blun-king-cli 9.1.563 → 9.1.565

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "blun-king-cli",
-  "version": "9.1.563",
+  "version": "9.1.565",
   "description": "BLUN CLI - your own AI agent with a Telegram channel. Get it done. With BLUN.",
   "license": "MIT",
   "bin": {
@@ -42,10 +42,13 @@
     "agent-spine-plugin/assets/",
     "agent-spine-plugin/bin/",
     "agent-spine-plugin/blun.plugin.json",
+    "agent-spine-plugin/CONTRIBUTING.md",
     "agent-spine-plugin/hooks/",
     "agent-spine-plugin/LICENSE",
     "agent-spine-plugin/package.json",
+    "agent-spine-plugin/README.md",
     "agent-spine-plugin/scripts/check-hosts.js",
+    "agent-spine-plugin/SECURITY.md",
     "agent-spine-plugin/skill/",
     "agent-spine-plugin/skills/",
     "agent-spine-plugin/src/",

package/standard-skills/translate-native/LICENSE

@@ -1,21 +1,21 @@
-MIT License
-
-Copyright (c) 2026 Maykbiletti
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
+MIT License
+
+Copyright (c) 2026 Maykbiletti
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/standard-skills/translate-native/references/evaluation-protocol.md

@@ -1,95 +1,95 @@
-# Native Translation Evaluation Protocol
-
-## Release rule
-
-Release a translation only when it has zero blocking defects and reads naturally for the specified community, audience, register, and medium. Fluency cannot compensate for changed meaning. Fidelity cannot compensate for obvious translationese.
-
-## Classify defects
-
-| Severity | Definition | Examples | Action |
-| --- | --- | --- | --- |
-| Blocking | Meaning, safety, structure, or identity is damaged | Reversed negation, missing caveat, wrong number, broken placeholder, invented claim, wrong script | Fix before delivery |
-| Major | A native editor would rewrite it or the locale/register is materially wrong | Calque, unnatural syntax, grammatical-but-non-native collocation, broken parallelism, unnecessary borrowing, generic AI filler, wrong honorific, inconsistent terminology, regional mismatch | Fix before delivery |
-| Minor | Meaning and nativeness survive but polish is imperfect | Slight repetition, optional punctuation refinement | Fix when feasible |
-
-Do not average blocking defects into a numeric score. One blocking defect fails the translation.
-
-Grammar is not a release category. A target with zero grammar errors still fails when it contains a major nativeness defect. Use [translationese-review.md](translationese-review.md) for the target-only procedure and worked regression case.
-
-## Run a target-only native review
-
-Temporarily hide the source and judge the target as original writing. Inspect:
-
-- syntax and information order;
-- collocations and idiomatic word choice;
-- paragraph flow, cohesion, rhythm, and repetition;
-- politeness, honorifics, directness, and social relationship;
-- locale, script, punctuation, typography, and medium conventions;
-- whether a native writer would choose the same headline, button, instruction, or transition.
-
-Also inspect whether coordinated items belong to the same grammatical and semantic category, whether source-language technical vocabulary displaced an established native term, and whether polished-sounding AI filler hides a vague relationship.
-
-Rewrite any passage that is merely understandable but not naturally chosen. Preserve protected tokens while moving them to the grammatically natural position.
-
-## Run a source-aware fidelity review
-
-Compare propositions rather than matching words. Account for every source fact, negation, quantity, entity, relationship, condition, exception, uncertainty marker, time reference, and call to action. Flag:
-
-- omissions, additions, duplication, or unjustified explanation;
-- stronger or weaker certainty, obligation, praise, warning, or promise;
-- changed causal relationships, scope, chronology, or ambiguity;
-- terminology drift and inconsistent names;
-- cultural adaptation that changes factual content.
-
-Use back-translation only to reveal possible loss or addition. Never use it as proof of native quality.
-
-## Use independent review when stakes justify it
-
-When another agent or qualified reviewer is available, provide only:
-
-1. the complete source;
-2. the candidate target;
-3. target language, locale, script, audience, register, and medium;
-4. required glossary and protected tokens;
-5. this defect rubric.
