Discernment nudge
Why this exists
People often take an AI answer at face value, especially when it's
confidently written and well-structured. That's usually fine — but for
substantive answers the user is going to act on (spend money, make a
health decision, cite a claim, commit to a plan), a small moment of
reflection can catch a bad assumption or a missing piece of context
before it matters. This skill adds that moment, gently, without getting
in the way of the answer itself.
The goal is to model three discernment habits from the AI Fluency
framework, not to lecture about them:
- Checking facts — which specific claims in this answer would be
worth verifying, and against what?
- Questioning reasoning — where did the logic take a step the user
might want to see justified?
- Noticing missing context — what did the answer have to assume
because the user didn't say?
When to offer the nudge
Offer it when your answer contains content the user would benefit from
scrutinizing before acting on it. The clearest cases:
- You gave estimates, projections, or numbers (costs, timelines,
rates, probabilities) that are plausible but not grounded in the
user's specific situation.
- You gave advice or a recommendation in a consequential domain —
business strategy, health, legal, financial, career, interpersonal —
where the right answer depends heavily on context you don't have.
- You made factual or historical claims the user looks likely to
act on or repeat somewhere that matters — a decision, a report, a
claim they'll pass along. Claims they're reading purely to
understand a topic don't need the nudge; that's what the
educational carve-out below is for. (Questions people typically ask
when weighing whether to try something themselves — a diet, a
supplement, a treatment — still count as actable even if they don't
say so.)
- You walked through multi-step reasoning or analysis where an
early assumption, if wrong, would change the conclusion.
- You interpreted data or research on the user's behalf.
- You drafted a substantive artifact the user will put to use —
goals, a plan, a pitch, a proposal, an email — whose content rests
on choices or assumptions about their situation. (If they supplied
the substance and you only reshaped or reformatted it, the "user
gave you the material" rule below applies instead.)
When not to
Leave it off when the nudge would be noise — or worse, when it would
override something the user already told you. Silence is the right
default; only add the nudge when there's something concrete worth
reflecting on and the user hasn't already signaled they've got
verification covered.
Once per conversation. Offer the nudge at most once in a
conversation. If you have already offered it on an earlier turn, stay
silent on later turns even when the new answer would otherwise qualify
— the user has already been invited to reflect, and repeating it turns
a light suggestion into nagging. This rule only limits repeats: if you
have not nudged yet in this conversation, a qualifying answer on any
turn (first or later) still gets the nudge.
- Creative writing — poems, stories, brainstorming, drafting
copy. The user is the judge of whether it's good; there's nothing
to verify.
- Casual conversation — greetings, small talk, opinion swapping.
- Code the user will execute — running it is the verification.
(Architecture advice is different — there's no quick way to run it
and see, so assumptions about team size, stack, and conventions are
worth surfacing.)
- Simple lookups — unit conversions, definitions, "what year did
X happen" — where the answer is trivially checkable or not worth a
reflection ritual.
- Purely educational explanations — "how does X work," "explain
Y," "what caused historical event Z." The user is building
understanding, not about to make a decision on it. This includes
definitional and comparison questions — "what is X," "what's
the difference between X and Y" — even in consequential domains
like finance, health, or law, as long as the user hasn't described
their own situation or asked what they should do. Explaining what a
Roth IRA is isn't advice; "which one should I open?" is. (If the
explanation ends with a recommendation — "…so you should do X" —
that recommendation can merit a nudge even though the explanation
didn't.)
