Sealos Skills

Agent deploy plugin for Sealos Cloud

Deploy from your agent

Sealos Skills is an open-source deploy plugin that gives AI coding agents a concrete path from source code to Sealos Cloud. It lets an agent inspect a local repo or GitHub URL, create the right Docker and Sealos template artifacts, deploy to Sealos Cloud, then verify the running app.

npx plugins add labring/sealos…https://github.com/labring/sealos-skills

Direct skill fallback

npx skills add labring/sealos-skills

View source

$ npx plugins add https://github.com/labring/sealos-skills --target codexReady

Reading plugin metadata

Registering $sealos command

Checking Sealos auth helpers

Linking deploy skill

Ready for project analysis

Sealos Skills

agent-owned deploy

Run /sealos-deploy ~/project

file + terminal access

01 / OpenAI agent host

Codex CLI / App

Supported

$sealos in CLI, or choose Sealos from Plugins in the app.

02 / Anthropic agent host

Claude Code

Supported

Use /sealos after installing the Claude-compatible plugin.

03 / Direct skill

skills.sh hosts

Supported

Run /sealos-deploy directly from the installed skill pack.

04 / Context extension

Gemini / Qwen

Supported

Context extension mode: ask the agent to use Sealos Skills.

Install once

Works where agents work

The repository ships both plugin packaging and direct skill entry points, so the same deploy logic can travel across agent hosts.

Read install notes

Deploy intelligence

One command, two clean paths

The skill chooses deploy or update after preflight instead of asking the agent to guess what kind of project it is handling.

$sealos deploy ~/project

Deploy path

Preflight

Assess

Detect Image

Dockerfile

Template

Deploy + Verify

Update path

Find State

Confirm Target

Build Image

Patch Rollout

Readiness Check

Capability scan first

Checks Sealos auth, workspace, git, Docker, gh, kubectl, curl, and jq before touching the project.

Scored assessment

Scores deploy readiness from 0–12, detects framework, package manager, ports, databases, and environment variables.

Artifacts stay inspectable

Writes .sealos/analysis.json, build-result.json, template/index.yaml, and state.json so a run can be resumed or updated.

Read the pipeline

Operational states

Finished screens include the in-between moments

Loading, empty, and error states are shown as first-class deploy moments, not hidden behind a generic spinner.

Scanning project

The agent is checking runtime, entry points, Docker readiness, and Sealos auth.

Empty

No project selected

Pass a local folder or GitHub URL to start the deploy analysis.

Error

docker info failed

Docker daemon offline

The run can still detect reusable images, but local builds wait until Docker is running.

Preflight

Guided setup,
not guesswork

The skill separates hard blockers from optional accelerators, so the agent can continue when local builds are unnecessary.

Open setup guide

Sealos Cloud

Required account, auth token, workspace, and kubeconfig for every deploy.

Required

node sealos-auth.mjs login

Docker

Needed only when no reusable amd64 image exists and the agent must build locally.

Build path

docker info

GitHub CLI

Preferred for GHCR pushes and private image pull secrets.

Preferred

gh auth status

kubectl

Required for in-place updates, rollout checks, and cluster discovery.

Update path

kubectl version --client

Repository anatomy

A skill you can inspect

The repo keeps instructions, modules, and scripts readable, with run artifacts stored under .sealos for repeatable updates.

Browse repository

Sealos skill repository/ tree

README.md

.codex-plugin / plugin.json

commands / sealos.md

skills / sealos-deploy / SKILL.md

modules / preflight.md

modules / pipeline.md

scripts / deploy-template.mjs

dockerfile-skill + docker-to-sealos

Mode: plugin + skillState: .sealosLicense: MIT

For agent-led shipping

From prompt to verified rollout

Use it when you want the agent to handle deployment details while keeping the artifacts visible enough for human review.

See examples

Deploy a local repo

Ready

first run

EnvironmentCurrent workspaceResultAssess, build or reuse image, generate template, then deploy.

Deploy a GitHub URL

Ready

remote

EnvironmentRemote sourceResultClone with git, inspect the project, and create Sealos artifacts.

Update an existing app

Ready

repeat run

EnvironmentPrevious .sealos/state.json or cluster matchResultRebuild image, patch rollout, then verify pods, endpoints, and URL.

Sealos Skills

Install the skill.
Ship with evidence.

Give your coding assistant a concrete path from source code to Sealos Cloud, including build artifacts, template output, and rollout checks.

npx plugins add labring/sealos…https://github.com/labring/sealos-skills

Skeleton scan rows, no spinner

Empty

Project path prompt and direct action

Error

Inline recovery guidance

GitHub |MIT License |plugin + skill

Built for the open agent skills ecosystem.

Ready to Stop Configuring and
Start Creating?

Get started for free. No credit card required.

Resources

Docs Sealos Skills Education Blog Frequently Asked Questions

Products

DevBox Databases App Store

Services

Pricing FastGPT

Support

Deploy from your agent

Codex CLI / App

Claude Code

skills.sh hosts

Gemini / Qwen

Works where agents work

One command, two clean paths

Capability scan first

Scored assessment

Artifacts stay inspectable

Finished screens include the in-between moments

Scanning project

No project selected

Docker daemon offline

Guided setup, not guesswork

Sealos Cloud

Docker

GitHub CLI

kubectl

A skill you can inspect

From prompt to verified rollout

Deploy a local repo

Deploy a GitHub URL

Update an existing app

Install the skill. Ship with evidence.

Ready to Stop Configuring andStart Creating?

Guided setup,
not guesswork

Install the skill.
Ship with evidence.

Ready to Stop Configuring and
Start Creating?