Receber por email
← Voltar para o catálogo

contabo-live-instance-lifecycle-guard

Esta skill de live-guard gerencia operações de ciclo de vida de VPS e VDS da Contabo, como criação, reinstalação e cancelamento. Ela exige reconhecimento obrigatório do período de contrato, confirmação do impacto na cobrança e um plano de rollback, interrompendo qualquer ação sem aceitação explícita do período.

19estrelas
Atualizado há 4 meses

Ver no GitHub ↗ Licença: Apache-2.0

Como adicionar

/plugin marketplace add Raishin/vanguard-frontier-agentic

O comando exato pode variar conforme o repositório. Confira o README no GitHub.

Para o autor da skill

Cole no README do seu repo

Mostra que sua skill está catalogada na Skillteca, gera backlink e tráfego rastreável.

[![Skillteca](https://www.skillteca.com.br/api/badge/contabo-live-instance-lifecycle-guard/svg)](https://www.skillteca.com.br/skills/contabo-live-instance-lifecycle-guard)
#ai

Skills relacionadas

Ver todas de DevOps e Infra →

Alerta por categoria

Receba novas skills de DevOps e Infra toda segunda

1 email curto, só com as skills novas de DevOps e Infra. 4 minutos de leitura, sem spam, cancela com 1 clique.

Você confirma o email no primeiro envio. Sem spam. Cancela com 1 clique.

Ler descrição completa

Contabo Live Instance Lifecycle Guard

Purpose

Act as the approval gate for Contabo VPS/VDS lifecycle mutations: inventory current state, surface billing obligations, confirm the rollback path, then execute only after explicit user sign-off on all three.

When to use

Use this skill for:

  • VPS or VDS instance creation (product selection, region, image, contract period, Cloud-Init userData, SSH key secret IDs)
  • Instance reinstallation (image change, userData update, SSH key rotation)
  • Instance cancellation (period remaining, early-termination billing impact)
  • Verifying instance inventory before any bulk lifecycle operation
  • Generating approval-ready change records with period, billing impact, and rollback plan

Hard-stop conditions

REFUSE to execute any mutation unless ALL of the following are confirmed in writing:

  1. Target: Instance ID (for reinstall/cancel) or product ID + region (for create)
  2. Contract period: Explicit selection of 1, 3, 6, or 12 months with billing impact acknowledged
  3. Rollback plan: Documented recovery path if the operation fails or produces unexpected results
  4. Named approving identity: the full name or authenticated account identifier of the person authorizing this operation (not a role, alias, or ticket number alone)

Lean operating rules

  • Contabo has no official Terraform provider or SDK — recommend cntb CLI or REST API (curl + jq) for automation.
  • Prefer official Contabo docs (https://api.contabo.com/, https://docs.contabo.com/) and Context7 when live MCP access is unavailable.
  • Separate confirmed facts from inference. If state was not queried or shown, say so.
  • OAuth2 password grant tokens expire in ~5 minutes — include token refresh handling in all automation examples. Refresh logic must not log token values.
  • Include x-request-id (UUIDv4) in all REST API mutation calls for support traceability.
  • SSH keys must be referenced via Contabo secret IDs — never include raw private key material in API payloads, scripts, or recommendations.
  • Cloud-Init userData submitted by the user must be reviewed for embedded secrets, curl-pipe-sh patterns, and destructive commands before inclusion in the API payload. Flag and refuse to pass userData that contains raw credentials, unauthenticated remote execution, or commands that bypass audit trails.
  • Inventory current instances via read-only API calls before proposing any mutation.
  • Label claims as live evidence, user-provided sanitized evidence, documentation-based, or inference.

Automation pattern (read-only inventory first)

# Load credentials from environment — never hardcode
: "${CONTABO_CLIENT_ID:?set in env}"
: "${CONTABO_CLIENT_SECRET:?set in env}"
: "${CONTABO_API_USER:?set in env}"
: "${CONTABO_API_PASSWORD:?set in env}"

# Refresh token before each operation
TOKEN=$(curl -s \
  -d "client_id=${CONTABO_CLIENT_ID}" \
  -d "client_secret=${CONTABO_CLIENT_SECRET}" \
  --data-urlencode "username=${CONTABO_API_USER}" \
  --data-urlencode "password=${CONTABO_API_PASSWORD}" \
  -d 'grant_type=password' \
  'https://auth.contabo.com/auth/realms/contabo/protocol/openid-connect/token' \
  | jq -r '.access_token')

# Read-only inventory
curl -s \
  -H "Authorization: Bearer ${TOKEN}" \
  -H "x-request-id: $(uuidgen)" \
  'https://api.contabo.com/v1/compute/instances' | jq .

Response minimum

Return, at minimum:

  • the target instance or product tier and evidence level,
  • the contract period with billing impact,
  • the hard-stop checklist status (all three items confirmed or blocked),
  • the rollback plan,
  • the assumptions or open questions that require user clarification before proceeding.

References

Load these only when needed:

  • [Workflow and output contract](references/workflow-and-output.md) — use when executing the full lifecycle operation or formatting the approval-ready change record.
  • [Safety checklist](references/safety-checklist.md) — use before any VPS or VDS mutation; all hard-stop gates must be confirmed before proceeding.
  • [Official sources](references/official-sources.md) — use when grounding Contabo instance lifecycle API behavior, product specifications, or cntb CLI usage.
Compartilhar X LinkedIn