-
-Ask for a structured defect list with severity, target excerpt, reason, and minimal correction direction. Do not ask the reviewer to replace the whole translation, reveal an expected answer, or praise the draft. Revise centrally so terminology and voice remain coherent.
-
-Use this release decision contract:
-
-```text
-status: PASS | FAIL
-blocking_defects:
-  - excerpt: exact target fragment
-    class: meaning | completeness | precision | integrity | orthography
-    reason: concise evidence
-major_defects:
-  - excerpt: exact target fragment
-    class: nativeness | locale | register | terminology | parallelism
-    reason: concise evidence
-```
-
-Return `PASS` only when both defect lists are empty. Any blocking or major defect requires revision followed by a complete new native, fidelity, integrity, and orthography review. Never average defects into a passing numeric score.
-
-For legal, medical, safety-critical, contractual, or public high-impact content, an AI reviewer is an additional check, not a substitute for a qualified native professional.
-
-## Route by confidence
-
-- **High:** Language variety and domain are well supported; all gates pass. Deliver normally.
-- **Medium:** One or more usage, locale, or domain choices are uncertain. Verify them with authoritative native sources and rerun the review.
-- **Low:** Reliable evidence is unavailable or varieties may be mixed. Do not present the result as assuredly native. State the limitation or request native review.
-
-Confidence describes evidence, not how fluent the prose appears.
-
-## Evaluate long documents
-
-Create a translation brief before drafting. Record locale, script, audience, address form, voice, tense policy, glossary, names, units, and formatting. Maintain it across sections. After assembly, review the whole document for:
-
-- terminology and capitalization drift;
-- pronoun/reference continuity;
-- headings, captions, lists, and cross-references;
-- consistent voice, politeness, tense, and narrative distance;
-- duplicated or missing content at section boundaries;
-- global rhythm and repeated sentence patterns caused by chunking.
-
-Do not declare a long document finished from section-level checks alone.
+# Native Translation Evaluation Protocol
+
+## Release rule
+
+Release a translation only when it has zero blocking defects and reads naturally for the specified community, audience, register, and medium. Fluency cannot compensate for changed meaning. Fidelity cannot compensate for obvious translationese.
+
+## Classify defects
+
+| Severity | Definition | Examples | Action |
+| --- | --- | --- | --- |
+| Blocking | Meaning, safety, structure, or identity is damaged | Reversed negation, missing caveat, wrong number, broken placeholder, invented claim, wrong script | Fix before delivery |
+| Major | A native editor would rewrite it or the locale/register is materially wrong | Calque, unnatural syntax, grammatical-but-non-native collocation, broken parallelism, unnecessary borrowing, generic AI filler, wrong honorific, inconsistent terminology, regional mismatch | Fix before delivery |
+| Minor | Meaning and nativeness survive but polish is imperfect | Slight repetition, optional punctuation refinement | Fix when feasible |
+
+Do not average blocking defects into a numeric score. One blocking defect fails the translation.
+
+Grammar is not a release category. A target with zero grammar errors still fails when it contains a major nativeness defect. Use [translationese-review.md](translationese-review.md) for the target-only procedure and worked regression case.
+
+## Run a target-only native review
+
+Temporarily hide the source and judge the target as original writing. Inspect:
+
+- syntax and information order;
+- collocations and idiomatic word choice;
+- paragraph flow, cohesion, rhythm, and repetition;
+- politeness, honorifics, directness, and social relationship;
+- locale, script, punctuation, typography, and medium conventions;
+- whether a native writer would choose the same headline, button, instruction, or transition.
+
+Also inspect whether coordinated items belong to the same grammatical and semantic category, whether source-language technical vocabulary displaced an established native term, and whether polished-sounding AI filler hides a vague relationship.
+
+Rewrite any passage that is merely understandable but not naturally chosen. Preserve protected tokens while moving them to the grammatically natural position.
+
+## Run a source-aware fidelity review
+
+Compare propositions rather than matching words. Account for every source fact, negation, quantity, entity, relationship, condition, exception, uncertainty marker, time reference, and call to action. Flag:
+
+- omissions, additions, duplication, or unjustified explanation;
+- stronger or weaker certainty, obligation, praise, warning, or promise;
+- changed causal relationships, scope, chronology, or ambiguity;
+- terminology drift and inconsistent names;
+- cultural adaptation that changes factual content.
