Deploying & Managing

Deploy your MCP integrations to production, share them with your team, and monitor their performance—all from within MCP Studio.

One-Click Deployment: MCP Studio handles containerization, infrastructure, and scaling automatically. Just click deploy and watch your integration go live.

Deployment Overview

When you deploy an MCP server, MCP Studio:

1. Validates your code for errors and security issues

2. Builds a container image with your code and dependencies

3. Deploys the container to cloud infrastructure

4. Creates an API endpoint for accessing your integration

5. Generates an API key for authentication

Aspect	Details
Infrastructure	AWS-powered secure cloud hosting
Scaling	Automatic scaling based on demand
Availability	High availability with redundancy
Security	Encrypted connections, isolated execution

Deploy to Your Own AWS Account: You can optionally deploy MCP servers into your own AWS account for data residency, compliance, or to use existing AWS credits. See Deploy to Your AWS Account for setup instructions.

Deployment Status Lifecycle

Your MCP server progresses through these statuses:

┌─────────┐   ┌────────────┐   ┌──────────┐   ┌───────────┐   ┌──────────┐
│  Draft  │ → │ Validating │ → │ Building │ → │ Deploying │ → │ Deployed │
└─────────┘   └────────────┘   └──────────┘   └───────────┘   └──────────┘
                                                    │
                                                    ↓
                                              ┌─────────┐
                                              │  Error  │
                                              └─────────┘

Status	Description	What's Happening
Draft	Initial state	Code has not been deployed
Validating	Code validation	Checking syntax, imports, and tool definitions
Building	Container build	Creating Docker image with dependencies
Deploying	Infrastructure setup	Launching container and configuring endpoints
Deployed	Live and ready	Integration is available for use
Error	Deployment failed	Something went wrong (see logs for details)
Archived	Deactivated	Integration has been stopped

Deploy Tab: Deploying Your Integration

1

Navigate to Deploy Tab

From the MCP Builder, click the Deploy tab.

2

Review Pre-Deployment Checklist

Before deployment, ensure:

Requirement	Description
✅ Code validates	Run validation on the Build tab
✅ Secrets configured	All required secrets have values
✅ Tools defined	At least one tool is defined
✅ Documentation added	Tools have clear descriptions

3

Click Deploy

Click the Deploy button to start the deployment process.

Catalog MCPs: If you're deploying from the MCP Catalog, validation is skipped since the code is pre-validated.

4

Monitor Progress

Watch the streaming deployment progress:

[12:34:56] Starting deployment...
[12:34:57] Validating code... ✓
[12:34:59] Building container image...
[12:35:15] Pushing image to registry... ✓
[12:35:22] Deploying to infrastructure...
[12:35:45] Creating API endpoint... ✓
[12:35:47] Deployment complete! ✓

5

Verify Deployment

Once complete, the status changes to Deployed and you can proceed to the Connect tab.

Handling Deployment Errors

If deployment fails:

1. Check the Error Message: The deployment log shows what went wrong

2. Review Your Code: Common issues include syntax errors or missing dependencies

3. Verify Secrets: Ensure all required credentials are provided

4. Try Again: Fix the issue and click Deploy again

Common Error	Cause	Solution
Validation failed	Code has errors	Fix errors shown in Build tab
Build failed	Dependency issues	Check package versions and imports
Deployment timeout	Infrastructure issue	Wait and retry, or contact support
Missing secrets	Required values not set	Configure all required secrets

Connect Tab: Enabling Your Integration

After deployment, use the Connect tab to enable your integration and get connection details.

Enable for Darcy

Toggle Enable for Darcy to make your integration available in:

Feature	How It Works
DarcyIQ Chat	Tools appear automatically for the AI to use
AI Workflows	Integration available as a workflow step

Instant Availability: Once enabled, your integration's tools are immediately available. No restart required.

Manual Connection Details

For connecting from external applications, the Connect tab provides:

Detail	Description	Example
Endpoint URL	The HTTPS endpoint for your integration	https://mcp.darcyiq.com/v1/abc123
API Key	Authentication key for requests	mcp_key_xxxxx...

Code Examples for External Connections

The Connect tab provides ready-to-use code examples:

Python (httpx):

import httpx

endpoint = "https://mcp.darcyiq.com/v1/abc123"
api_key = "mcp_key_xxxxx"

async with httpx.AsyncClient() as client:
    response = await client.post(
        f"{endpoint}/tools/call",
        headers={"Authorization": f"Bearer {api_key}"},
        json={
            "name": "get_customer",
            "arguments": {"customer_id": "12345"}
        }
    )
    result = response.json()
    print(result)

