DigitalOcean MCP Server

Official MCP server for managing DigitalOcean cloud infrastructure: droplets, databases, Kubernetes, apps, networking, storage, and GenAI services.

Cloud Infrastructure by DigitalOcean Bearer Token active

Connect (Remote Server) GitHub Docs npm: @digitalocean/mcp

Overview

The DigitalOcean MCP server is the official integration maintained by DigitalOcean Labs that exposes the DigitalOcean API to AI agents via the Model Context Protocol. It provides a comprehensive interface to manage cloud infrastructure across 19 distinct services including App Platform, Droplets, Managed Databases, DigitalOcean Kubernetes (DOKS), Container Registry (DOCR), Spaces, Volumes, NFS, Networking (VPCs, firewalls, CDN, load balancers), Serverless Functions, and the GenAI Platform (inference, custom models, evaluation).

The integration is offered in two forms. The recommended option is a set of remote MCP endpoints hosted by DigitalOcean (one URL per service domain, e.g. https://droplets.mcp.digitalocean.com/mcp), which require only a DigitalOcean API token passed as a bearer header. Alternatively, the server can be run locally via the @digitalocean/mcp npm package, where the --services flag selects which service modules to enable in order to limit context size.

Because each service has its own endpoint, you can scope an AI agent's access to only the parts of your DigitalOcean account it needs. Tools include operations such as creating apps from a spec, resizing droplets, enabling backups, creating SSH keys, managing DNS records, peering VPCs, flushing CDN caches, provisioning databases, managing Kubernetes clusters, and searching DigitalOcean documentation.

Tools

Tool	Description
`Apps service`	Manage DigitalOcean App Platform: create apps from a spec, deploy, list, and manage app configuration.
`Droplets service`	Create, resize, snapshot, enable backups for, and manage Droplet virtual machines and related resources.
`Databases service`	Provision and manage DigitalOcean Managed Databases, including monitoring and configuration.
`DOKS service`	Manage DigitalOcean Kubernetes Service clusters and node pools.
`DOCR service`	Manage the DigitalOcean Container Registry: repositories, tags, and garbage collection.
`Networking service`	Manage VPCs, VPC peering, firewalls, load balancers, reserved IPs, and CDN endpoints (e.g. cdn-flush-cache, vpc-peering-create/delete).
`Domains service`	Manage DNS domains and records (e.g. domain-create).
`Accounts service`	Manage account-level resources including SSH keys (key-create), team membership, and billing context.
`Spaces service`	Manage DigitalOcean Spaces object storage buckets and keys.
`Volumes service`	Create, attach, detach, and snapshot block storage Volumes.
`NFS service`	Manage shared NFS file storage volumes.
`Functions service`	Deploy and manage DigitalOcean Functions serverless workloads.
`Marketplace service`	Browse and deploy applications from the DigitalOcean Marketplace.
`Insights service`	Retrieve monitoring metrics and alerts for DigitalOcean resources.
`Docs service`	Search and retrieve content from the DigitalOcean documentation.
`GenAI Batch Inference service`	Run batch inference jobs on the DigitalOcean GenAI platform.
`GenAI Custom Models service`	Manage custom GenAI models hosted on DigitalOcean.
`Dedicated Inference service`	Provision and manage dedicated inference endpoints.
`Inference Model Catalog service`	Browse the catalog of available GenAI inference models.
`GenAI Evaluation service`	Run evaluations against GenAI models and inspect results.

Setup Guide

Prerequisites

A DigitalOcean account with a Personal Access Token generated at https://cloud.digitalocean.com/account/api/tokens
Export the token as an environment variable: export DIGITALOCEAN_API_TOKEN=dop_v1_...
For local installs: Node.js v18+ and npm v8+

Option 1: Remote MCP (recommended)

DigitalOcean hosts one MCP endpoint per service. Add only the services you need. Example for the Apps service:

{
  "mcpServers": {
    "digitalocean-apps": {
      "url": "https://apps.mcp.digitalocean.com/mcp",
      "headers": {
        "Authorization": "Bearer ${DIGITALOCEAN_API_TOKEN}"
      }
    }
  }
}

Available endpoints include apps, accounts, databases, doks, droplets, docr, networking, functions, spaces, volumes, nfs, marketplace, insights, docs, plus GenAI endpoints (genai-batchinference, genai-custom-models, dedicated-inference, inference-modelcatalog, genai-evaluation).

Option 2: Claude Code CLI

claude mcp add --transport http digitalocean-apps https://apps.mcp.digitalocean.com/mcp \
  --header "Authorization: Bearer $DIGITALOCEAN_API_TOKEN"

Option 3: Local install via npm

{
  "mcpServers": {
    "digitalocean": {
      "command": "npx",
      "args": ["@digitalocean/mcp", "--services", "apps,droplets,databases"],
      "env": {
        "DIGITALOCEAN_API_TOKEN": "${DIGITALOCEAN_API_TOKEN}"
      }
    }
  }
}

Use the --services flag with a comma-separated list to limit which service modules are loaded and keep context small.

Config file locations

Claude Desktop (macOS): ~/Library/Application Support/Claude/claude_desktop_config.json
Cursor: ~/.cursor/config.json
VS Code: .vscode/mcp.json

The README emphasizes storing the token in an environment variable rather than hardcoding it in config files.

Use Cases

Provision and deploy a new App Platform application from a spec, then watch the deployment status from your IDE
Scale or resize Droplets and enable automated backups in response to operational alerts
Create, peer, and tear down VPCs and firewall rules when standing up new environments
Manage DNS records on domains hosted in DigitalOcean directly from chat (create A/CNAME/MX records)
Provision managed Postgres or Redis databases and a DOKS cluster as part of a bootstrapping workflow
Search the DigitalOcean documentation inline while debugging an infrastructure problem

Example Prompts

"Create a new $12/month Droplet in NYC3 running Ubuntu 22.04 and enable automatic backups."
"Deploy this app.yaml spec to DigitalOcean App Platform and report the live URL when it's ready."
"Add an A record for api.example.com pointing to my reserved IP in DigitalOcean DNS."
"List all my DOKS clusters and show CPU usage for the past 24 hours."
"Search the DigitalOcean docs for how to configure VPC peering between two regions."

Pros

Official server maintained by DigitalOcean Labs, with broad coverage across 19 services
Provider-hosted remote endpoints remove the need to run anything locally and use standard bearer auth
Per-service endpoints (and a local --services flag) let you scope agent access and keep context size manageable
Includes a Docs service so agents can self-serve DigitalOcean documentation while operating on resources

Limitations

A personal access token grants broad account access; scoping must be done by selecting which service endpoints you wire up
The README warns explicitly that letting an AI act on your DigitalOcean account carries real risk (destructive operations are possible)
Tool inventory is large and split across 19 endpoints, which can be confusing to configure for first-time users

Alternatives

AWS MCP servers from awslabs/mcp for equivalent operations on AWS
Cloudflare's official MCP servers for DNS, Workers, and edge networking
The godo Go SDK or doctl CLI driven by a generic shell MCP server for users who prefer scripting over a dedicated MCP integration