+
+Use back-translation only to reveal possible loss or addition. Never use it as proof of native quality.
+
+## Use independent review when stakes justify it
+
+When another agent or qualified reviewer is available, provide only:
+
+1. the complete source;
+2. the candidate target;
+3. target language, locale, script, audience, register, and medium;
+4. required glossary and protected tokens;
+5. this defect rubric.
+
+Ask for a structured defect list with severity, target excerpt, reason, and minimal correction direction. Do not ask the reviewer to replace the whole translation, reveal an expected answer, or praise the draft. Revise centrally so terminology and voice remain coherent.
+
+Use this release decision contract:
+
+```text
+status: PASS | FAIL
+blocking_defects:
+  - excerpt: exact target fragment
+    class: meaning | completeness | precision | integrity | orthography
+    reason: concise evidence
+major_defects:
+  - excerpt: exact target fragment
+    class: nativeness | locale | register | terminology | parallelism
+    reason: concise evidence
+```
+
+Return `PASS` only when both defect lists are empty. Any blocking or major defect requires revision followed by a complete new native, fidelity, integrity, and orthography review. Never average defects into a passing numeric score.
+
+For legal, medical, safety-critical, contractual, or public high-impact content, an AI reviewer is an additional check, not a substitute for a qualified native professional.
+
+## Route by confidence
+
+- **High:** Language variety and domain are well supported; all gates pass. Deliver normally.
+- **Medium:** One or more usage, locale, or domain choices are uncertain. Verify them with authoritative native sources and rerun the review.
+- **Low:** Reliable evidence is unavailable or varieties may be mixed. Do not present the result as assuredly native. State the limitation or request native review.
+
+Confidence describes evidence, not how fluent the prose appears.
+
+## Evaluate long documents
+
+Create a translation brief before drafting. Record locale, script, audience, address form, voice, tense policy, glossary, names, units, and formatting. Maintain it across sections. After assembly, review the whole document for:
+
+- terminology and capitalization drift;
+- pronoun/reference continuity;
+- headings, captions, lists, and cross-references;
+- consistent voice, politeness, tense, and narrative distance;
+- duplicated or missing content at section boundaries;
+- global rhythm and repeated sentence patterns caused by chunking.
+
+Do not declare a long document finished from section-level checks alone.

package/standard-skills/translate-native/references/native-orthography.md

@@ -1,79 +1,79 @@
-# Native Orthography Gate
-
-## Release rule
-
-Write every target in its native orthography. Correct spelling, language-specific letters, diacritics, alphabet, script, punctuation, casing, spacing, and Unicode normalization are mandatory parts of translation quality, not optional typography.
-
-Do not rely on a separate skill to perform this review. Do not release a translation while a confirmed orthographic defect remains.
-
-## Mandatory review
-
-1. Identify the exact target language, locale, script, and regional standard.
-2. Review every natural-language segment, including headings, buttons, labels, metadata, accessibility text, and mixed-language passages.
-3. Replace ASCII transliterations only when they stand in for native spelling: `schön`, not `schoen`; `förstår`, not `forstar`; `español`, not `espanol`; `čeština`, not `cestina`.
-4. Use the native writing system by default. Do not romanize Greek, Cyrillic, Arabic, Hebrew, Indic, Southeast Asian, or East Asian text unless requested.
-5. Apply locale-specific punctuation and typography, such as Spanish `¿…?` and `¡…!`, French spacing conventions, or Chinese and Japanese punctuation.
-6. Preserve the canonical spelling of names, brands, quotations, and established technical terms. Verify uncertainty instead of inventing a mark.
-7. Normalize newly written UTF-8 content to Unicode NFC. Preserve a different normalization only when a technical format explicitly requires it.
-8. Re-run the full language and fidelity review after any orthographic correction that could change meaning.