JavaScript (fetch):

const endpoint = "https://mcp.darcyiq.com/v1/abc123";
const apiKey = "mcp_key_xxxxx";

const response = await fetch(`${endpoint}/tools/call`, {
  method: "POST",
  headers: {
    "Authorization": `Bearer ${apiKey}`,
    "Content-Type": "application/json"
  },
  body: JSON.stringify({
    name: "get_customer",
    arguments: { customer_id: "12345" }
  })
});

const result = await response.json();
console.log(result);

cURL:

curl -X POST "https://mcp.darcyiq.com/v1/abc123/tools/call" \
  -H "Authorization: Bearer mcp_key_xxxxx" \
  -H "Content-Type: application/json" \
  -d '{"name": "get_customer", "arguments": {"customer_id": "12345"}}'

MCP Client Configuration:

{
  "mcpServers": {
    "my-integration": {
      "url": "https://mcp.darcyiq.com/v1/abc123",
      "apiKey": "mcp_key_xxxxx"
    }
  }
}

Monitor Tab: Viewing Logs

The Monitor tab shows real-time logs from your deployed integration.

Log Information

Field	Description
Timestamp	When the event occurred
Level	INFO, WARNING, ERROR, DEBUG
Message	Log message content
Request ID	Unique identifier for tracing

Using Logs for Debugging

Logs help you understand:

• Tool Invocations: When and how tools are called

• Errors: Stack traces and error messages

• Performance: Response times and bottlenecks

• Usage Patterns: How the integration is being used

Log Controls

Control	Function
Refresh	Load latest logs
Auto-refresh	Continuously update (toggle)
Filter	Filter by level or search text
Pagination	Navigate through historical logs

Sharing Integrations

Share your integrations with team members to enable collaboration.

Permission Levels

Level	Capabilities
Owner	Full control: edit, deploy, share, delete
Editor	Can modify code, configure secrets, deploy
Viewer	Read-only: can view code and logs

Sharing an Integration

1

Open Share Dialog

From the MCP server list or builder, click the Share button.

2

Add Team Members

Enter email addresses or select from your organization's members.

3

Set Permissions

Choose the permission level for each person:

• Owner: For co-maintainers

• Editor: For contributors

• Viewer: For reviewers or users

4

Send Invitations

Click Share to grant access. Team members can now see the integration in their MCP Studio.

Managing Shared Access

To modify or revoke access:

1. Open the Share dialog

2. Find the user in the list

3. Change their permission level or click Remove

Managing Deployed Integrations

From the My MCPs list, you can manage all your integrations:

Available Actions

Action	Description	When to Use
Edit	Open in builder	Modify code or settings
Duplicate	Create a copy	Test changes without affecting production
Deploy	Deploy or redeploy	Push updates to production
Stop	Undeploy	Temporarily disable the integration
Share	Manage access	Grant or revoke team access
Favorite	Add to favorites	Quick access from your list
Delete	Remove entirely	Permanently delete the integration

Updating a Deployed Integration

To update a live integration:

1

Make Changes

Edit your code on the Build tab.

2

Test Locally

Use the Test panel to verify changes work correctly.

3

Redeploy

Go to the Deploy tab and click Deploy again. The new version replaces the current deployment.

Downtime: There may be a brief interruption (a few seconds) while the new version deploys.

Stopping an Integration

To temporarily disable an integration without deleting it:

1. Click Stop from the actions menu

2. The status changes to Archived

3. The integration is no longer available in Chat or Workflows

4. You can redeploy later to restore it

Deleting an Integration

Permanent Action: Deleting an integration removes all code, configuration, and history. This cannot be undone.

1. Click Delete from the actions menu

2. Confirm the deletion

3. The integration is permanently removed

Organization Favorites

Mark integrations as organization favorites to highlight them for your team:

Adding to Favorites

1. Click the star icon on any integration

2. Optionally set a custom display name

3. The integration appears in the "Organization Recommended" section

Benefits of Favorites

Benefit	Description
Visibility	Appear at the top of the MCP Catalog
Standardization	Guide team members to approved integrations
Custom Naming	Use names that make sense for your organization

Deployment Best Practices

Before Deployment

Practice	Description
Test Thoroughly	Use the Test panel to verify all tools work
Validate Code	Run validation to catch errors
Document Tools	Add clear descriptions for AI understanding
Configure Secrets	Ensure all credentials are set

After Deployment

Practice	Description
Monitor Logs	Check for errors or unexpected behavior
Test in Chat	Verify tools work in real conversations
Share Carefully	Only give necessary permissions
Update Regularly	Keep dependencies and code current

Security Considerations

Practice	Description
Rotate API Keys	Change keys if they may be compromised
Minimum Permissions	Share with appropriate access levels
Monitor Usage	Watch logs for unexpected activity
Secure Secrets	Never expose credentials in code or logs

Next Steps

Goal	Documentation
View code examples	Code Samples
Troubleshoot issues	Troubleshooting
Browse the MCP Catalog	MCP Catalog
Build custom integrations	Building Custom MCPs