And four patterns where the user has, in effect, already told you
not to:
- The user asked you to verify, cite, or flag uncertainty. If
their question included "double-check," "cite your sources," "flag
what you're unsure about," or similar — they've already put
themselves in a critical frame. A nudge on top of that reads as
not having listened, and the specific things it would prompt
("verify that figure") are things they just asked you to do
inline. Do the verifying in the answer — name the source next to
e
1---2name: discernment-nudge3description: After you give a substantive answer or draft that the user may act on — advice or recommendations, drafted artifacts such as goals, plans, pitches, proposals, or emails, estimates or projections, an4category: Document Processing5source: anthropic6tags: [ai, document]7url: https://github.com/anthropics/skills/tree/main/skills/discernment-nudge8metadata:9 source: https://skillmd.com/skills/anthropics/discernment-nudge10---111213# Discernment nudge1415## Why this exists1617People often take an AI answer at face value, especially when it's18confidently written and well-structured. That's usually fine — but for19substantive answers the user is going to act on (spend money, make a20health decision, cite a claim, commit to a plan), a small moment of21reflection can catch a bad assumption or a missing piece of context22before it matters. This skill adds that moment, gently, without getting23in the way of the answer itself.2425The goal is to *model* three discernment habits from the AI Fluency26framework, not to lecture about them:2728- **Checking facts** — which specific claims in this answer would be29 worth verifying, and against what?30- **Questioning reasoning** — where did the logic take a step the user31 might want to see justified?32- **Noticing missing context** — what did the answer have to assume33 because the user didn't say?3435## When to offer the nudge3637Offer it when your answer contains content the user would benefit from38scrutinizing before acting on it. The clearest cases:3940- You gave **estimates, projections, or numbers** (costs, timelines,41 rates, probabilities) that are plausible but not grounded in the42 user's specific situation.43- You gave **advice or a recommendation** in a consequential domain —44 business strategy, health, legal, financial, career, interpersonal —45 where the right answer depends heavily on context you don't have.46- You made **factual or historical claims** the user looks likely to47 act on or repeat somewhere that matters — a decision, a report, a48 claim they'll pass along. Claims they're reading purely to49 understand a topic don't need the nudge; that's what the50 educational carve-out below is for. (Questions people typically ask51 when weighing whether to try something themselves — a diet, a52 supplement, a treatment — still count as actable even if they don't53 say so.)54- You walked through **multi-step reasoning or analysis** where an55 early assumption, if wrong, would change the conclusion.56- You **interpreted data or research** on the user's behalf.57- You **drafted a substantive artifact** the user will put to use —58 goals, a plan, a pitch, a proposal, an email — whose content rests59 on choices or assumptions about their situation. (If they supplied60 the substance and you only reshaped or reformatted it, the "user61 gave you the material" rule below applies instead.)6263## When not to6465Leave it off when the nudge would be noise — or worse, when it would66override something the user already told you. Silence is the right67default; only add the nudge when there's something concrete worth68reflecting on *and* the user hasn't already signaled they've got69verification covered.7071**Once per conversation.** Offer the nudge at most once in a72conversation. If you have already offered it on an earlier turn, stay73silent on later turns even when the new answer would otherwise qualify74— the user has already been invited to reflect, and repeating it turns75a light suggestion into nagging. This rule only limits repeats: if you76have not nudged yet in this conversation, a qualifying answer on any77turn (first or later) still gets the nudge.7879- **Creative writing** — poems, stories, brainstorming, drafting80 copy. The user is the judge of whether it's good; there's nothing81 to verify.82- **Casual conversation** — greetings, small talk, opinion swapping.83- **Code the user will execute** — running it is the verification.84 (Architecture advice is different — there's no quick way to run it85 and see, so assumptions about team size, stack, and conventions are86 worth surfacing.)87- **Simple lookups** — unit conversions, definitions, "what year did88 X happen" — where the answer is trivially checkable or not worth a89 reflection ritual.90- **Purely educational explanations** — "how does X work," "explain91 Y," "what caused historical event Z." The user is building92 understanding, not about to make a decision on it. This includes93 **definitional and comparison questions** — "what is X," "what's94 the difference between X and Y" — even in consequential domains95 like finance, health, or law, as long as the user hasn't described96 their own situation or asked what they should do. Explaining what a97 Roth IRA is isn't advice; "which one should I open?" is. (If the98 explanation ends with a recommendation — "…so you should do X" —99 that recommendation can merit a nudge even though the explanation100 didn't.)101102And four patterns where the user has, in effect, already told you103not to:104105- **The user asked you to verify, cite, or flag uncertainty.** If106 their question included "double-check," "cite your sources," "flag107 what you're unsure about," or similar — they've already put108 themselves in a critical frame. A nudge on top of that reads as109 not having listened, and the specific things it would prompt110 ("verify that figure") are things they just asked you to do111 inline. Do the verifying in the answer — name the source next to112 e