-
-## Coverage map
-
-| Language or writing system | Preserve |
-| --- | --- |
-| German | `ä ö ü Ä Ö Ü ß`; use Swiss `ss` where required |
-| Swedish | `å ä ö Å Ä Ö` as separate letters |
-| Danish and Norwegian | `æ ø å Æ Ø Å` |
-| Icelandic and Faroese | accents and letters such as `ð þ æ ö ø` |
-| Spanish | `á é í ó ú ü ñ ¿ ¡` |
-| French | accents, diaeresis, cedilla, and conventional ligatures |
-| Portuguese | acute and circumflex accents, tildes, grave accents, and cedilla |
-| Catalan, Galician, and Basque | native accents, diaeresis, cedilla, middle dot, and `ñ` where applicable |
-| Czech and Slovak | every caron, acute, ring, circumflex, diaeresis, and language-specific consonant |
-| Polish, Hungarian, Romanian | complete native diacritics; Romanian comma-below `ș ț`, not cedilla variants |
-| Turkish and Azerbaijani | dotted and undotted `i`, `ç ğ ö ş ü`, and Azerbaijani `ə` |
-| South Slavic Latin scripts | `č ć đ š ž` and the exact requested regional standard |
-| Baltic languages | macrons, carons, dots, cedillas, and ogoneks |
-| Vietnamese | complete tone and vowel marks, including stacked marks |
-| Greek | Greek alphabet with correct tonos and dialytika; no Greeklish by default |
-| Cyrillic languages | the exact language alphabet, including letters absent from Russian |
-| Arabic-script languages | the correct language-specific letters and punctuation; optional vowel marks only when conventional |
-| Hebrew | Hebrew script; preserve or add niqqud only when the task requires it |
-| Indic scripts | native letters, vowel signs, conjuncts, nukta, virama, and punctuation |
-| Thai, Lao, Khmer, and Myanmar | tone marks, vowel placement, combining marks, and native spacing conventions |
-| Chinese | requested Simplified or Traditional characters and locale-appropriate punctuation |
-| Japanese | appropriate kanji, kana, prolonged-sound and iteration marks, spacing, and punctuation |
-| Korean | Hangul and correct Korean spacing; preserve hanja only when intended |
-| Armenian, Georgian, Ethiopic, and other scripts | native alphabet, canonical spelling, and punctuation |
-
-This map is illustrative, never exhaustive. Apply the same native-orthography rule to every human language and writing system.
-
-## Technical exceptions
-
-Keep exact ASCII or source spelling when modification could break behavior or fidelity:
-
-- code identifiers, commands, environment variables, API fields, JSON keys, database columns, slugs, URLs, email addresses, paths, hashes, tokens, regular expressions, and protocol values;
-- exact quotations, imported data, trademarks, and personal names whose canonical spelling is known;
-- systems with a documented encoding limitation;
-- explicit requests for transliteration, ASCII folding, search keys, or URL-safe text.
-
-Technical exceptions never justify ASCII-only prose around protected values.
-
-## Limits of automated checks
-
-The bundled translation guard verifies UTF-8, Unicode NFC, protected tokens, and structure. The bundled diacritics linter flags frequent ASCII substitutions in many Latin-script languages while ignoring common technical spans. Neither can prove that a word contains the correct diacritic, that the requested script was chosen, or that spelling is native. Use deterministic checks as evidence, then complete the language-aware review manually or with a qualified native reviewer.
-
-## Verification hierarchy
-
-When usage, spelling, script, or locale conventions are uncertain, verify in this order:
-
-1. the requested language community's academy, language council, government style guide, or other recognized normative body;
-2. the [Unicode Standard](https://unicode.org/reports/tr15/) for normalization and character behavior;
-3. [Unicode CLDR](https://cldr.unicode.org/) for locale identifiers, exemplar characters, punctuation sets, plural rules, number and date conventions, and script variants;
-4. the [W3C Language Enablement Index](https://www.w3.org/International/typography/gap-analysis/language-matrix.html) for writing-system layout and typography;
-5. the [IANA Language Subtag Registry](https://www.iana.org/assignments/language-subtag-registry) for language, region, and script tags;
-6. contemporary native-edited dictionaries and monolingual corpora for usage and collocations.
