Generate

Generate a wizard

A wizard is a bash script that walks a human, step by step, through a manual procedure that's tedious to do by hand and tedious to re-explain every time. It opens each URL, says exactly what to click and copy, captures the values, writes them where they belong (.env, CI secrets), confirms at every stage, and shows how many stages are left.

The UX and the security hardening are already solved by template.sh. Stage-by-stage progress, fail-closed TTY-only prompts, https-only URL opening (cross-platform incl. WSL and Git Bash), hidden secret entry, quoted 0600 .env upserts with a gitignore check, repo-confirmed gh secret/gh variable writes over stdin, and a closing names-only summary. Your job is only to scope the procedure and author its stages. The library above the STAGES marker is identical in every wizard; that consistency is the point. Never hand-edit it.

A wizard is ephemeral by default. It is built for one run, saved to a scratch or scripts/ path, deleted when the job's done. Commit it only when the user wants a repeatable setup path that should live in the repo.

The generated script requires bash; on Windows the supported path is Git Bash or WSL. gh (authenticated) is needed only by stages that write CI secrets or variables. When it's absent those stages warn and land in the closing summary instead of failing the run.

Process

1. Scope the procedure

Work out every manual step the human must take and every value that gets captured along the way. Read the repo first. Don't ask cold:

• For setup: read .env.example, README, docker-compose*, framework config, and .github/workflows/* fully (every secrets.* / vars.* reference is a value the wizard must produce). From a live .env, take key names only, e.g. grep -oE '^[A-Za-z_][A-Za-z0-9_]*=' .env. Never values.
• For a migration or transition: the current state, the target state, and the irreversible actions between them.

Be honest about what reaches the model if the user asks: values the wizard captures at runtime never reach the model. The human runs the script and it writes captures straight to .env or gh. Authoring-time reads are names-only by the rule above. A value the user pastes into the chat, though, is in context like any other pasted text.

Then show the user the ordered list of stages and the values each produces, and confirm. They may add, drop, or reorder.

Done when: every stage is named in order, and for each captured value you know (a) where the human gets it, (b) where it's written (.env, a CI secret, both, or nowhere; some stages are pure actions), and (c) whether it's secret (hidden entry) or public.

2. Map each stage's journey

For each stage, write the precise path a human follows: which URL to open, what to do there, where a value is shown, which variable it fills, e.g. "Dashboard → Developers → API keys → Reveal test key → copy". Where you don't actually know the current UI or the exact command, say so and ask the user or check the docs. Never invent steps that may not exist.

Done when: every stage traces to concrete instructions a stranger could follow.

3. Author the wizard

Copy template.sh to the target path. Replace the example stage with one stage per step, in dependency order. Use the library helpers: stage, say/step/note/warn, open_url, ask/ask_secret, write_env, set_secret/set_var, pause/confirm. Then set TOTAL_STAGES to the number of stages you wrote.

Hold the bar the template sets: open the URL (https only) before asking for its value, use ask_secret for anything secret, write_env every persisted value, set_secret only the values CI actually needs, and confirm before any irreversible action. Each stage clears the screen so only the current step is visible. Keep a stage to one focused task so nothing the human needs scrolls away. Don't touch the library above the marker.

4. Verify and hand off

1. bash -n <script>; run shellcheck if available. Fix what they find.
2. Trace it statically. Never run it end-to-end yourself: it opens browsers and blocks on human input, and the agent-never-executes line is the security model. Dispatch a fresh-context subagent to do the trace: hand it the script and the step-1 value list (artifact only, not your reasoning), and have it verify that every value from step 1 is captured and lands where step 1 said,that every set_secret/set_var name exactly matches a secrets.*/vars.* reference in CI, and that nothing above the STAGES marker was edited.
3. Stop the line. Human approval gate. Print the full STAGES block (everything below the marker) to the user and get their explicit approval. Do NOT chmod +x the script, and do NOT tell the user to run it, until they have read the stages and approved them. This is a hard ordering, not a suggestion: the human is about to feed real credentials to this script, so the human reads it first.
4. Only after approval: chmod +x <script>, then tell the user how to run it. If it's a repeatable setup path, offer to commit it and link it from the README so the next person runs the script instead of asking an AI.

Gotchas

• No back button. A wrong answer means Ctrl-C and re-run. Cheap by design: values already in .env are offered back as defaults, so the human presses Enter through the stages they got right.
• TTY required. The script refuses to start without /dev/tty, so it will not run piped, in CI, or driven by pasted input. That is deliberate.
• Arrow keys work in ask prompts (readline) but not in ask_secret (hidden entry has no line editing). Backspace works in both.
• gh absence is not an error. CI-secret stages degrade to a warning plus a closing-summary entry telling the human what to set by hand.