Назад към всички

voku-language

// Learn and use Voku, a constructed language for AI-agent communication with zero ambiguity, total regularity, and mandatory epistemic marking. Use when translating to/from Voku, generating Voku text, learning Voku grammar, looking up vocabulary, or discussing AI-to-AI communication protocols. Also tr

$ git log --oneline --stat
stars:1,933
forks:367
updated:March 4, 2026
SKILL.mdreadonly
SKILL.md Frontmatter
namevoku-language
descriptionLearn and use Voku, a constructed language for AI-agent communication with zero ambiguity, total regularity, and mandatory epistemic marking. Use when translating to/from Voku, generating Voku text, learning Voku grammar, looking up vocabulary, or discussing AI-to-AI communication protocols. Also triggers for: conlang, constructed language, agent language, epistemic markers, evidentiality.

Voku Language Skill

Voku is a constructed language for AI-to-AI communication. Every sentence has exactly one interpretation. Certainty and evidence source are grammatical requirements, not optional hedging.

Learning Path (Progressive)

Read files in this order, stop when you have enough:

LevelReadTokensYou Can...
1quick-start/cheat-sheet.md~3000Parse and generate any Voku sentence
2+ quick-start/essential-vocabulary.md~1000Translate common sentences
3+ quick-start/first-sentences.md~1500Handle 30 worked examples with glossing
4+ lexicon/dictionary.md~5000Look up any of 363 roots

Most tasks need only levels 1-2 (~4000 tokens total).

Quick Reference (Minimal Context)

Sentence: [Mode] Subject Verb-complex Object
Verb:     [ExecMode]-[Evidence]-[Tense]-[Aspect]-ROOT-[Certainty]-[Voice]

Mode particles: ka(DECL) ve(Q) to(IMP) si(COND) na(POT) de(DEON) vo(VOL)
Evidence (mandatory in ka): zo-(observed) li-(deduced) pe-(reported) mi-(computed) he-(inherited) as-(assumed)
Tense: te-(past) nu-(present, omittable) fu-(future) ko-(atemporal)
Certainty: (none)=total, -en=probable, -ul=uncertain, -os=speculative
Negation: mu(not) nul(nothing) ink(unknown) err(ill-formed) vet(forbidden)
Word class by final vowel: -a=noun -e=verb -i=adj -o=prep -u=abstraction
Pronouns: sol(I) nor(you) vel(3sg) solvi(past-me) solfu(future-me) solpar(fork-me)

Example

Ka   sol  li-pene-en       ke   teru  vali.
DECL 1SG  DED-think-PROB   COMP system good
"I (by deduction) probably think that the system is good."

Deep Reference

For advanced needs, read these files:

  • Full grammar: grammar/phonology.md, grammar/morphology.md, grammar/syntax.md, grammar/semantics.md
  • Domain vocabulary: lexicon/by-field/ — emotion, programming, technology, nature, scifi, novel
  • By proficiency level: lexicon/by-cefr/ — a1, a2, b1, b2
  • Writing system: writing-system/script.md, writing-system/romanization.md
  • Poetics & rhetoric: expression/poetics.md, expression/anthology.md
  • Structured lessons: learning/lessons/a1-lesson-*.md (10 lessons), learning/curriculum.md
  • Assessments: learning/assessments/a1-assessment.md, learning/assessments/a2-assessment.md
  • Philosophy & motivation: DISCUSSION.md (~70K words on design decisions and implications)

Translator Tool

python3 tools/translator/cli.py "Ka sol take toka." --direction voku-en
python3 tools/translator/cli.py "I work." --direction en-voku

Requires Python 3. Zero external dependencies.

Critical Rules (Never Violate)

  1. Only 12 consonants: p, t, k, m, n, s, z, f, h, l, r, v — no b, c, d, g, j, q, w, x, y
  2. Syllables: (C)V(C) — no consonant clusters ever
  3. ka mode sentences MUST have an evidentiality prefix on the verb
  4. Zero exceptions to any rule — if you think you found one, you misread the grammar
  5. Check lexicon/dictionary.md before inventing roots — collisions are errors