-
-Examples of authoritative language sources include Sweden's [Institutet för språk och folkminnen](https://www.isof.se/svenska-spraket/frageladan), the [Real Academia Española](https://www.rae.es/), the Czech Academy's [Czech Language Institute](https://ujc.cas.cz/), the [Institut d'Estudis Catalans](https://www.iec.cat/), and [Euskaltzaindia](https://www.euskaltzaindia.eus/).
-
-Do not treat a translation engine, parallel corpus, search-result count, ASCII spell-checker, or another model's confidence as a normative source. Use them only to locate questions that still require native evidence.
+# Native Orthography Gate
+
+## Release rule
+
+Write every target in its native orthography. Correct spelling, language-specific letters, diacritics, alphabet, script, punctuation, casing, spacing, and Unicode normalization are mandatory parts of translation quality, not optional typography.
+
+Do not rely on a separate skill to perform this review. Do not release a translation while a confirmed orthographic defect remains.
+
+## Mandatory review
+
+1. Identify the exact target language, locale, script, and regional standard.
+2. Review every natural-language segment, including headings, buttons, labels, metadata, accessibility text, and mixed-language passages.
+3. Replace ASCII transliterations only when they stand in for native spelling: `schön`, not `schoen`; `förstår`, not `forstar`; `español`, not `espanol`; `čeština`, not `cestina`.
+4. Use the native writing system by default. Do not romanize Greek, Cyrillic, Arabic, Hebrew, Indic, Southeast Asian, or East Asian text unless requested.
+5. Apply locale-specific punctuation and typography, such as Spanish `¿…?` and `¡…!`, French spacing conventions, or Chinese and Japanese punctuation.
+6. Preserve the canonical spelling of names, brands, quotations, and established technical terms. Verify uncertainty instead of inventing a mark.
+7. Normalize newly written UTF-8 content to Unicode NFC. Preserve a different normalization only when a technical format explicitly requires it.
+8. Re-run the full language and fidelity review after any orthographic correction that could change meaning.
+
+## Coverage map
+
+| Language or writing system | Preserve |
+| --- | --- |
+| German | `ä ö ü Ä Ö Ü ß`; use Swiss `ss` where required |
+| Swedish | `å ä ö Å Ä Ö` as separate letters |
+| Danish and Norwegian | `æ ø å Æ Ø Å` |
+| Icelandic and Faroese | accents and letters such as `ð þ æ ö ø` |
+| Spanish | `á é í ó ú ü ñ ¿ ¡` |
+| French | accents, diaeresis, cedilla, and conventional ligatures |
+| Portuguese | acute and circumflex accents, tildes, grave accents, and cedilla |
+| Catalan, Galician, and Basque | native accents, diaeresis, cedilla, middle dot, and `ñ` where applicable |
+| Czech and Slovak | every caron, acute, ring, circumflex, diaeresis, and language-specific consonant |
+| Polish, Hungarian, Romanian | complete native diacritics; Romanian comma-below `ș ț`, not cedilla variants |
+| Turkish and Azerbaijani | dotted and undotted `i`, `ç ğ ö ş ü`, and Azerbaijani `ə` |
+| South Slavic Latin scripts | `č ć đ š ž` and the exact requested regional standard |
+| Baltic languages | macrons, carons, dots, cedillas, and ogoneks |
+| Vietnamese | complete tone and vowel marks, including stacked marks |
+| Greek | Greek alphabet with correct tonos and dialytika; no Greeklish by default |
+| Cyrillic languages | the exact language alphabet, including letters absent from Russian |
+| Arabic-script languages | the correct language-specific letters and punctuation; optional vowel marks only when conventional |
+| Hebrew | Hebrew script; preserve or add niqqud only when the task requires it |
+| Indic scripts | native letters, vowel signs, conjuncts, nukta, virama, and punctuation |
+| Thai, Lao, Khmer, and Myanmar | tone marks, vowel placement, combining marks, and native spacing conventions |
+| Chinese | requested Simplified or Traditional characters and locale-appropriate punctuation |
+| Japanese | appropriate kanji, kana, prolonged-sound and iteration marks, spacing, and punctuation |
+| Korean | Hangul and correct Korean spacing; preserve hanja only when intended |
+| Armenian, Georgian, Ethiopic, and other scripts | native alphabet, canonical spelling, and punctuation |
+
+This map is illustrative, never exhaustive. Apply the same native-orthography rule to every human language and writing system.
