Skip to content

SeaCloud SDK Overview

SeaCloud SDKs are pure-code packages for applications, automation scripts, and agents. The public docs currently cover the released TypeScript and Python SDKs for LLM chat, multimodal generation, model specs, async tasks, and SkillHub.

SDKs do not provide UI, CLI prompts, config-file discovery, or credential storage. Callers must pass an API key explicitly when creating a client; the SDKs do not read credentials from environment variables by themselves.

Quickstart Install an SDK, create a client, and run the first chat or generation call.

Operation Manual Use generation tasks, dry run, task polling, model specs, SkillHub, and errors.

Released SDKs

Language	Package / module	Use case
TypeScript	`@seacloudai/sdk`	Node.js, browser-proxy, and TypeScript applications with ESM and CommonJS support.
Python	`seacloudai-sdk`, imported as `seacloud_sdk`	Python services, scripts, and async applications powered by `httpx`.

Shared Capability Map

Capability	TypeScript	Python
Create client	`new SeaCloud({ apiKey })`	`SeaCloud(api_key=...)`
Offline SDK docs	`getSeaCloudDocs()`	`getSeaCloudDocs()`
Chat	`client.chat.send()`	`client.chat.send()`
Create async task	`client.run()`	`client.run()`
Wait for result	`client.runSync()`	`client.run_sync()`
Read task status	`client.tasks.get()`	`client.tasks.get()`
Read task response	`client.tasks.getResponse()`	`client.tasks.get_response()`
List models	`client.models.list()`	`client.models.list()`
Read model spec	`client.models.getSpec()`	`client.models.get_spec()`
Search SkillHub	`client.skills.find()`	`client.skills.find()`
List SkillHub	`client.skills.list()`	`client.skills.list()`

When To Use An SDK

You need a long-lived integration instead of one-off curl calls.
You want consistent API key handling, timeouts, error mapping, task polling, and normalized outputs.
You want agents to inspect local SDK usage through getSeaCloudDocs().
You need dryRun to preview the queue request before submitting a generation task.

For terminal sign-in, model inspection, or temporary model execution, use SeaCloud CLI. For direct HTTP integration, use Model API.