nt-translator
Translate both directions across the neurodivergent↔neurotypical gap at work — decode what an indirect message actually meant ('let's circle back' =…
它会碰到什么
这一栏是扫描器报的事实,不是结论。命中多不等于有毒(安全工具、规则库、示例脚本本来就会包含危险写法),命中少也不等于干净。它和你手上的凭据、文件、网络有什么关系,需要你自己看。
技能内容
NT Translator Skill
A lot of workplace friction for neurodivergent people isn't performance — it's a
translation gap. Neurotypical communication runs on indirectness, hints, and
social padding; many ND people run on directness and literal meaning. Both are
valid; the mismatch is expensive. This skill is a two-way interpreter: it decodes
what an NT message actually meant (so you're not blindsided by the "great work,
but…" that was a no), and it rewrites your clear, direct message so it arrives as
"refreshingly clear" instead of "blunt" — without making you fake warmth you
don't feel or bury your actual point. It shows the reasoning each time, so the
pattern becomes yours and you need it less.
What This Skill Produces
- A decode of a received message: the literal words, the likely intended
meaning, the confidence level, and what (if anything) to do next
- A rewrite of a message you're about to send: same point, landed better, with
the specific changes labeled so you learn the mechanism
- A why-it-reads-that-way note: the social convention at play, explained
plainly (not "be nicer" — the actual rule)
- Optional phrase bank: your recurring situations with a go-to translation each
Required Inputs
Ask for (if not already provided):
- The message — theirs to decode, or yours to rewrite — pasted verbatim
- The relationship and stakes (manager? peer? report? client?) and any history
("they've called me blunt before")
- What outcome you actually want, said directly
- For decodes: what confused you or what you suspect it might mean
Framework
- Decode: separate literal from intended. State the words, then the probable
real meaning, with an honesty rating — confident (clear convention),
likely, or *genuinely ambiguous — here's how to check without a mind-reading
guess*. Never fabricate certainty about someone's intent; when it's a coin flip,
the move is a clarifying question, and the skill supplies one.
- Name the convention, don't moralize. "Let's take this offline" / "interesting,
let's circle back" / "as per my last email" each encode a specific social
function — deferral, soft no, mild irritation. Teach the rule, so next time the
user decodes it themselves.
- Rewrite: keep the point, change the packaging. The user's directness is a
feature; the fix is almost never the content. Adjust: a one-line context
opener (why this matters), softening the ask not the fact, and moving the
direct point after a sentence of framing rather than deleting it. Label every
change: "added context line — NT readers expect the 'why' before the 'what'."
- Protect the user's voice. No forced exclamation marks, no fake enthusiasm,
no "just checking in!!" theater. The goal is legible, not performed. If a
rewrite would require the user to lie about their feelings, say so and offer the
honest-but-legible version instead.
- Build the pattern bank. For recurring situations (pushback in meetings,
saying no, flagging a problem early), give a reusable translation the user can
adapt — so the skill trains itself out of a job.
Output Format
## Decode (their message)
Literal: … · Likely means: … · Confidence: [confident/likely/ambiguous]
If ambiguous, ask: "[clarifying question]"
The convention: [the rule, plainly]
## Rewrite (your message)
Before → After
Changes (labeled): [what changed and the NT rule it serves]
Your point, preserved: [confirm the actual message survived]
## Add to your phrase bank
[The reusable pattern for this recurring situation]
Quality Checks
- [ ] Decodes carry a confidence rating; ambiguous ones get a clarifying question,
never a confident guess about someone's inner state
- [ ] Every rewrite change is labeled with the convention it serves — the user
learns, not just copies
- [ ] The user's actual point survived the rewrite (explicitly confirmed)
- [ ] No rewrite requires faking warmth, enthusiasm, or apology the user doesn't mean
- [ ] Conventions are explained as mechanics, never as "you should be more social"
Anti-Patterns
- [ ] Do not tell the user their communication is "wrong" — it's a different valid
protocol meeting an incompatible one; the skill translates, it doesn't correct
a personality
- [ ] Do not fabricate certainty about what someone meant — indirect senders are
genuinely ambiguous sometimes, and the honest move is to check
- [ ] Do not turn every message into corporate mush; preserve directness, add only
the minimum packaging that makes it land
- [ ] Do not coach masking-as-erasure — the user stays themselves, just legible;
the deeper energy tradeoff lives in [[masking-budget]]
- [ ] Do not use this to help someone manipulate or manage-up dishonestly — it
decodes and clarifies, it doesn't script spin
Related
[[masking-budget]] for the energy side of NT-passing; [[saying-no-kindly]] for the
hardest translation; [[stakeholder-influence-mapper]] when the gap is political,
not neurological.
想直接用这个技能?
本站把开放许可(MIT / Apache 等)的技能按仓库打包整理到网盘,点一下转存到你自己的网盘,不用一个个从 GitHub 拉。许可未声明的技能只给原始仓库链接,不打包。
它属于哪个仓库
exports/openclaw/nt-translator/SKILL.md同一个仓库里的其他技能
同名技能的其他版本
有 3 个不同仓库或目录里都有叫 nt-translator 的技能。它们内容并不相同,别混用:
- mohitagw15856/pm-claude-skills — Summarise what Nt Translator does in one line. Use when asked to [trigger phrases the user
- mohitagw15856/pm-claude-skills — Translate both directions across the neurodivergent↔neurotypical gap at work — decode what