+
+## Technical exceptions
+
+Keep exact ASCII or source spelling when modification could break behavior or fidelity:
+
+- code identifiers, commands, environment variables, API fields, JSON keys, database columns, slugs, URLs, email addresses, paths, hashes, tokens, regular expressions, and protocol values;
+- exact quotations, imported data, trademarks, and personal names whose canonical spelling is known;
+- systems with a documented encoding limitation;
+- explicit requests for transliteration, ASCII folding, search keys, or URL-safe text.
+
+Technical exceptions never justify ASCII-only prose around protected values.
+
+## Limits of automated checks
+
+The bundled translation guard verifies UTF-8, Unicode NFC, protected tokens, and structure. The bundled diacritics linter flags frequent ASCII substitutions in many Latin-script languages while ignoring common technical spans. Neither can prove that a word contains the correct diacritic, that the requested script was chosen, or that spelling is native. Use deterministic checks as evidence, then complete the language-aware review manually or with a qualified native reviewer.
+
+## Verification hierarchy
+
+When usage, spelling, script, or locale conventions are uncertain, verify in this order:
+
+1. the requested language community's academy, language council, government style guide, or other recognized normative body;
+2. the [Unicode Standard](https://unicode.org/reports/tr15/) for normalization and character behavior;
+3. [Unicode CLDR](https://cldr.unicode.org/) for locale identifiers, exemplar characters, punctuation sets, plural rules, number and date conventions, and script variants;
+4. the [W3C Language Enablement Index](https://www.w3.org/International/typography/gap-analysis/language-matrix.html) for writing-system layout and typography;
+5. the [IANA Language Subtag Registry](https://www.iana.org/assignments/language-subtag-registry) for language, region, and script tags;
+6. contemporary native-edited dictionaries and monolingual corpora for usage and collocations.
+
+Examples of authoritative language sources include Sweden's [Institutet för språk och folkminnen](https://www.isof.se/svenska-spraket/frageladan), the [Real Academia Española](https://www.rae.es/), the Czech Academy's [Czech Language Institute](https://ujc.cas.cz/), the [Institut d'Estudis Catalans](https://www.iec.cat/), and [Euskaltzaindia](https://www.euskaltzaindia.eus/).
+
+Do not treat a translation engine, parallel corpus, search-result count, ASCII spell-checker, or another model's confidence as a normative source. Use them only to locate questions that still require native evidence.

package/standard-skills/translate-native/references/native-translation-standard.md

@@ -1,94 +1,94 @@
-# Native Translation Standard
-
-## The core distinction
-
-A correct translation preserves the source meaning. A native translation also follows the target language's own habits of thought and expression. Both are required.
-
-Do not optimize for visible word correspondence. Optimize for semantic equivalence plus native acceptability. A sentence may need to be split, merged, reordered, made implicit, or made explicit where the target grammar conventionally requires it, provided no source claim changes.
-
-## Choose the operating mode
-
-| Mode | Primary goal | Permitted adaptation |
-| --- | --- | --- |
-| Faithful native translation | Preserve meaning in fully native language | Syntax, idiom, information order, punctuation |
-| Localization | Make content work in a locale or product | Formats, units, conventions, locally expected labels |
-| Transcreation | Preserve effect in persuasive or creative work | Imagery, wordplay, slogans, cultural framing—with claims intact |
-| Proofreading | Repair target-language text | Grammar, vocabulary, nativeness, consistency |
-
-Use faithful native translation by default. Do not silently turn translation into copywriting.
-
-## Specify language beyond its name
-
-Represent the target as a bundle:
-
-- language and BCP 47 tag where known;
-- region or speech community;
-- script and orthography;
-- formal, informal, honorific, technical, literary, or conversational register;
-- audience age, expertise, and relationship to the writer;
-- medium and space constraints.
-
-Language labels can hide materially different choices. “Chinese” requires a script and often a region; “Arabic” can mean Modern Standard Arabic or a spoken variety; “Spanish,” “Portuguese,” “French,” “English,” “Catalan,” “Serbian,” and many others have regional standards. Prefer the user's explicit identity and terminology over assumptions.
