agentleFS
M1688-cpuM1688-cpu/ask-first/skills/clarify-first/SKILL.md

clarify-first

Skill1 starsChanged 38 days agoMITView on GitHub

Interview the user one question at a time before starting an ambiguous task, then confirm a written brief before acting. Use when a request is underspecified: the goal, scope, deliverable, or a key decision could each be read several ways, or the work is large enough that guessing wrong wastes real effort. Skip for clear, small, mechanical requests.

4 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions4 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 4

    1. FaisalAlqarni/sp-ecc★ 9same name
    2. DmiyDing/clarify-first★ 6same name
    3. geniusBrainLsm/clarify-first★ 4same name
    4. OrionicsLtd/claude-skills★ 1same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repositoryOne other file it gives its agents.

The file

We split this skill into the rules it gives an agent

11 rules. 0 are common, 11 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 11

    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:

    After reading the prompt, rate four things: goal, scope, deliverable, key decisions.

    line 21
  2. Distinctive:

    Look it up yourself: anything the codebase, docs, or the web can answer.

    line 33
  3. Distinctive:

    Only the user knows: intent, taste, priorities, constraints of their environment, budget of time or money.

    line 35
  4. Distinctive:

    Ask it alone with AskUserQuestion: one question per call, 2-4 options, each option with a one-line description.

    line 43
  5. Distinctive:

    Stop when the remaining unknowns could no longer change the approach or the deliverable.

    line 49
  6. Distinctive:

    Ask in the language the user writes in.

    line 53
  7. Distinctive:

    If the user picks the recommended option three times in a row, offer to accept all remaining recommendations at once.

    line 59
  8. Distinctive:

    Never batch several questions into one AskUserQuestion call during the interview.

    line 61
  9. Distinctive:

    Follow-ups depend on earlier answers.

    line 62
  10. Distinctive:

    Write a task brief and get one confirmation:

    line 67
  11. Distinctive:

    Only after a yes does work begin.

    line 80

“Common” is measured: kept by at least 6 in 10 of the 4 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

clarify-first: ask until the task is clear, then start

One vague instruction can send an agent hours in the wrong direction. This skill trades a few focused questions for that risk.

Step 0: decide whether to interview

After reading the prompt, rate four things: goal, scope, deliverable, key decisions.

  • Two or more are unclear, or the task is large and hard to reverse: interview, continue below.
  • Everything is clear: skip the interview, do the task. State your understanding in one sentence before starting.
  • The user says "don't ask, just do it": skip to the brief (step 3), no interview.

Step 1: build the question list

List what you would need to know to pick one approach over another. Sort every item into two piles before asking anything:

  • Look it up yourself: anything the codebase, docs, or the web can answer. These are never asked.
  • Only the user knows: intent, taste, priorities, constraints of their environment, budget of time or money. These are the interview questions.

Order them so an early answer can remove or reshape the questions after it.

Step 2: ask one at a time

  1. Pick the highest-value open question.
  2. Ask it alone with AskUserQuestion: one question per call, 2-4 options, each option with a one-line description. Put your recommended option first with "(Recommended)" at the end of its label. The user can always pick "Other" and type freely. If the host has no AskUserQuestion tool, ask the same single question in chat, listing the options as a numbered list.
  3. Derive follow-up questions from the answer and add them to the list.
  4. Stop when the remaining unknowns could no longer change the approach or the deliverable. That is the 95% bar: the questions left are ones whose answers would not change what gets built.

Ask in the language the user writes in.

Guard rails:

  • About 7 questions per task is the ceiling. Past that, you are asking things you could have looked up.
  • If the user picks the recommended option three times in a row, offer to accept all remaining recommendations at once.
  • Never batch several questions into one AskUserQuestion call during the interview. Follow-ups depend on earlier answers; that dependency is the point of asking one at a time.

Step 3: confirm the brief

Write a task brief and get one confirmation:

  • Goal: one sentence.
  • In scope: short list.
  • Out of scope: what you will deliberately not touch.
  • Deliverable: what exists for the user at the end.
  • Assumptions: anything you decided without asking.

Then ask with AskUserQuestion, question "Ready to start on this brief?":

  • "Yes, start" (Recommended)
  • "Change something" (ask what, fix the brief, confirm again)

Only after a yes does work begin. If new ambiguity appears mid-task, solve small ones yourself and record them under Assumptions; come back and ask only when the answer would change the deliverable.

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.