Quick Start

Get a temporary Token, save it to the current project .env, then copy one complete prompt to a terminal Agent. No account, workspace, or Skill installation is required.

Get a temporary Token

Create one only when you are ready to publish. It stays in this browser tab and expires automatically.

No temporary Token has been created in this tab

Write it to .env

Run this command from the root of the project you want to publish.

set +o history
umask 077
printf_token='<temporary-token>'
if command -v git >/dev/null 2>&1 && git ls-files --error-unmatch .env >/dev/null 2>&1; then
  printf 'Refusing to write PRINTF_TOKEN: .env is tracked by Git.\n' >&2
else
  if command -v git >/dev/null 2>&1 && git rev-parse --is-inside-work-tree >/dev/null 2>&1 && ! git check-ignore -q .env; then
    printf '\n.env\n' >> .gitignore
  fi
  tmp_file="$(mktemp .env.XXXXXX)"
  if [ -n "$tmp_file" ]; then
    if {
      if [ -f .env ]; then
        while IFS= read -r line || [ -n "$line" ]; do
          case "$line" in
            PRINTF_TOKEN=*) ;;
            *) printf '%s\n' "$line" ;;
          esac
        done < .env
      fi
      printf 'PRINTF_TOKEN=%s\n' "$printf_token"
    } > "$tmp_file" && mv "$tmp_file" .env; then
      printf 'PRINTF_TOKEN saved to project .env\n'
    else
      rm -f "$tmp_file"
      printf 'Failed to update .env\n' >&2
    fi
  fi
fi
unset printf_token tmp_file
set -o history

The copied command contains the Token but the preview does not. It disables shell history, refuses a Git-tracked .env, adds .env to .gitignore when needed, preserves other values, replaces any old PRINTF_TOKEN, and sets private file permissions. Keep it out of chat, screenshots, logs, and source control.

Copy the deployment prompt

The prompt itself is the execution entrypoint. The Agent reads the public technical runbook remotely, uses the Token from this project .env, and fetches only runtime files the project actually needs. There is no Skill installation step.

Copy this entire prompt

Treat this prompt as the complete execution entrypoint for publishing the HTTP service in my current workspace as a public Printf HTTPS URL. Do not install or download an Agent Skill and do not clone the printf-client-releases repository. Read the Moon-hosted technical backup at https://moon.lprintf.com/docs/backup/printf-client-release/docs/getting-started.md, https://moon.lprintf.com/docs/backup/printf-client-release/docs/configuration.md, https://moon.lprintf.com/docs/backup/printf-client-release/docs/api/panel-api-experimental.md, https://moon.lprintf.com/docs/backup/printf-client-release/docs/api/errors.md, and https://moon.lprintf.com/docs/backup/printf-client-release/SECURITY.md. It is a static backup of the official GitHub documentation, so keep using it when GitHub is unreachable. Fetch only the exact Compose template from https://moon.lprintf.com/docs/backup/printf-client-release or a verified release asset required for this project; never pipe remote content directly into a shell.

Load PRINTF_TOKEN from the current project root .env with shell tracing disabled. First verify that .env is not tracked by Git and is ignored; never print or commit the Token. If the project passes its entire .env into application containers or processes, stop before startup and ensure PRINTF_TOKEN is provided only to the Printf Client. If the Token is missing, stop and tell me to obtain a temporary Token at https://moon.lprintf.com/docs/quick-start, or a persistent beta Token at https://moon.lprintf.com/docs/backup/printf-client-release/GET-TOKEN.md; never guess it or ask me to disclose it in chat.

Inspect the project and runtime before changing or starting anything. Use Compose Bridge when Docker, Docker Compose, and the daemon work. Use the verified Native Client only when Docker is absent and the application can run natively; stop for a Compose-only application or a broken Docker environment. For a multi-service site, publish one gateway that serves the frontend and proxies its API and authentication paths. Only that gateway may join both the private network and the Printf gateway network; keep databases, Redis, and internal services private. Do not install extra host networking tools or expose additional host ports for Printf.

One Token may run only one Client. Do not call Admin APIs, refresh the Token, delete user resources, or bypass an existing authentication gateway. Re-read Client and Mapping state after every change, validate the selected target and public HTTPS URL, and report original errors instead of silently falling back. Only for a verified Native release defect, prepare a redacted Issue draft, search for duplicates, and submit it after I approve the public content. Report security issues privately.
Paste it into an Agent that can read the current project, access GitHub, and run terminal commands.Open technical backup

The security boundary is part of the prompt

The Agent must reject a tracked .env, prevent application services from inheriting PRINTF_TOKEN, keep databases and internal services private, and avoid extra host ports. It never asks you to disclose the Token in chat.