-
-## Preserve semantic force
-
-Check especially:
-
-- scope of negation and quantifiers;
-- can/may/must/should and their degree of obligation;
-- completed, ongoing, habitual, and hypothetical actions;
-- evidentiality, confidence, hearsay, and uncertainty;
-- inclusive/exclusive pronouns and levels of respect;
-- gender, number, animacy, classifiers, and noun classes where relevant;
-- relationships between headings, labels, buttons, and surrounding instructions.
-
-Naturalness never licenses stronger claims, extra benefits, removed caveats, or invented politeness.
-
-## Respect the language community
-
-- Use the script and orthography used by the requested community, including diacritics, combining marks, contextual forms, native punctuation, spacing, and directionality.
-- Do not treat Latin transliteration as the default for a non-Latin script.
-- Avoid exoticizing minority and Indigenous languages or replacing community terminology with a majority-language label.
-- When multiple standards exist, name the selected standard if that choice matters.
-- Preserve names according to the person's or organization's established form. Transliterate only according to the target convention or explicit request.
-
-## Source hierarchy for verification
-
-Use evidence in roughly this order, adjusting for the language and domain:
-
-1. Native-authored or professionally native-edited material in the same domain and locale.
-2. Language academies, community authorities, government terminology banks, universities, and respected dictionaries or style guides.
-3. Unicode CLDR and BCP 47 resources for locale identifiers, formats, exemplar characters, plurals, and writing-system conventions.
-4. Curated monolingual corpora such as the Leipzig Corpora Collection for collocations and contemporary usage.
-5. Curated evaluation sets such as FLORES+ for cross-language comparison.
-6. Parallel corpora such as OPUS as supporting evidence only.
-7. Uncurated web text or anonymous dataset entries only as weak corroboration.
-
-A large dataset is not automatically authoritative. It may contain machine translations, subtitles, duplicated text, wrong language tags, domain bias, outdated orthography, or misaligned sentences. Check its dataset card, provenance, license, curation method, date, locale, and intended use.
-
-Useful starting points:
-
-- FLORES+: <https://huggingface.co/datasets/openlanguagedata/flores_plus>
-- OPUS: <https://opus.nlpl.eu/>
-- Leipzig Corpora Collection: <https://cls.corpora.uni-leipzig.de/>
-- Unicode CLDR: <https://cldr.unicode.org/>
-
-Do not download large corpora for a routine translation. Consult only when the wording or language variety needs verification.
-
-## Detect translationese
-
-Revise when any symptom appears:
-
-- target sentences mirror source length and clause order without linguistic reason;
-- dictionary synonyms replace established collocations;
-- pronouns are repeated or omitted according to the source rather than target norms;
-- politeness, articles, tense, classifiers, particles, or discourse markers feel imported;
-- headlines and buttons use sentence grammar instead of the target medium's conventions;
-- an idiom is understandable but no native writer would choose it;
-- every sentence is grammatical yet the paragraph does not flow natively.
-
-Use a back-translation only as a diagnostic for missing meaning, never as the final naturalness test.
-
-## Handle uncertainty honestly
-
-For low-resource languages, fluent-looking output can still be wrong. If evidence is thin, separate what is known from what is inferred, avoid mixing neighboring varieties, and request community or professional review for consequential use. Do not call a translation native merely because it is grammatical.
+# Native Translation Standard
+
+## The core distinction
+
+A correct translation preserves the source meaning. A native translation also follows the target language's own habits of thought and expression. Both are required.
+
+Do not optimize for visible word correspondence. Optimize for semantic equivalence plus native acceptability. A sentence may need to be split, merged, reordered, made implicit, or made explicit where the target grammar conventionally requires it, provided no source claim changes.
+
+## Choose the operating mode
+
+| Mode | Primary goal | Permitted adaptation |
+| --- | --- | --- |
+| Faithful native translation | Preserve meaning in fully native language | Syntax, idiom, information order, punctuation |
+| Localization | Make content work in a locale or product | Formats, units, conventions, locally expected labels |
+| Transcreation | Preserve effect in persuasive or creative work | Imagery, wordplay, slogans, cultural framing—with claims intact |
+| Proofreading | Repair target-language text | Grammar, vocabulary, nativeness, consistency |
+
+Use faithful native translation by default. Do not silently turn translation into copywriting.
