unasked-questions
Skill0 starsChanged 36 days agoMITView on GitHub
Find the question a prompt is really asking under the one it states, and write it under every prompt in an application file. Use when extracting questions from an application, before drafting, and when auditing a draft.
In this repository9 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- application-typesSkill
- application-writingSkill
- honesty-guardSkill
- interviewing-youSkill
- mock-interviewSkill
- no-slop-writingSkill
- org-researchSkill
- reader-povSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 6 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Every prompt has a stated ask and an unstated one.
- Distinctive: line 17
Ask four questions about the prompt:
- Distinctive: line 23
2. What is the failure mode they are screening out?: Every prompt exists because somebody bad got in once.
- Distinctive: line 32
4. What will the reader think and never write down?: If the applicant has shipped real things, the reader is quietly asking whether this …
- Distinctive: line 55
If the honest read is uncomfortable, write it anyway.
- Distinctive: line 86
Before drafting: read the unasked list for that question first.
- Very specific: line 95
When the same unasked question runs under prompts at four different orgs, put it in applications/INDEX.md once, at the top, instead of …
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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.
Files
The unasked question
Every prompt has a stated ask and an unstated one. The unstated one is what gets scored. An org asks "tell us about a time you worked on a team" and decides something about you that has nothing to do with teams.
This is the single highest-leverage thing in the kit. Do it for every prompt in every application file, before any drafting starts.
How to find it
Ask four questions about the prompt:
1. What decision is this org making with this answer? Not "what do they want to know." What are they deciding. A pro-bono consulting org asking about a social issue is deciding whether you will still show up in week nine of an unglamorous project.
2. What is the failure mode they are screening out? Every prompt exists because somebody bad got in once. "Describe a conflict with a teammate" exists because they admitted someone who could not be told they were wrong.
3. What does this specific org worry about, given who applies to them? A startup org worries about people who love the idea of startups. A service org worries about résumé-builders. An org that rejected you last year is asking, without saying it, what is different now.
4. What will the reader think and never write down? If the applicant has shipped real things, the reader is quietly asking whether this person will take direction. If the applicant is a freshman, whether they will follow through. Name that.
How to write it
Under each prompt in the applications/ file, in a quoted block:
### Q2. Describe a product that changed how you think or act. (200 words)
> **Unasked:** Do you have taste, or do you just consume? Anyone can name an app they
> like. They want evidence you notice mechanism: why it works on you, what it was
> designed to do, whether you can separate the two. And underneath: is your interest
> in building things actually about anything, or is it about wanting to have built
> something.Rules for these blocks:
- Two to four sentences. Longer and it stops being usable at write time.
- Name the decision, not the theme. "They are deciding whether you can be told you are wrong" beats "this is about teamwork."
- If the honest read is uncomfortable, write it anyway. That is the entire value.
- One per prompt, including the fun ones. Especially the fun ones.
Patterns worth knowing
These run under prompts in nearly every type:
| Stated prompt | Usually unasked |
| Why us | Why not the five others you are obviously also applying to |
| Tell us about yourself | Can you pick one thing and go deep, or do you list |
| A time you failed | Can you look at yourself honestly without turning it into a win |
| A challenge you overcame | Do you know the difference between hard and inconvenient |
| A time you disagreed | Can you be wrong out loud |
| What do you want to learn | Are you here to learn or to be impressive |
| Your strengths and weaknesses | Do you know how you are hard to work with |
| Reapplying? Tell us more | What is actually different since we told you no |
| Optional additional materials | Do you take the free opportunity or not |
The optional-materials one deserves emphasis. Declining an optional invitation while reapplying answers the real question badly. Almost nobody takes it, which is exactly why taking it works.
Then read the table in this application's reference/ file. Each type has its own
set, and they are more useful than the general list because they carry what that
specific kind of reader worries about. A grant reviewer reading your budget and a
student-org officer reading your favorite snack are both asking something, and it is
not the same something.
Using them
Before drafting: read the unasked list for that question first. It usually changes which story you pick, not just how you tell it.
After drafting: hand a reader only your answer, nothing else. Could they answer each unasked question from it? Three misses is your revision, and the revision is usually a swap of story rather than a rewrite of sentences.
Roll the repeats up
When the same unasked question runs under prompts at four different orgs, put it in
applications/INDEX.md once, at the top, instead of repeating the analysis. Those
repeated ones are the real shape of the whole cycle, and seeing them together tells
the applicant what they are actually being evaluated on this fall.
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
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.

