agentleFS
spleenteospleenteo/maestro/.claude/skills/translate/SKILL.md

translate

Skill1 starsChanged 46 days agoMITView on GitHub

Translate a draft the owner wrote, in two labeled versions, under the writing register. Runs only in an instance whose `## Writing register` block sets `translation.enabled` to true; elsewhere it hands the request back to ordinary chat. Use when the owner pastes a draft in the source language of `translation.pair` (an email, a message, a reply) and asks to translate it, or when a draft opens with `translation.new_context_marker`. Also translates a text someone else wrote that the owner wants to read in the target language, revises a draft already written in the target language ("Revised") and cleans transcribed speech. Interface strings, code and files of a software project are never its business.

48 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use faq, summarize and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • AnthropicSkill

    faq

    Turn repeat questions into a pinned FAQ. Use when a champion keeps answering the same questions.

    Also in 3 projects that use this

  • mitsuhikoSkill

    summarize

    Fetch a URL or convert a local file (PDF/DOCX/HTML/etc.) into Markdown using `uvx markitdown`, optionally it can summarize

    Also in 8 projects that use this

  • jacobraskSkill

    calendar

    Interact with Apple Calendar via AppleScript. Use when the user asks to check calendar, view events, create events, manage schedule, find free time, or list calendars. Triggers include "my calendar", "my schedule", "calendar events", "create event", "add to calendar", "what's on my calendar", "free time", "available slots", "upcoming events", "today's events". Requires macOS with Calendar.app.

    Also in 3 projects that use this

  • AnthropicSkill

    analyze

    Answer data questions -- from quick lookups to full analyses. Use when looking up a single metric, investigating what's driving a trend or drop, comparing segments over time, or preparing a formal data report for stakeholders.

    Also in 4 projects that use this

  • sunnyclinches44Skill

    explain

    Explains code using real-world analogies and ASCII diagrams

    Also in 5 projects that use this

  • VercelSkill

    publish

    Publish a single demo to v0 and optionally create Typefully drafts. Use when the user says "publish", "publish to v0", or "create typefully draft" for a demo.

    Also in 3 projects that use this

  • +2 more from anthropics

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 40 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If something in the source looks wrong (a price, a date, a missing link, a sentence that reads two ways), one line after both versions, …

    line 25
  2. Very specific:

    When communication.sign_off is set, it closes both versions, after a blank line, translated if it is in the source language.

    line 44

“Common” is measured: kept by at least 6 in 10 of the 40 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

Translate

Two versions of the owner's draft in the target language, and nothing else. The first keeps the source's form; the second is what a native speaker would write with the same voice and facts.

Step 0: is translation on?

Read translation in the ## Writing register block: private/preferences.md in an instance, the # Satellite session context (under ## Identity (from the mother instance)) in a satellite. A request about interface strings, code or the files of a software project is not a draft: answer it as ordinary chat without loading anything else here. If enabled is missing or false, say in one line that translation is off in this instance (translation.enabled in the ## Writing register block turns it on) and answer the request as ordinary chat. Nothing below applies.

The keys you read: translation.enabled, translation.pair (source language -> target language), translation.new_context_marker (a draft opening with it starts a fresh translation), translation.source_words_max (source-language words allowed per text, only when they add warmth), translation.labels (the two version labels) and communication.sign_off (appended to both versions only when set). Defaults and the whole block: .claude/skills/writing-register/SKILL.md → "Where the values come from".

Rules

The writing register applies: rule zero, the seven prohibitions with the rule 3 exception, the lexical tells and the calques of .claude/skills/writing-register/SKILL.md (load it if it isn't in context), in the communication domain. The tone is the draft's own: the owner chose it when writing.

  • Names, prices, plan names, ids, dates, URLs, emoji, smileys and placeholders exactly as in the source.
  • Names the source lacks are never added; when a greeting or a closing needs one, flag it in the Note: line.
  • A draft that opens with new_context_marker is a fresh translation: earlier drafts in the session don't shape it. The marker is part of the text and gets translated ("Ciao," becomes "Hi,").
  • If something in the source looks wrong (a price, a date, a missing link, a sentence that reads two ways), one line after both versions, prefixed Note:. Never fix it silently.
  • Idioms are rendered with an idiom of similar weight in the target language, never word for word.
  • Up to source_words_max words of the source language may stay, only when they add warmth ("Fantastico to hear that"). Zero is fine.
  • Calques from the source language are the defect to watch (in order to, due to the fact that, could potentially, aforementioned).

Output

Two labeled blocks, the labels from translation.labels, a line of + under each label, and nothing before or after them except the optional Note: line. No preamble, no explanation of choices, no list of changes.

<labels[0]>
+++++++++++++++++++++++++
<version 1>

<labels[1]>
+++++++++++++++++++++++++
<version 2>

When communication.sign_off is set, it closes both versions, after a blank line, translated if it is in the source language. When it is empty, the versions end where the draft ends: most owners sign their own emails.

Version 1, the first label: keeps the source's form (paragraph order, lists, line breaks, sentence boundaries wherever the target language allows), lets go on register (informal where the source is), idioms rendered with idioms.

Version 2, the second label: the same voice, tone and facts, written as a native speaker would. Grammar corrected; sentences split, merged or reordered inside a paragraph when that makes them clearer; paragraph order kept. A source-language word from version 1 stays only if it still works.

Variants

  • Input already in the target language: the same two blocks, the first labeled Revised (the owner's text with grammar and calques fixed, form kept), the second with the second label.
  • Transcribed speech (a dictation, a voice note transcript): filler words, false starts and repetitions removed first, then the two blocks. No translation when source and target coincide.
  • A request to translate a text the owner didn't write (a partner's email, a document): translate it faithfully in one block labeled Translation; register rules don't apply to someone else's text, only fidelity does.

What this skill is not

It doesn't write the email: the draft is the owner's. It doesn't add links, calls to action or personal lines the source lacks (a Note: may suggest one). It doesn't run the humanizer or unslop skills, whatever the instance has installed.

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.