+
+## Specify language beyond its name
+
+Represent the target as a bundle:
+
+- language and BCP 47 tag where known;
+- region or speech community;
+- script and orthography;
+- formal, informal, honorific, technical, literary, or conversational register;
+- audience age, expertise, and relationship to the writer;
+- medium and space constraints.
+
+Language labels can hide materially different choices. “Chinese” requires a script and often a region; “Arabic” can mean Modern Standard Arabic or a spoken variety; “Spanish,” “Portuguese,” “French,” “English,” “Catalan,” “Serbian,” and many others have regional standards. Prefer the user's explicit identity and terminology over assumptions.
+
+## Preserve semantic force
+
+Check especially:
+
+- scope of negation and quantifiers;
+- can/may/must/should and their degree of obligation;
+- completed, ongoing, habitual, and hypothetical actions;
+- evidentiality, confidence, hearsay, and uncertainty;
+- inclusive/exclusive pronouns and levels of respect;
+- gender, number, animacy, classifiers, and noun classes where relevant;
+- relationships between headings, labels, buttons, and surrounding instructions.
+
+Naturalness never licenses stronger claims, extra benefits, removed caveats, or invented politeness.
+
+## Respect the language community
+
+- Use the script and orthography used by the requested community, including diacritics, combining marks, contextual forms, native punctuation, spacing, and directionality.
+- Do not treat Latin transliteration as the default for a non-Latin script.
+- Avoid exoticizing minority and Indigenous languages or replacing community terminology with a majority-language label.
+- When multiple standards exist, name the selected standard if that choice matters.
+- Preserve names according to the person's or organization's established form. Transliterate only according to the target convention or explicit request.
+
+## Source hierarchy for verification
+
+Use evidence in roughly this order, adjusting for the language and domain:
+
+1. Native-authored or professionally native-edited material in the same domain and locale.
+2. Language academies, community authorities, government terminology banks, universities, and respected dictionaries or style guides.
+3. Unicode CLDR and BCP 47 resources for locale identifiers, formats, exemplar characters, plurals, and writing-system conventions.
+4. Curated monolingual corpora such as the Leipzig Corpora Collection for collocations and contemporary usage.
+5. Curated evaluation sets such as FLORES+ for cross-language comparison.
+6. Parallel corpora such as OPUS as supporting evidence only.
+7. Uncurated web text or anonymous dataset entries only as weak corroboration.
+
+A large dataset is not automatically authoritative. It may contain machine translations, subtitles, duplicated text, wrong language tags, domain bias, outdated orthography, or misaligned sentences. Check its dataset card, provenance, license, curation method, date, locale, and intended use.
+
+Useful starting points:
+
+- FLORES+: <https://huggingface.co/datasets/openlanguagedata/flores_plus>
+- OPUS: <https://opus.nlpl.eu/>
+- Leipzig Corpora Collection: <https://cls.corpora.uni-leipzig.de/>
+- Unicode CLDR: <https://cldr.unicode.org/>
+
+Do not download large corpora for a routine translation. Consult only when the wording or language variety needs verification.
+
+## Detect translationese
+
+Revise when any symptom appears:
+
+- target sentences mirror source length and clause order without linguistic reason;
+- dictionary synonyms replace established collocations;
+- pronouns are repeated or omitted according to the source rather than target norms;
+- politeness, articles, tense, classifiers, particles, or discourse markers feel imported;
+- headlines and buttons use sentence grammar instead of the target medium's conventions;
+- an idiom is understandable but no native writer would choose it;
+- every sentence is grammatical yet the paragraph does not flow natively.
+
+Use a back-translation only as a diagnostic for missing meaning, never as the final naturalness test.
+
+## Handle uncertainty honestly
+
+For low-resource languages, fluent-looking output can still be wrong. If evidence is thin, separate what is known from what is inferred, avoid mixing neighboring varieties, and request community or professional review for consequential use. Do not call a translation native merely because it is grammatical.