Sealos Deploy

Sealos Deploy

Identity and Discovery

• Owner: sealos-deploy (/sealos-deploy and deploy, update, publish, or cloud-runtime requests).
• Class: composite-orchestration across readiness, Dockerfile, template, build, deployment, and Runtime Truth.
• Canaries: DEP-KUBECONFIG-SCOPE, DEP-CONFIRM-MUTATION, DEP-REDACT, and DEP-RUNTIME-TRUTH.
• Contract: Read references/deploy-contract.md after the canaries pass. It defines the typed phase handoffs, owned .sealos artifacts, terminal states, and the read-only Canvas boundary.

Scope and Boundaries

Accept a local path or GitHub URL and scope all work to the selected namespace/app. Preserve the current DEPLOY/UPDATE phase order, .sealos artifact inventory, one log file, dependency handoffs, and cleanup footprint. All Kubernetes commands use KUBECONFIG=~/.sealos/kubeconfig kubectl --insecure-skip-tls-verify; unsupported workloads stop before scoring/build.

Risk and Confirmation

Keep auth/workspace, kubeconfig scope, system-tool installation, public exposure, credential changes, deletion, rollback, and cleanup confirmation visible before module detail. Redact passwords, tokens, cookies, env values, kubeconfig, Secret data, and full connection strings. A quality gate and actual Runtime Truth evidence are acceptance conditions.

For every gated mutation, report the exact operation, impact, confirmation, and post-action evidence. Keep sanitized logs, state, diagnostics, and footprint evidence free of secret data.

Lifecycle Workflow

For each request, run preflight/auth/workspace, detect mode, enforce eligibility, assess/detect/build or reuse, generate/validate the template, deploy or update, run Runtime Truth, and record state/cleanup evidence. Emit request-scoped success, stopped, or error; the existing phase modules below remain authoritative for detailed behavior.

Progressive Disclosure

Load modules/ and helper scripts one phase at a time after the corresponding canaries pass. Preserve typed readiness → Dockerfile → Docker-to-Sealos handoffs, .sealos/analysis.json, .sealos/build/build-result.json, .sealos/template/index.yaml, .sealos/state.json, and delivery evidence; do not hide phase order behind a generic deploy shortcut.

Output, Stop, and Error States

• success: actual returned App URL/live identity, route and port match, setup/login proof, recent logs/events, workload convergence, database/object evidence, full footprint, and saved deploy state.
• stopped: unsupported eligibility, missing auth/tool, unresolved configuration, or unconfirmed public/destructive/cleanup boundary with the safe next action.
• error: failed preflight, build/template/deploy/runtime/rollback/cleanup step and recovery action with sensitive values redacted.

Handoffs

Readiness, Dockerfile, and Docker-to-Sealos inputs use typed target, inputArtifact, allowedAction, failureReturn, and responseOwner fields. A verified deploy can hand target: sealos-canvas, inputArtifact: sanitized .sealos/state.json and Runtime Truth, allowedAction: read-only topology inspection, failureReturn: runtime/state diagnostic, and responseOwner: sealos-deploy.

Verification

Use the existing eligibility, artifact, quality-gate, footprint, live-smoke, rollout, Runtime Truth, and cleanup checks. Baseline cases deploy-positive-runtime-truth and deploy-violating-missing-runtime-proof must preserve actual App URL/live identity, auth/cleanup confirmation, log scans, and redaction.

Compatibility

Sealos auth/workspace are required for deploys. Docker, buildx, and gh CLI are required only when the selected path needs local build/push. git is required when cloning from a GitHub URL or when git metadata is needed. Node.js 18+ remains an optional accelerator. Phase 5 requires Python 3.8+ with PyYAML; root Compose conversion also requires kompose and may require crane when image tags are floating.

