pks deploy

stable

Reference for pks deploy, a cosmetic Spectre.Console demo of a deployment flow with no build, registry push, or cluster call behind it.

Author: Poul Kjeldager
Usage: pks deploy [options]
Category: infrastructure

Examples

$ pks deploy

Run the simulated flow against the default dev environment

$ pks deploy --environment staging --replicas 5

Same flow with different labels in the printed panel

$ pks deploy --watch

Add a 20-second live metrics table after the results panel

pks deploy renders a scripted, cosmetic deployment sequence — a pre-flight checks table, four animated progress bars, a results panel, and an optional live metrics view — using Spectre.Console's UI primitives. It does not build code, does not push to a container registry, and does not talk to Kubernetes, Coolify, Azure, or any cluster API.

Overview

pks deploy is a single leaf command (DeployCommand, registered in Program.cs with only a description, no .WithExample() or .WithAlias() entries). Every step in its output is either hardcoded or driven by Random and Thread.Sleep calls:

  • The pre-deployment checks table always shows four green rows, including the literal string "127/127 tests passing" — no test suite, security scan, or dependency check actually runs.
  • The build/push/deploy/health progress bars advance on timed increments with no underlying build, registry push, or network call.
  • The results panel always reports success, a fixed ~42ms average response time, and fabricated endpoint URLs (https://api-{environment}.example.com, https://web-{environment}.example.com) that are not real or resolvable.
  • Execute() always returns 0 — there is no failure path, because nothing real is attempted that could fail.

When to use it

Use pks deploy only to script a terminal recording or demo of what a deployment CLI experience could look like — progress bars, a checks table, a results panel, a live metrics view — for product screenshots, GIFs, or as a Spectre.Console UI reference. Do not run it expecting any real infrastructure change: nothing is built or pushed anywhere, and no service is contacted.

For an actual deployment need in this CLI, use one of these instead:

  • pks coolify — registers a real Coolify instance for application deployment.
  • pks ssh — registers real SSH remote targets, including for devcontainer deployment.

Prerequisites

None. pks deploy has no external dependencies, no authentication, no network calls, and no file I/O — it runs standalone on any machine with pks installed.

Synopsis

pks deploy [options]

deploy    Runs a scripted, simulated deployment sequence for demo purposes

pks deploy has no subcommands.

Options

FlagDefaultDescription
-e, --environment <ENV>devTarget environment label. Accepts any string — the code only uppercases it for display (settings.Environment.ToUpper()); there is no validation restricting it to dev/staging/prod despite the flag description implying those three.
-w, --watchfalseAfter the results panel, run a fixed 20-second live-refreshing metrics table.
--ai-optimizefalsePrint two extra lines, including a fixed "23% faster startup, 15% less memory usage" claim. No model or service call is made.
-r, --replicas <COUNT>3Echoed into the results panel as "{Replicas} instances running". Does not affect any real scaling — no orchestrator call exists to scale.

Examples

pks deploy

Runs the simulated flow against the default dev environment with 3 replicas: a green checks table, four animated progress bars, then a results panel with fabricated endpoint URLs and a fixed ~42ms response time.

pks deploy --environment staging --replicas 5

Same simulated flow, but the results panel echoes STAGING and 5 instances running. --environment and --replicas are cosmetic string substitutions here, not real targeting or scaling parameters.

pks deploy --ai-optimize

Adds "AI optimization enabled - analyzing optimal deployment strategy..." and the fixed "23% faster startup, 15% less memory usage" line to the results panel. No AI or model call happens anywhere in the command.

pks deploy --watch

After the results panel, enters a 20-iteration (about 20 seconds), 1-second-interval live table of Random-generated CPU, memory, requests, and status values for three fictitious services — API Gateway, Database, and Cache — then exits on its own.

Troubleshooting

Note. pks deploy cannot fail. Execute() always returns 0 because no real build, push, or deployment is attempted — there is nothing for the command to fail at. A green results panel is not evidence that any application was deployed.

"The printed endpoint URLs don't resolve." They are literal placeholder text (https://api-{environment}.example.com, https://web-{environment}.example.com), not real or project-specific URLs. Do not copy them into user-facing output or documentation as if they were real.

"--watch didn't respond to Ctrl+C the way the on-screen hint suggested." The watch loop is bounded to 20 iterations and exits automatically; it was never designed to run indefinitely, so there is no graceful-stop behavior to trigger. Ctrl+C during the loop kills the process, as it would for any command.

"I need pks deploy to actually deploy something." It never will — there is no build, registry push, or cluster call in its implementation. Use pks coolify to register a real Coolify instance, or pks ssh to register a real SSH target for devcontainer deployment.

See also

  • pks coolify — the command group that actually deploys applications, via a registered Coolify instance.
  • pks ssh — register real SSH remote targets, including for devcontainer deployment.
  • pks — the full command surface this group belongs to.