agentleFS
grinbuddygrinbuddy/claude-skills/skills/ask-me/SKILL.md

ask-me

Skill0 starsChanged 35 days agoMITView on GitHub

Turn everything currently waiting on the user into multiple-choice questions, each with a recommended answer listed first. Use when the user says "ask-me", "ask me", "what do you need from me" as a request to be asked, "walk me through the decisions", "give me choices", or otherwise wants pending decisions put to them as pick-one questions instead of prose.

15 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use handoff.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • MicrosoftSkill

    handoff

    Compact the current conversation into a handoff document for another agent to pick up.

    Also in 4 projects that use this

The file

We split this skill into the rules it gives an agent

12 rules. 0 are common, 12 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 11 edited versions we read keep these.

  • 12

    Distinctive

    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:

    Look at this conversation, then at the project's own records of open items (a handover, a closeout or "still to do" list, an "owed by" …

    line 14
  2. Distinctive:

    Leave out anything you can settle yourself from the request, the code or a sensible default.

    line 23
  3. Distinctive:

    If nothing is waiting on them, say so in one line and stop.

    line 26
  4. Distinctive:

    Use the question tool that shows tappable options (AskUserQuestion).

    line 30
  5. Distinctive:

    Plain language: No ticket ids or shorthand without a plain description beside it.

    line 38
  6. Distinctive:

    One decision per question: If two things are tied together, ask the first and follow up.

    line 41
  7. Distinctive:

    A recommendation is required even when it is close: If you have none, say which way you lean and why, and mark that option as recommended.

    line 46
  8. Distinctive:

    Ask in batches of up to four questions (the tool's limit), in consecutive calls, until the list is done.

    line 52
  9. Distinctive:

    Someone else's decision: If an item belongs to another person, do not ask the user to answer for them.

    line 63
  10. Distinctive:

    Read every answer back in a short list: the question in a few words, the choice made, and "as recommended" or "differs from the …

    line 68
  11. Distinctive:

    If an answer raises a follow-up question, ask it the same way.

    line 76
  12. Distinctive:

    If the user asks for this to be the default ("always ask me like this"), treat any later point where you need their input the same way: …

    line 80

“Common” is measured: kept by at least 6 in 10 of the 11 edited versions of this file we could read. “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

Ask me

The user wants to clear what is waiting on them by tapping answers, not by reading paragraphs and composing replies. Collect every open item that is theirs, put each to them as a pick-one question with your recommendation first, then act on the answers.

1. Gather what is waiting on them

Look at this conversation, then at the project's own records of open items (a handover, a closeout or "still to do" list, an "owed by" section, the latest decision entry, open questions in a plan). List every item that needs the user specifically:

  • a decision between options;
  • an approval or go-ahead;
  • a fact only they know;
  • an action only they can take (a login, a look at something on their device).

Leave out anything you can settle yourself from the request, the code or a sensible default. Settle those, and say in one line what you chose.

If nothing is waiting on them, say so in one line and stop. Do not invent questions.

2. Turn each item into a question

Use the question tool that shows tappable options (AskUserQuestion). Rules:

  • Recommendation first. The first option is the one you recommend, with "(Recommended)" at the end of its label. Its description says why in one sentence.
  • Two to four options, each a real, different choice. No joke options, no "other": the tool adds a free-text answer on its own.
  • Each description states the consequence of picking it: what happens next, what it costs, what it risks. A reader should be able to choose from the options alone.
  • Plain language. No ticket ids or shorthand without a plain description beside it. The question must stand alone for someone who has lost the thread: give the one sentence of context it needs.
  • One decision per question. If two things are tied together, ask the first and follow up; do not make the user answer a compound.
  • Facts and actions too. A fact becomes a question with the likely answers as options ("Only the two of us" / "Someone else as well" / "Not sure, I'll check"). An action becomes "Done" / "Will do by <when>" / "Skip it", with what each means.
  • A recommendation is required even when it is close. If you have none, say which way you lean and why, and mark that option as recommended. Never present options as equal just to avoid choosing.
  • Order by what blocks the most. The item holding up other work goes first. Things that are safe to leave go last.

Ask in batches of up to four questions (the tool's limit), in consecutive calls, until the list is done. Before the first batch, say in one line how many questions there are and how many are routine.

3. Things that must not be multiple choice

  • A step that is hard to undo or reaches outside (a production deploy, sending a message to other people, deleting data, spending money) still needs the user's own explicit words wherever the project requires them. A tapped option does not replace a typed go-ahead that the project's rules ask for. Ask the question to learn their intent, then show the exact line to type.
  • Someone else's decision. If an item belongs to another person, do not ask the user to answer for them. Ask whether to put it to that person, and how.

4. After the answers

  • Read every answer back in a short list: the question in a few words, the choice made, and "as recommended" or "differs from the recommendation" for each.
  • Where an answer differs from your recommendation, say what changes because of it. Do not argue the point again.
  • Record the answers wherever the project keeps decisions (a decision log, the plan, a handover), in the project's own format, quoting any typed words verbatim. If the project has no such place, the read-back is the record.
  • Then carry on with whatever the answers unblocked, without waiting to be told.
  • If an answer raises a follow-up question, ask it the same way.

5. Standing use

If the user asks for this to be the default ("always ask me like this"), treat any later point where you need their input the same way: recommendation first, options with consequences, no prose question when a pick-one will do. Save that preference where the session keeps the user's preferences.

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.