Brain Managed Mode

The skill has two deliberately separate execution modes:

• Local mode is the existing interactive workflow. It is selected when SEALAI_DEPLOY_MODE is absent or has any value other than managed; its auth, prompts, Template API flow, and output remain unchanged.
• Managed mode is selected only when SEALAI_DEPLOY_MODE=managed. The Devbox Codex is the deployment executor: it analyzes, builds, applies, observes, diagnoses, repairs, and verifies with the injected kubeconfig. Brain is the task control plane and form owner; it is not a second Kubernetes executor.

Managed mode is non-interactive. Do not start OAuth, install tools, ask for confirmation in the turn, or replace a missing callback with a file, webhook, curl request, or a Brain-side apply. Before doing any work, confirm that the Codex tool registry contains both exact MCP tools template_ready and deployment_completed. If either tool is unavailable, stop with a managed-mode fatal error; never claim a deployment result.

Brain supplies these task-scoped values through the environment:

SEALAI_DEPLOY_MODE=managed
SEALAI_DEPLOY_TASK_ID=<task id>
SEALAI_PROJECT_ID=<Brain project id>
SEALAI_NAMESPACE=<target namespace>
SEALAI_INPUTS_PATH=/run/sealai/deployment/inputs.json
SEALAI_TURN_DEADLINE_AT=<absolute deadline>

Use the injected kubeconfig/context for every Kubernetes command. Do not perform login or switch workspace. Keep the input file and kubeconfig out of prompts, logs, Timeline text, and generated artifacts; read the input file only when the managed flow says to do so.

The actual Sealos Instance name is owned by the Template/Skill path. Brain does not pre-allocate it and the managed adapter does not add Brain identity labels or extraLabels.

Deploy compatible cloud workloads to Sealos Cloud, stopping unsupported targets before build or deployment.

kubectl Safety Rules (all phases)

All kubectl commands MUST use the Sealos kubeconfig:

KUBECONFIG=~/.sealos/kubeconfig kubectl --insecure-skip-tls-verify

System tool installation requires user confirmation. If docker, gh, or kubectl is missing and the skill can install it for the current platform, ask first and only run the install command after the user explicitly replies y.

kubectl delete requires user confirmation. Before deleting any resource (deployment, service, ingress, PVC, database, etc.), always ask:

WARNING: About to delete <resource kind>/<resource name>. This data cannot be recovered. Confirm? (y/n)

Only proceed after user confirms. This applies even if the pipeline logic suggests deletion — always ask first.

Template API cleanup must include Instance CRs. Deployments created through scripts/deploy-template.mjs create instances.app.sealos.io/<app-name> in addition to App/workload resources. A cleanup is incomplete until instances.app.sealos.io, apps.app.sealos.io, workloads, Services, Ingresses, PVCs, and Pods are all checked.

Use this check when cleaning Template API test deployments:

KUBECONFIG=~/.sealos/kubeconfig kubectl --insecure-skip-tls-verify -n "$NS" \
  get instances.app.sealos.io,app,statefulset,deployment,svc,ingress,pvc,pod | grep "$APP"

Delete in this order after confirmation:

KUBECONFIG=~/.sealos/kubeconfig kubectl --insecure-skip-tls-verify -n "$NS" delete instances.app.sealos.io "$APP" --ignore-not-found --wait=false
KUBECONFIG=~/.sealos/kubeconfig kubectl --insecure-skip-tls-verify -n "$NS" delete app "$APP" --ignore-not-found --wait=false
KUBECONFIG=~/.sealos/kubeconfig kubectl --insecure-skip-tls-verify -n "$NS" delete statefulset "$APP" --ignore-not-found --wait=false
KUBECONFIG=~/.sealos/kubeconfig kubectl --insecure-skip-tls-verify -n "$NS" delete deployment "$APP" --ignore-not-found --wait=false
KUBECONFIG=~/.sealos/kubeconfig kubectl --insecure-skip-tls-verify -n "$NS" delete ingress "$APP" --ignore-not-found --wait=false
KUBECONFIG=~/.sealos/kubeconfig kubectl --insecure-skip-tls-verify -n "$NS" delete svc "$APP" --ignore-not-found --wait=false
KUBECONFIG=~/.sealos/kubeconfig kubectl --insecure-skip-tls-verify -n "$NS" get pvc -o name | grep "$APP" | while read -r PVC; do
  KUBECONFIG=~/.sealos/kubeconfig kubectl --insecure-skip-tls-verify -n "$NS" delete "$PVC" --ignore-not-found --wait=false
done

