird's eye view of your hardware. ### query_logs Search through Loki logs to find specific events. This is your go-to for troubleshooting infrastructure issues. ### update_gateway Modify an Inference Gateway to change routing rules. Use this to update how traffic hits your models. ### update_vpc Edit the settings of an existing VPC. You can change CIDR blocks or other network properties here. ### get_vpc See the specific details of a Virtual Private Cloud. This helps you check your network configuration quickly. ### create_capacity_claim Request a new Inference Capacity Claim for your models. It ensures you have the resources needed for heavy inference. ### create_cluster Provision a new CoreWeave Kubernetes Service (CKS) cluster. This sets up your bare-metal compute for AI workloads. ## Prompt Examples **Prompt:** ``` List all my active CoreWeave clusters. ``` **Response:** ``` I've retrieved your CKS clusters. You have 2 active clusters: | Cluster Name | ID | Status | | :--- | :--- | :--- | | production-gpu-1 | cks-7721 | Active | | research-test-bed | cks-8832 | Active | Would you like details on either of them? ``` **Prompt:** ``` Show me the details for VPC ID vpc-99402. ``` **Response:** ``` Fetching VPC details... **VPC vpc-99402** is currently active with CIDR block **10.0.0.0/16**. It is configured for high-bandwidth interconnects between your GPU nodes. ``` **Prompt:** ``` List all inference deployments and gateways currently configured. ``` **Response:** ``` I've compiled the list. You have 3 active inference gateways routing traffic to 5 deployments, including your **llama-3-70b-prod** service. **Gateways:** * Gateway-Prod-01 (Healthy) * Gateway-Test-01 (Healthy) * Gateway-Backup (Healthy) All gateways are reporting healthy status. ``` ## Capabilities ### Spin up GPU clusters Create bare-metal Kubernetes clusters for intensive AI and ML workloads. ### Isolate network traffic Configure Virtual Private Clouds to keep your compute resources secure. ### Route inference traffic Manage gateways for routing and authenticating traffic to your models. ### Monitor deployment health Inspect inference deployments to see if your services are running well. ### Automate lifecycle tasks Perform CRUD operations on clusters, VPCs, and gateways to scale your setup. ## Use Cases ### Scaling training clusters An ML Engineer needs to spin up 5 more GPU nodes for a large training run. They ask the agent to list_clusters to check current capacity then provision new ones. ### Network isolation for production A DevOps lead needs to isolate a production model. They tell the agent to create_vpc to ensure the compute resources are on a private network. ### Routing traffic to new models A researcher wants to route traffic to a new model version. They ask the agent to list_gateways and then update the routing rules. ### Real-time deployment monitoring A developer wants to know if the production deployment is healthy. They ask the agent to list_deployments and then check the status. ## Benefits - Manage GPU clusters without a dashboard by using create_cluster to spin up hardware on demand. - Secure your network faster by creating VPCs with create_vpc to isolate your compute resources. - Audit your inference traffic by listing all gateways with list_gateways to see how users reach your models. - Troubleshoot issues in seconds by pulling data with query_logs and query_metrics to find errors. - Automate your entire resource lifecycle using CRUD tools like create_deployment and delete_deployment to keep your environment clean. ## How It Works The bottom line is you get a hands-free way to manage high-performance GPU hardware. 1. Subscribe to the CoreWeave MCP on Vinkius. 2. Enter your CoreWeave API Token in your client settings. 3. Start managing your GPU cloud from your preferred AI client. ## Frequently Asked Questions **Can I use CoreWeave MCP to manage my GPU clusters?** Yes, it lets you provision and manage CKS clusters. You can create new ones for training or tear them down when you're done to save on costs. **Does this Connector support VPC configuration?** Yes, you can create and update VPCs for network isolation. This helps keep your compute resources secure and organized. **Can I use this to monitor my AI models?** Yes, it lets you list deployments and query metrics. You can see which models are live and check their performance in real time. **How do I connect my CoreWeave account?** Just provide your CoreWeave API Token in your client settings. From there, your agent can perform actions on your behalf. **Can I troubleshoot my infrastructure with this?** Yes, you can query Loki logs to find specific events and use Prometheus metrics to see performance data for your clusters. **Is this for production AI workloads?** Yes, it handles gateways and clusters for production-grade inference. It's designed for teams running heavy AI workloads. **Can I delete old clusters to save money?** Yes, you can use the delete_cluster tool to remove clusters you no longer need. This helps you stop billing for unused bare-metal hardware. **Can I list all my active Kubernetes clusters across the CoreWeave infrastructure?** Yes. By using the `list_clusters` tool, your agent will retrieve a complete list of all bare-metal Kubernetes clusters (CKS) managed under your account. **How do I check the specific network configuration of a VPC?** You can use the `get_vpc` tool by providing the specific VPC ID. The agent will return detailed information about network isolation and configuration for that resource. **Is it possible to create a new inference gateway via the AI agent?** Absolutely. Use the `create_gateway` tool with the required specification JSON. This allows you to set up routing and authentication for your AI model traffic programmatically.