digitalocean/dots

digitalocean/dots

Description: Official DigitalOcean Typescript Client based on the DO OpenAPIv3 specification

Language: TypeScript

License: Apache-2.0

Stars: 6

Forks: 1

Open issues: 13

Created: 2025-04-07T18:26:24Z

Pushed: 2026-06-08T18:23:37Z

Default branch: main

Fork: no

Archived: no

README:

DoTs

DoTs is the official DigitalOcean Typescript Client based on the DO OpenAPIv3 specification.

> New in v1.11.0 — DoTs now ships first-class support for DigitalOcean Serverless Inference: streaming chat completions, image generation, and model listing. Jump to [AI & Inference](#ai--inference) for usage, or browse runnable scripts in [examples/inference/](./examples/inference).

Getting Started

Prerequisites

• NodeJS 20.10 or above
• TypeScript 5 or above
• A DigitalOcean account with an active subscription. Along with a DigitalOcean token with proper permissions to manage DigitalOcean resources.
• "type": "module" in your package.json (for ES module support)

Installation

> The package is published on npmjs.com.

npm i @digitalocean/dots

Documentation

> Note: Comprehensive documentation quality is currently lacking and under development. The documentation is actively being improved and will be updated once fixed.

https://digitaloceandots.readthedocs.io/en/latest/

Basic Usage

> A quick guide to getting started with client

Authenticating

dots must be initialized with createDigitalOceanClient(). A DigitalOcean token is required. This token can be passed in via DigitalOceanApiKeyAuthenticationProvider(), an example below:

const authProvider = new DigitalOceanApiKeyAuthenticationProvider(token!);
// Create request adapter using the fetch-based implementation
const adapter = new FetchRequestAdapter(authProvider);
// Create the API client
const client = createDigitalOceanClient(adapter);

Managing DigitalOcean Resources

Find below a working example of creating a DigitalOcean volume via dots:

import { createDigitalOceanClient, DigitalOceanApiKeyAuthenticationProvider,
FetchRequestAdapter, Volumes_ext4 } from "@digitalocean/dots";

const token = process.env.DIGITALOCEAN_TOKEN;
if (!token) {
throw new Error("DIGITALOCEAN_TOKEN not set");
}

const authProvider = new DigitalOceanApiKeyAuthenticationProvider(token!);
// Create request adapter using the fetch-based implementation
const adapter = new FetchRequestAdapter(authProvider);
// Create the API client
const client = createDigitalOceanClient(adapter);

async function main(): Promise {
try {
const volumeReq: Volumes_ext4 = {
sizeGigabytes: 10,
name: `test-volume`,
description: "Block storage testing",
region: "nyc3",
filesystemType: "ext4",
};

const resp = await client.v2.volumes.post(volumeReq);
if (resp && resp.volume) {
const volume = resp.volume;
console.log(`Created volume ${volume.name} `);
} else {
throw new Error("Failed to create volume or volume is undefined");
}

console.log("Done!");

} catch (err) {
console.error(err);
}
}

main();

Running the above code snippet with tsc && node index.js would output the following:

Created volume test-volume
Done!

More working examples can be found in dots/examples.

AI & Inference

> Added in v1.11.0. Talk to DigitalOcean Serverless Inference directly from DoTs — streaming chat, image generation, model listing, and more.

Authentication

Inference APIs are gated by a separate credential from the v2 control-plane token. You must use one of the following:

• A DigitalOcean Personal Access Token (PAT) with _full access_ scope — read/write across all resources, including GenAI/Inference. Tokens scoped only to specific resource types will be rejected.
• A Model Access Key issued from the DigitalOcean Cloud console under *GenAI Platform → Model Access Keys*. These are the recommended credential for production workloads since they're scoped only to inference.

The credential is passed as apiKey to InferenceClient. The same DIGITALOCEAN_TOKEN env var used elsewhere in this README works here only if that PAT has full-access scope; otherwise use a Model Access Key.

import { InferenceClient } from "@digitalocean/dots";

const client = new InferenceClient({
apiKey: process.env.DIGITALOCEAN_TOKEN!, // full-access PAT or Model Access Key
});

Streaming chat completions

import { InferenceClient } from "@digitalocean/dots";

const client = new InferenceClient({ apiKey: process.env.DIGITALOCEAN_TOKEN! });

const stream = await client.chat.completions.create({
model: "llama3.3-70b-instruct",
messages: [{ role: "user", content: "Write two short lines about DigitalOcean." }],
stream: true,
});

for await (const chunk of stream) {
process.stdout.write(chunk.choices?.[0]?.delta?.content ?? "");
}
process.stdout.write("\n");

Callback-style streaming is also supported:

await client.chat.completions.create(
{ model: "llama3.3-70b-instruct", messages, stream: true },
{
onData: (chunk) => process.stdout.write(chunk.choices?.[0]?.delta?.content ?? ""),
onComplete: () => console.log("\n[done]"),
onError: (err) => console.error(err),
},
);

Image generation

import { InferenceClient } from "@digitalocean/dots";

const client = new InferenceClient({ apiKey: process.env.DIGITALOCEAN_TOKEN! });

const result = await client.images.generate({
model: "stable-diffusion-3.5-large", // any image model from `client.models.list()`
prompt: "An isometric illustration of a serverless data center, soft pastel colors",
n: 1,
size: "1024x1024",
});

const img = result.data?.[0];
if (img?.url) {
console.log(img.url);
} else if (img?.b64_json) {
// Stable Diffusion returns base64 — decode and save to disk.
const fs = await import("node:fs");
fs.writeFileSync("generated-image.png", Buffer.from(img.b64_json, "base64"));
console.log("Saved generated-image.png");
}

Listing available models

import { InferenceClient } from "@digitalocean/dots";

const client = new InferenceClient({ apiKey: process.env.DIGITALOCEAN_TOKEN! });

const models = await client.models.list();
for (const m of models.data ?? []) {
console.log(`${m.id}\t${m.owned_by ?? ""}`);
}

Minimal, runnable versions of all three snippets live under [examples/inference/](./examples/inference):

-…