AWS Bedrock AgentCore

Deploy an MCP server from the Solo Enterprise for agentregistry catalog to AWS Bedrock AgentCore.

Before you begin

Follow the AWS Bedrock AgentCore quickstart guide to install Solo Enterprise for agentregistry, deploy a managed EC2 gateway, and register the AWS Bedrock AgentCore runtime. Make sure that the following networking details are stored in environment variables:
```
echo $SUBNET_IDS_YAML
echo $SG_ID
echo $AWS_REGION
```
Create an MCP server.
Store the MCP server source code in a GitHub repository.
For testing purposes, you can skip this step and use the example MCP server in Solo.io’s sample repo instead, https://github.com/solo-io/doc-examples.
Publish the MCP server to the catalog.
Port-forward the Solo Enterprise for agentregistry service on your local machine so that you can interact with the registry.
```
kubectl -n agentregistry-system port-forward svc/agentregistry-enterprise-server 12121:12121
```

Deploy the MCP server

Open the MCP server catalog in the Solo Enterprise for agentregistry UI and select the MCP server that you want to deploy.
Click Deploy to open the deployment wizard.
Select the runtime and region that you want to deploy your MCP server to. Then, click Deploy Server.
Monitor the progress of the deployment in the Instance Logs section of your MCP server deployment. Note that the deployment can take a few minutes to complete.

List the runtimes that are connected to Solo Enterprise for agentregistry and note the name of the runtime that you want to deploy your MCP server to. In the following example, the runtime name is agentcore.
```
arctl get runtimes
```
Example output:
```
NAME                 TYPE
agentcore            BedrockAgentCore
kubernetes-default   kubernetes
local                local
```

Create a Deployment for the MCP server. The registry fetches the server source code and Dockerfile from the GitHub URL that you provided when you published the MCP server to the registry catalog. Then, the registry packages the code and builds an image that is compatible to run on AWS Bedrock AgentCore, and uses that image to deploy the runtime inside your VPC.

arctl apply -f - <<EOF
apiVersion: ar.dev/v1alpha1
kind: Deployment
metadata:
  name: mymcp
spec:
  targetRef:
    kind: MCPServer
    name: mymcp
    tag: "latest"
  runtimeRef:
    kind: Runtime
    name: agentcore
  runtimeConfig:
    region: $AWS_REGION
    entryPointFile: src/main.py
    networkMode: vpc
    subnetIds:
${SUBNET_IDS_YAML}
    securityGroupIds:
      - ${SG_ID}
EOF

Setting	Description
`targetRef`	The MCP server that you want to deploy.
`runtimeRef`	The name of the runtime to deploy to.
`runtimeConfig.region`	The AWS region to deploy to.
`runtimeConfig.entryPointFile`	The entry point file that AWS Bedrock AgentCore uses to start the MCP server.
`runtimeConfig.networkMode`	Set to `vpc` to deploy the MCP server inside your private VPC. Required for the agent to reach the MCP server over the private network.
`runtimeConfig.subnetIds`	The private subnets in which the MCP server runtime is deployed.
`runtimeConfig.securityGroupIds`	The security groups applied to the MCP server runtime.

Example output:

✓ Deployment/mymcp created

Open the Instances view in the UI and select the MCP server that you deployed. You can monitor the progress of the deployment in the Instance Logs section. Note that the deployment can take a few minutes to complete.

Solo Enterprise for kagent