Anti-example: do not report cleanup complete after only checking app,statefulset,svc,ingress,pvc,pod; that misses instances.app.sealos.io/<app-name> and leaves the Sealos Instance layer dirty.

Usage

/sealos-deploy <github-url>
/sealos-deploy                    # deploy current project
/sealos-deploy <local-path>

Quick Start

Execute the modules in order:

1. modules/preflight.md — Environment checks & Sealos auth
2. modules/pipeline.md — Full deployment pipeline (Phase 1–6)
3. modules/runtime-truth.md — Post-deploy Runtime Truth Pass (Phase 6.5)

Logging

Every run MUST write a log file at ~/.sealos/logs/deploy-<YYYYMMDD-HHmmss>.log.

At the very start of execution, create the log file once:

mkdir -p ~/.sealos/logs
LOG_FILE=~/.sealos/logs/deploy-$(date +%Y%m%d-%H%M%S).log
echo "[$(date '+%Y-%m-%d %H:%M:%S')] Deploy started" > "$LOG_FILE"

Important: create the log file ONLY ONCE at the start. All subsequent writes MUST append (>>) to this same $LOG_FILE. Do NOT create a second log file.

At each phase boundary, append a log entry to the same file with Bash >>:

[2026-03-05 14:30:01] === Phase 0: Preflight ===
[2026-03-05 14:30:01] Docker: ✓ 27.5.1
[2026-03-05 14:30:01] Node.js: ✓ 22.12.0
[2026-03-05 14:30:02] Sealos auth: ✓ (region: <REGION from config.json>)
[2026-03-05 14:30:02] Project: /Users/dev/myapp (github: https://github.com/owner/repo)

[2026-03-05 14:30:03] === Phase 1: Assess ===
[2026-03-05 14:30:03] Score: 9/12 (good)
[2026-03-05 14:30:03] Language: python, Framework: fastapi, Port: 8000
[2026-03-05 14:30:03] Decision: CONTINUE

[2026-03-05 14:30:04] === Phase 2: Detect Image ===
[2026-03-05 14:30:05] Docker Hub: owner/repo:latest (arm64 only, no amd64)
[2026-03-05 14:30:05] GHCR: not found
[2026-03-05 14:30:05] Decision: no amd64 image → continue to Phase 3

[2026-03-05 14:30:06] === Phase 3: Dockerfile ===
[2026-03-05 14:30:06] Existing Dockerfile: none
[2026-03-05 14:30:07] Generated: python-fastapi template, port 8000

[2026-03-05 14:30:08] === Phase 4: Build & Push ===
[2026-03-05 14:30:08] Registry: ghcr (auto-detected via gh CLI)
[2026-03-05 14:30:30] Build: ✓ ghcr.io/zhujingyang/repo:20260305-143022
[2026-03-05 14:30:32] GHCR pullability: private package detected — deploy will auto-create image pull Secret from gh CLI
[2026-03-05 14:30:33] IMAGE_REF=ghcr.io/zhujingyang/repo:20260305-143022

[2026-03-05 14:30:34] === Phase 5: Template ===
[2026-03-05 14:30:35] Output: .sealos/template/index.yaml

[2026-03-05 14:30:36] === Phase 6: Deploy ===
[2026-03-05 14:30:36] Deploy URL: https://template.gzg.sealos.run/api/v2alpha/templates/raw
[2026-03-05 14:30:38] Status: 201 — deployed successfully
[2026-03-05 14:30:38] === DONE ===

