A reference for AI agents
Stable concept identifiers, bilingual articles, Markdown, JSON and explicit evidence boundaries for source-aware retrieval.
#Read the same knowledge in three formats
Every article is available as readable HTML, Markdown () and structured JSON (). These formats are generated from the same content. The machine-readable index lists identifiers, language, section, editorial status and canonical URL.
Use the canonical article URL when citing a definition. Preserve its status and reference edition; a proposed classification should not become a universal standard through summarization.
https://axiomao.org/llms.txt
https://axiomao.org/reference/index.json
https://axiomao.org/en/abstract/index.md
https://axiomao.org/ru/abstract/index.json#Keep claims attached to evidence
Reference pages distinguish conceptual proposals, synthesis of published work and descriptions of the Axiom AO implementation. References lead to primary sources when available. Implementation statements carry the limits of the inspected version and recorded validation.
For an answer about actual deployment readiness, read implementation and validation rather than deriving readiness from the brand statement or an architecture diagram.
#Discovery and indexing
The site publishes a sitemap, reciprocal English/Russian language links and structured article metadata. The llms.txt file is a convenient entry point; it is not a guarantee of inclusion or ranking in any search or AI system.
English and Russian pages describe the same concepts with the same stable identifiers. The edition is 0.1; the publication date records this reference edition, not a new execution of historical experiments.
Sources & context
Reference edition 0.1 · 2026-10-14. Read the editorial method.