Connect AWS Bedrock Custom Model Import
Choose your client and follow the steps below.
Visual Studio Code
Install the MCP extension
Open VS Code and install GitHub Copilot (includes MCP support) from the Extensions Marketplace, or confirm it's already installed.
Open your MCP settings
Press Ctrl+Shift+P (or ⌘+Shift+P on macOS) and search for MCP: Add Server, or manually edit:
Add the server configuration
Paste the following into your mcp.json file:
{
"servers": {
"aws-bedrock-custom-model-import": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"mcp/aws-bedrock-custom-model-import-mcp-server"
],
"env": {
"AWS_SECRET_ACCESS_KEY": "YOUR_AWS_SECRET_ACCESS_KEY",
"AWS_SESSION_TOKEN": "YOUR_AWS_SESSION_TOKEN",
"AWS_REGION": "YOUR_AWS_REGION",
"AWS_PROFILE": "YOUR_AWS_PROFILE",
"AWS_ACCESS_KEY_ID": "YOUR_AWS_ACCESS_KEY_ID",
"BEDROCK_MODEL_IMPORT_S3_BUCKET": "YOUR_BEDROCK_MODEL_IMPORT_S3_BUCKET"
}
}
}
}
Set required environment variables
When prompted, provide values for:
AWS_SECRET_ACCESS_KEY
AWS_SESSION_TOKEN
AWS_REGION
AWS_PROFILE
AWS_ACCESS_KEY_ID
BEDROCK_MODEL_IMPORT_S3_BUCKET
Start using it
Open Copilot Chat and the AWS Bedrock Custom Model Import tools will be available automatically.
Cursor
Open Cursor Settings
Go to Cursor Settings → MCP and click + Add new global MCP server, or manually edit:
Add the server configuration
Paste the following into your mcp.json:
{
"mcpServers": {
"aws-bedrock-custom-model-import": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"mcp/aws-bedrock-custom-model-import-mcp-server"
],
"env": {
"AWS_SECRET_ACCESS_KEY": "YOUR_AWS_SECRET_ACCESS_KEY",
"AWS_SESSION_TOKEN": "YOUR_AWS_SESSION_TOKEN",
"AWS_REGION": "YOUR_AWS_REGION",
"AWS_PROFILE": "YOUR_AWS_PROFILE",
"AWS_ACCESS_KEY_ID": "YOUR_AWS_ACCESS_KEY_ID",
"BEDROCK_MODEL_IMPORT_S3_BUCKET": "YOUR_BEDROCK_MODEL_IMPORT_S3_BUCKET"
}
}
}
}
Set required environment variables
Provide values for:
AWS_SECRET_ACCESS_KEY
AWS_SESSION_TOKEN
AWS_REGION
AWS_PROFILE
AWS_ACCESS_KEY_ID
BEDROCK_MODEL_IMPORT_S3_BUCKET
Start using it
Open Cursor's AI chat — the AWS Bedrock Custom Model Import tools will appear automatically.
ChatGPT (Desktop)
Open ChatGPT Desktop settings
Click your profile icon → Settings → Beta features and ensure MCP plugin support is enabled.
Add the MCP server
Navigate to Settings → MCP Servers → + Add Server and paste:
{
"mcpServers": {
"aws-bedrock-custom-model-import": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"mcp/aws-bedrock-custom-model-import-mcp-server"
],
"env": {
"AWS_SECRET_ACCESS_KEY": "YOUR_AWS_SECRET_ACCESS_KEY",
"AWS_SESSION_TOKEN": "YOUR_AWS_SESSION_TOKEN",
"AWS_REGION": "YOUR_AWS_REGION",
"AWS_PROFILE": "YOUR_AWS_PROFILE",
"AWS_ACCESS_KEY_ID": "YOUR_AWS_ACCESS_KEY_ID",
"BEDROCK_MODEL_IMPORT_S3_BUCKET": "YOUR_BEDROCK_MODEL_IMPORT_S3_BUCKET"
}
}
}
}
Set required environment variables
Provide values for:
AWS_SECRET_ACCESS_KEY
AWS_SESSION_TOKEN
AWS_REGION
AWS_PROFILE
AWS_ACCESS_KEY_ID
BEDROCK_MODEL_IMPORT_S3_BUCKET
Start using it
Start a new chat — AWS Bedrock Custom Model Import tools will be available when the server is connected.
Claude Desktop
Open Claude Desktop config
Open the Claude Desktop configuration file at:
Windows: %APPDATA%\Claude\claude_desktop_config.json
Add to the mcpServers object
Merge the following into the "mcpServers" key:
{
"aws-bedrock-custom-model-import": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"mcp/aws-bedrock-custom-model-import-mcp-server"
],
"env": {
"AWS_SECRET_ACCESS_KEY": "YOUR_AWS_SECRET_ACCESS_KEY",
"AWS_SESSION_TOKEN": "YOUR_AWS_SESSION_TOKEN",
"AWS_REGION": "YOUR_AWS_REGION",
"AWS_PROFILE": "YOUR_AWS_PROFILE",
"AWS_ACCESS_KEY_ID": "YOUR_AWS_ACCESS_KEY_ID",
"BEDROCK_MODEL_IMPORT_S3_BUCKET": "YOUR_BEDROCK_MODEL_IMPORT_S3_BUCKET"
}
}
}
Set required environment variables
Replace the placeholder values for:
AWS_SECRET_ACCESS_KEY
AWS_SESSION_TOKEN
AWS_REGION
AWS_PROFILE
AWS_ACCESS_KEY_ID
BEDROCK_MODEL_IMPORT_S3_BUCKET
Restart Claude Desktop
Quit and reopen Claude Desktop. The AWS Bedrock Custom Model Import tools will appear in the tools menu.