SeaCloud CLI · v0.0.19

SeaCloud,
from the command line

A CLI built for agents to discover models, run multimodal tasks, and manage workflow skills with a single API key.

npm
npm install -g @seacloudai/seacloud-cli
Also available viadirect binary

Interactive Demo

See the CLI
in action

Real commands, real output. Click any tab to explore workflows end-to-end

seacloud - run
LIVE
1
seacloud@~/projects (main) $ seacloud run happyhorse_1.0_i2v --param 'media=[{"type":"first_frame","url":"https://example.com/frame.jpg"}]' --param prompt="dancer in neon rain" --output url
3
* Submitting task to happyhorse_1.0_i2v ...
4
Task ID: tsk_a9f23d18c4b7e601
6
Generating | elapsed 0:23
Progress100%
9
[done] Generation complete
11
Output URL:
12
https://cdn.seacloud.ai/output/tsk_a9f23d18c4b7e601.mp4
14
seacloud@~/projects $

Capabilities

One CLI
Every modality

Instant Generation

Submit multimodal tasks directly in your terminal. For image-to-video, replace the example media URL with a public first-frame image URL

>
seacloud run happyhorse_1.0_i2v --param 'media=[{"type":"first_frame","url":"https://example.com/frame.jpg"}]'

Model Discovery

Browse 20+ AI models by type, search by keyword, and inspect full parameter specs before running

>
seacloud models list --type video --output id

Task Management

Track any generation task by ID. Poll status, retrieve output URLs, and pipe results into your workflow

>
seacloud task status <id> --output json

SkillHub Integration

Install AI agent skills from the SkillHub marketplace. Auto-detects Claude Code, Cursor, and Codex targets

>
seacloud skills find "video"

Secure Auth

Store a SeaCloud API key for model execution, or use browser login for full interactive sessions

>
seacloud auth set-key <api-key>

Dry Run Mode

Preview exactly what would be sent to the API without making any requests. Safe for testing and debugging

>
seacloud --dry-run run <model> --param key=value

Workflow

From prompt to output in four steps

The complete process from authentication to result retrieval

01

Authenticate

Store an API key for model execution, or use browser login for sandbox and template workflows

$seacloud auth set-key <api-key>
02

Pick a Model

Browse models by type, inspect parameters, and know exactly what you are running

$seacloud models list
03

Generate

Submit your prompt and watch the generation happen live in your terminal

$seacloud run <model>
04

Retrieve & Pipe

Get output URLs, pipe JSON downstream, and integrate with any automation

$seacloud task status <id>

Command Reference

Every command, clearly documented

Click any command to see subcommands, flags, and runnable examples

seacloud auth

Manage authentication with your SeaCloud account

Syntax
$ seacloud auth <subcommand>
Subcommands
seacloud auth set-key <api-key>Store an API key for model execution
seacloud auth loginLog in via browser OAuth2. Stores credentials locally
seacloud auth logoutRevoke credentials and remove local token
seacloud auth statusDisplay current state, account tier, and credits
Examples
Set API key
$ seacloud auth set-key <api-key>
Interactive login
$ seacloud auth login

CI/CD Integration

Built for automation

Store an API key with seacloud auth set-key before running headless jobs. The --output json flag makes every command scriptable

Secure Auth

Use seacloud auth set-key for model execution, or browser login for full interactive sessions

JSON First

Every command supports --output json, so jobs can pipe to jq or downstream systems

Standard Exits

Exit code 0 means success and 1 means error. It behaves like a normal POSIX tool

github-actions - ci.yml
LIVE
1
name: Generate Marketing Video
3
on
4
push
5
branches: [main]
7
jobs
8
generate
9
runs-on: ubuntu-latest
10
steps
11
- uses: actions/checkout@v4
13
- name: Install SeaCloud CLI
14
run: npm install -g @seacloudai/seacloud-cli
16
- name: Generate video
17
env
18
SEACLOUD_API_KEY: ${{ secrets.SEACLOUD_API_KEY }}
19
run: |
20
seacloud auth set-key "$SEACLOUD_API_KEY"
21
VIDEO_URL=$(seacloud run happyhorse_1.0_i2v \
22
--param 'media=[{"type":"first_frame","url":"https://example.com/frame.jpg"}]' \
23
--param prompt="product launch" \
24
--output url)

Ready to Build?

Generate anything
From the terminal

Start building with multimodal AI today. No credit card required

Get StartedView Docs