DigitalOcean MCP Server – Cloud Resource Control
DigitalOcean's official MCP servers give AI agents structured access to DigitalOcean services and API-backed cloud operations. Use them to inspect, deploy, configure, and manage supported infrastructure from an MCP-compatible client, either through hosted service endpoints or a unified local server.
Overview
DigitalOcean's official MCP servers connect compatible AI development tools to
DigitalOcean services. They expose structured, API-backed tools that let an
agent reason about and manage current cloud resources instead of relying only
on static model knowledge or manually copied dashboard information.
What the MCP server enables
DigitalOcean provides service-specific MCP capabilities across its cloud and AI
platform. Depending on the enabled service and API token permissions, an AI
agent can:
- Deploy and manage App Platform applications, builds, deployments, and variables.
- Create and control Droplets, snapshots, images, and power actions.
- Inspect and manage Kubernetes clusters, nodes, workloads, and configuration.
- Manage databases, users, backups, and engine-specific settings.
- Configure domains, DNS records, firewalls, load balancers, reserved IPs, and VPCs.
- Work with container registries, block storage, Spaces, and network file storage.
- Manage uptime checks, alert policies, metrics, and selected account resources.
- Use AI Platform services such as model catalog, knowledge bases, evaluations,
batch inference, and dedicated inference. - Search and retrieve DigitalOcean documentation without authentication.
When to use it
Use DigitalOcean MCP when an agent needs live DigitalOcean context or must
perform an authorized cloud-management task. Typical workflows include
provisioning development infrastructure, deploying an app, resizing a database,
checking Kubernetes resources, configuring networking, inspecting account or
billing information, managing object storage, or troubleshooting an operational
issue from an AI coding assistant.
Connection and authentication
DigitalOcean offers separate hosted Streamable HTTP endpoints for each service,
such as App Platform at https://apps.mcp.digitalocean.com/mcp, Droplets at
https://droplets.mcp.digitalocean.com/mcp, and Databases at
https://databases.mcp.digitalocean.com/mcp. Most hosted services authenticate
with a DigitalOcean API token in the Authorization header as a Bearer token.
The documentation endpoint does not require a token.
DigitalOcean also publishes the unified local @digitalocean/mcp npm package.
It runs over stdio, reads DIGITALOCEAN_API_TOKEN, and enables only the services
listed with the --services option.
Key considerations
Tool access is limited by the token's scopes and by service or regional
availability. Use the least privilege required, enable only necessary services,
and require human approval for destructive, billable, or production changes.
Keep tokens outside source control and protect MCP configuration files. Remote
services are separate endpoints rather than one universal hosted URL, while the
local package provides a unified process. Start with small operations, retain
request logs for troubleshooting, and verify that requested services are
available in the intended region.
Supported Transports
streamable_http
URL: https://apps.mcp.digitalocean.com/mcp
stdio
Command: npx
Args:
-y@digitalocean/mcp--servicesapps,droplets,databases
Frequently Asked Questions
- When should an AI agent use the DigitalOcean MCP servers?
- Use them when a workflow needs current DigitalOcean resource context or approved cloud operations, such as deploying apps, managing Droplets or Kubernetes, configuring databases and networking, inspecting metrics, or working with storage and AI Platform services.
- What do the DigitalOcean MCP servers add to an AI agent's capabilities?
- They give the agent structured access to live DigitalOcean APIs and service-specific tools, allowing it to inspect and manage current cloud resources instead of relying only on static model knowledge or manually copied control-panel information.
- What can an AI agent access or manage through DigitalOcean MCP?
- Depending on enabled services and token permissions, the agent can work with App Platform, Droplets, Kubernetes, databases, networking, container registries, block storage, Spaces, file storage, insights, accounts, marketplace resources, and selected AI Platform services.
- How is authentication configured for DigitalOcean MCP?
- Most hosted service endpoints require a DigitalOcean API token sent as a Bearer token. The local server reads DIGITALOCEAN_API_TOKEN. The documentation MCP endpoint is an exception and can be used without authentication. Always use minimum scopes and store the token securely.
- Which transport should be used for DigitalOcean MCP?
- Use Streamable HTTP when connecting directly to one of DigitalOcean's hosted service-specific endpoints. Use stdio with the official @digitalocean/mcp package when a single local process should expose multiple selected DigitalOcean services.