agentleFS
TsukiNi22TsukiNi22/ai-utils/skills/ai/interactive-questions/SKILL.md

interactive-questions

Skill1 starsChanged 3 days agoMITView on GitHub

Ask the user through the built-in interactive questions (AskUserQuestion, a QCM / Q&A popup with options) instead of a list in the text - questions, proposals, ideas, choices, rounds of ideas, a decision between approaches, what to do next - in Tsukini's way (French labels, 1-4 questions per call, 2-4 options each, the recommended one first, multi choice when they add up, previews to compare, the answers read and followed exactly). Use whenever the user says "propose en qcm", "q&a", "demande-moi", "pose-moi des questions", "propose plusieurs choix / idées", "laisse-moi choisir", or when several options exist and the choice is really theirs.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When the user has to choose, ask with the AskUserQuestion tool (the built-in popup: options, *Other* typed, multi choice, previews), never …

    line 10
  2. Distinctive:

    Not for what can be decided from the request, the code or a sensible default (pick it, say it in one line), nor to ask "should I go on?" / …

    line 20
  3. Distinctive:

    Read them exactly: a free text (*Other*) or the notes often change the option (a constraint, "mais...", "et aussi...", a mix of two): …

    line 40
  4. Distinctive:

    No answer to a question (skipped): don't guess what matters.

    line 42

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Interactive questions (QCM / Q&A)

When the user has to choose, ask with the AskUserQuestion tool (the built-in popup: options, Other typed, multi choice, previews), never with a numbered list in the answer. It also works in claude-cluster: the question of a sub-session pops up there (terminal, window, web page, phone, the voice), and Global answers lets its global session answer it.

When

  • The user asks for it: "propose en qcm / q&a", "demande-moi", "pose-moi des questions", "propose plusieurs idées / rendus / choix", "laisse-moi choisir", "si tu as d'autres idées propose-les".
  • A real decision of theirs: several valid approaches with different costs, a name, a visual style, what to keep, the scope of a change, which ideas of a brainstorm go in the backlog.
  • Not for what can be decided from the request, the code or a sensible default (pick it, say it in one line), nor to ask "should I go on?" / "is it OK?" (do it; a plan is confirmed by its own mechanism).

How

  • One call with 1 to 4 questions asked together (related ones in the same popup), each with 2 to 4 options; Other is added by the tool: never write an "Other" option.
  • Text in the language of the user (French here): question ends with ?, header ≤ 12 characters (a chip: "Nom", "Style", "Portée"), label 1-5 words, description = what it means, its trade-off (cost, risk, time).
  • The recommended option first, its label ending with (Recommended); at most one per question.
  • multiSelect: true when the options add up (ideas to keep, features, files, notifications...); false when they exclude each other.
  • preview (single choice only) to compare concrete things side by side: ASCII mockups of a UI, code, a config, a diagram, an icon / logo drawn in text.
  • A whole ideation (new ideas on a subject, recommendations, an opinion, variants of ideas): the ideation skill.
  • Many ideas: several rounds (4 questions × 4 options max per call), grouped by theme ("Claude Code", "Confort", "Contrôles"...), then a short recap of what was kept (and stored as tasks / backlog when asked).
  • A question whose answer needs explaining first: explain in a few lines, then ask. "Explique plus" / "clarifie" as an answer: detail each option (what happens, an example, its risks), then ask again with the same options refined.

The answers

  • Read them exactly: a free text (Other) or the notes often change the option (a constraint, "mais...", "et aussi...", a mix of two): follow what is written, not the label alone.
  • No answer to a question (skipped): don't guess what matters; ask it again alone, or take the recommended one only when it is harmless and say so.
  • Remember them for the rest of the conversation (never ask the same thing twice); a choice that must last (preferences, style, a name) goes where it lasts (memory, CLAUDE.md, the project config) when the user wants it.
  • Then act at once on the choices, and say in one line what was chosen.

Example

{"questions": [
  {"question": "Quel nom pour le skill ?", "header": "Nom", "multiSelect": false, "options": [
    {"label": "ask-choices (Recommended)", "description": "Court et explicite"},
    {"label": "qcm", "description": "Le mot que tu utilises, moins parlant en anglais"}]},
  {"question": "Quelles idées garder ?", "header": "Idées", "multiSelect": true, "options": [
    {"label": "Rewind", "description": "Revenir à un point de la conversation"},
    {"label": "Revue du plan", "description": "Valider un plan avant qu'il soit exécuté"}]}
]}

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.