promptify

name	promptify
description	USER-INVOKED ONLY. Use only when Augustin explicitly invokes $promptify or explicitly names this skill with one raw prompt argument. Transform a rough dictated instruction into a clean GPT-5.5-optimized prompt. If blocking ambiguities remain, do not produce the final prompt; route to $grill-me first.

Promptify

Input

One parameter only:

prompt: <raw dictated instruction>

If no prompt is provided, ask for the raw prompt. Do not ask a questionnaire upfront.

Purpose

Turn a messy intention into a prompt that GPT-5.5 can execute reliably: explicit goal, context, constraints, source-of-truth boundaries, allowed actions, output shape, validation, and stopping rules.

GPT-5.5 Prompt Shape

Prefer a compact, outcome-first prompt:

Objective:
Context:
Inputs:
Constraints:
Allowed actions:
Clarify first if:
Output:
Validation:
Stop when:

Only include sections that matter. Keep the final prompt shorter than the raw prompt when possible.

Ambiguity Gate

Before writing the final prompt, check for blockers:

• unclear objective or success criteria;
• missing target audience, repo/path, channel, or artifact;
• unclear write/send/commit permission;
• ambiguous source of truth;
• hidden tradeoff that changes the work;
• missing output format;
• high-stakes or irreversible action without confirmation.

If any blocker remains, do not output the final prompt. Instead, invoke $grill-me with the smallest clarification brief:

$grill-me
Goal: clarify this raw prompt before Promptify writes the final GPT-5.5 prompt.
Raw prompt: <raw prompt>
Blocking ambiguities: <short list>
Ask one question at a time. Stop when the final prompt can be written safely.

After $grill-me resolves the blockers, resume and produce the final prompt.

Rewrite Rules

• Preserve the user's real intent; do not add ambition, scope, or claims.
• Remove filler, repetitions, speech tics, and chronology noise.
• Convert vague verbs into concrete actions.
• Make side effects explicit: edit files, send mail, commit, browse, ask, or report-only.
• Put uncertainty in Clarify first if, not hidden assumptions.
• Add validation only when it changes reliability.
• Do not include chain-of-thought requests.

Output

If clear, output only the final copy-pasteable prompt. Do not add labels, headers, introductions, explanations, fences, or trailing commentary.

<copy-pasteable prompt>

If not clear:

Ambiguites bloquantes:
- ...

Je lance $grill-me avant de generer le prompt final.