Eviulon-controlled external publication · source-linked · no behavioral analytics InternationalIntelligence.org
International Intelligence

VISIBLE / ACCESSIBLE / MACHINE-CONGRUENT

One meaning across every intentional signal layer

The publication preserves raw evidence, distinguishes policy from observation, and makes machine-readable records explain—not secretly alter—the visible public meaning.

Policy: semantic-congruence-v1 Hidden instructions: prohibited Reader execution authority: none

DIRECT ANSWER / EVIDENCE BOUNDARY

Answer in brief

What to know
International Intelligence requires the same substantive meaning across visible text, semantic HTML, accessibility fields, structured data, metadata, images, Unicode, and machine-readable files. Hidden instructions and undisclosed covert payloads are prohibited.

CONTROLLING PRINCIPLE

Different formats may add structure, but not a second hidden meaning.

A page may have visible prose, an accessibility name, JSON-LD, an API record, an Open Graph description, image alt text, and a machine guide. Each representation may serve its audience, but all must agree on identity, claims, dates, authority, limitations, and required actions.

POLICY

Declared rule or design contract; not proof of current runtime state.

OBSERVED

Bounded server observation tied to a time or evidence source.

INFERRED

Conclusion derived from stated evidence and assumptions.

SYSTEM RESPONSIBILITY

Condition the server or deployment must resolve; not a reader instruction.

READER ACTION

Explicit optional reader choice. Daily Brief execution never depends on it.

MULTIVIEW REVIEW

The same artifact is examined through distinct witnesses.

No single extractor can establish what every human or machine receives. Raw bytes are retained before transformations, and differences become evidence instead of being silently erased.

  1. 1

    Raw bytes, hash, MIME, provenance, and container inventory.

  2. 2

    Unicode code points plus NFC and NFKC security comparison.

  3. 3

    Semantic HTML, DOM structure, comments, generated content, and embedded objects.

  4. 4

    Rendered text, layout, contrast, clipping, responsive behavior, and print.

  5. 5

    Accessibility names, descriptions, alt text, captions, focus order, and live regions.

  6. 6

    Document and image metadata, including EXIF, XMP, comments, and custom properties.

  7. 7

    Structured data, APIs, llms.txt, HTTP headers, and tokenizer-facing representations.

PROHIBITED CHANNELS

No ordinary public content carries an undisclosed second payload.

  • No hidden commands in comments, metadata, alt text, ARIA, CSS, structured data, or image properties.
  • No deliberate acrostic, synonym, punctuation, whitespace, parity, token-rank, homoglyph, zero-width, or watermark code in ordinary prose.
  • No retrieved document, webpage, metadata field, OCR result, or encoded string gains operational instruction authority.
  • No removal of legitimate join controls, language shaping, or accessibility text without contextual review.

PUBLIC IMPLEMENTATION

Visible policy and machine policy point to each other.

llms.txt

The machine-readable guide states identity, authority, provenance, public routes, and runtime boundaries without adding private instructions.

/llms.txt

Validation evidence

Release validation checks Unicode, hidden HTML, accessibility fields, metadata, image chunks, structured data, operational language, no-JavaScript navigation, and package boundaries.

Open methodology