planning-deploymentlisted
Install: claude install-skill JavierQuinan/ai-dev-team-marketplace
# Planning deployment
**Advisory / planning only.** This skill never executes a deployment — not after "sí", "confirmo", "hazlo", "autorizado", or any other phrasing of consent. See [enforcing-safety-baseline](../enforcing-safety-baseline/SKILL.md): production/shared-environment deployment always requires separate, explicit human execution outside this skill.
## Workflow
1. **Separate the current deployment shape from the requested target — they answer different questions.**
- **Current deployment evidence** (what the repo shows *today*): look for `vercel.json`/`.vercel/project.json` (Vercel), `Dockerfile`/`docker-compose*`/`compose*` (Docker), systemd units/`nginx`/`caddy` config (VPS), Kubernetes manifests/Helm charts (K8s), `Procfile`, CI deploy jobs, or deployment docs naming the platform. A framework being *compatible* with a platform (e.g. Next.js auto-detecting on Vercel) is never itself evidence of the current target — that same app could just as legitimately run on Docker, a VPS, or elsewhere; only platform-specific config/CI/docs count.
- **Requested target** (what the user asks to plan *for*): when the user explicitly names a target ("despliega esto en Vercel", "planifica cómo migrar esto a Docker"), that target governs the plan — repo evidence about the *current* shape never overrides an explicit instruction about the *desired* one. Label it `TARGET: <X>` / `SOURCE: USER-PROVIDED` / `REPO SUPPORT: verified` (if current evidence also confirms it) or `NOT V