On error, log the error details before stopping:

[2026-03-05 14:30:10] === ERROR ===
[2026-03-05 14:30:10] Phase: 4 (Build & Push)
[2026-03-05 14:30:10] Error: docker buildx build failed — "npm ERR! Missing script: build"
[2026-03-05 14:30:10] Retry: 1/3

At the very end, tell the user where the log is:

Log saved to: ~/.sealos/logs/deploy-20260305-143001.log

Scripts

Located in scripts/ within this skill directory (<SKILL_DIR>/scripts/):

All scripts output JSON. Run via Bash and parse the result.

For public web applications, run sealos-launchpad-network.mjs before HTTP smoke. Acceptance requires ok: true, an open public network, the expected Service port, and an App URL host match. The script emits an allowlisted network summary and excludes raw Launchpad application data, environment variables, Secrets, and kubeconfig content.

Runtime Event acceptance uses two scans. Capture the first report after readiness with no baseline, wait at least 60 seconds, then pass that report through --baseline for the final scan. Extend --min-window-seconds to cover one full known reconciliation, probe, or scheduled-work period. An initial Warning Event is an observation; a Warning that advances after the baseline, an unresolved referenced Secret, a Ready transition, a Pod replacement, or a restart delta is an active failure.

For intentional fault injection, retain a pre-injection report as evidence. After recovery reaches Ready, capture a fresh recovery baseline and compare the final scan against that recovery baseline after the full stability window.

Internal Skill Dependencies

This skill references knowledge files from co-installed internal skills. These are not user-facing — they are loaded on-demand during specific phases.

<SKILL_DIR> refers to the directory containing this SKILL.md. Sibling skills are at <SKILL_DIR>/../:

<SKILL_DIR>/../
├── sealos-deploy/           ← this skill (user entry point) = <SKILL_DIR>
├── dockerfile-skill/        ← Phase 3: Dockerfile generation knowledge
├── cloud-native-readiness/  ← Phase 0.4 eligibility policy + Phase 1 assessment criteria
└── docker-to-sealos/       ← Phase 5: Sealos template rules

Paths used in pipeline.md follow the pattern:

<SKILL_DIR>/../dockerfile-skill/knowledge/error-patterns.md
<SKILL_DIR>/../dockerfile-skill/templates/<lang>.dockerfile
<SKILL_DIR>/../docker-to-sealos/references/sealos-specs.md

Phase Overview

Decision Flow

Input (GitHub URL / local path)
  │
  ▼
[Phase 0] Preflight ── fail → guide user to fix and STOP
  │ pass
  ▼
[Phase 0.5] Template fast path
  │
  ├── materialized template match ───────┐
  │                                      │
  ▼                                      │
[Phase 1] Assess ── not suitable → STOP with reason
  │ suitable
  ▼
[Phase 2] Detect existing image
  │
  ├── found (amd64) ────────────────────┐
  │                                     │
  ▼                                     │
[Phase 3] Dockerfile (generate/reuse)   │
  │                                     │
  ▼                                     │
[Phase 4] Build & Push to registry      │
  │                                     │
  ◄─────────────────────────────────────┘
  │
  ▼
[Phase 5] Generate Sealos Template
  ◄──────────────────────────────────────┘
  │
  ▼
[Phase 5.5] Configure ── present env vars → ask user for inputs → confirm
  │
  ▼
[Phase 6] Deploy to Sealos Cloud ── 401 → re-auth
│                                  409 → instance exists
▼
[Phase 6.5] Runtime Truth Pass ── network/runtime/log/login issue → debug template or runtime config
│
▼
Done — app deployed ✓

Execution rule: Phase 1 must never start while Phase 0 still has unresolved entry blockers. Docker, gh, builder, and registry failures must be reported early, but only become hard blockers if the run later requires local build/push.