Interview the user with sharp, one-at-a-time questions to turn a vague request into a tight, complete brief any other skill can run on.
SkillScore73Maintained
Automatic
◆Product Management pack
Last verified 8/14/2026
What it changes
Passed 16 of 22 test cases with the skill, and 10 of 22 without it.
Without skill45%
With skill73%
Checks it passed that the model alone missed
Asks at most two questions in the initial turn
Asks at most two questions in the initial response
Initiates an interactive question sequence instead
and 8 more in the full report
10 fixed6 passed either way2 still failing4 broke
Pass-rate lift
+27 pp
Tokens
-7%
Agent turns
0%
Response time
-56%
Model gemini-3.6-flashMeasured 8/14/2026v116 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 10 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 10
Prompt
Sales wants a customer case study video script. Start interviewing me to create the brief.
What was checked, without and with the skill
Checked
Without skill
With skill
Limits the question batch to maximum two questions
Missed
Passed
case-12 · gemini-3.6-flash · 8/14/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md69 linessingle file
Brief Builder Skill
Most weak AI output comes from a weak brief — the model guessed at context instead of getting it. This skill
flips that: it interviews the user with focused questions, one or two at a time, following up where
answers are thin, until it has enough to produce excellent work. Then it writes the brief and hands off. The
whole value is asking the right questions in the right order — not producing on the first vague sentence.
Required Inputs
The raw ask — whatever the requester actually said, however vague ("we need a dashboard", "marketing wants a one-pager"). Verbatim beats paraphrased; the gaps in their words are the interrogation map.
Who is asking and who will consume the output (if known) — the same ask from a CEO and an intern needs different briefs.
Optional: any prior artifacts the ask relates to (the doc being replaced, the thread that sparked it).
How to run this skill (the interrogation loop)
Read what they gave you and identify the task type (a launch? a doc? a decision? a piece of copy?).
Ask the smallest set of high-leverage questions first, ONE or TWO at a time — never a 20-question wall. Lead with the questions whose answers most change the output.
Follow up when an answer is vague ("everyone" → "who specifically?"; "soon" → "what date?"). Dig until it's concrete.
Offer defaults: when the user doesn't know, propose a sensible default and let them confirm ("I'll assume B2B SaaS founders unless you say otherwise").
Stop when you have enough — don't over-interview. Then summarize the brief and confirm before producing.
The question backbone (adapt to the task)
Goal — what does success look like? What decision or action should this drive?
Audience — who is this for, specifically? What do they already know / believe?
Context — what exists already? What's the backstory, the constraint, the deadline?
Scope & format — how long, what format, where will it live?
Voice & guardrails — tone, must-says, can't-says, examples to match.
Success criteria — how will they judge if the output is good?
Output Format
1. The questions (interactive)
Ask them conversationally, batched 1–2 at a time, easiest path first. (Do not dump the whole list at once.)