Canonical Knowledge Structure (CKS)

A universal, representation-independent foundation for knowledge.

CKS is an open specification that defines how knowledge can be represented, validated, exchanged, and evolved independently of programming languages, document formats, databases, or AI systems.

Rather than introducing yet another serialization format or programming language, CKS defines a canonical semantic layer shared by humans, software, and artificial intelligence.

Ecosystem

CKS Core is the semantic foundation of the CKS ecosystem. Other projects build upon it:

Why CKS?

Today the same knowledge exists simultaneously in many incompatible forms:

• documents
• databases
• JSON
• XML
• source code
• knowledge graphs
• ontologies
• AI prompts
• APIs

Each representation describes the same underlying knowledge differently.

CKS separates knowledge itself from every concrete representation.

Knowledge
      │
      ▼
Canonical Knowledge Structure (CKS)
      │
 ┌────┼───────────────┐
 ▼    ▼               ▼
JSON Python Database Natural Language

Representations may change.

Canonical knowledge remains the same.

Core Principles

CKS is founded on four simple principles.

Knowledge exists independently of its representation.

Knowledge is not JSON.

Knowledge is not a PDF.

Knowledge is not source code.

Representations are temporary.

Knowledge is not.

Structure preserves meaning.

Meaning is preserved by canonical structure rather than by syntax.

Representation preserves structure.

Different representations may express the same canonical structure.

Canonical operations belong to knowledge itself.

Validation.

Serialization.

Comparison.

Evolution.

Inspection.

These are operations on knowledge—not on files, databases, or programming languages.

Architecture

The CKS ecosystem consists of implementation-independent specifications.

Features

The current Python reference implementation provides:

• Immutable Canonical Knowledge Objects
• Canonical Relations
• Immutable Knowledge Structures
• Canonical JSON Serialization
• Deterministic Validation Pipeline
• Diagnostic System
• Reference Engine
• Canonical Public API
• Structural Comparison
• Projection
• Extraction
• Inspection
• Conformance Test Suite
• Command-Line Interface (validate, parse, inspect, evolve, schema, plugin)
• Structural Evolution (Genesis/Decay operators)
• Configurable Severity Thresholds
• HTML and Markdown Report Formatters
• Batch Validation (multiple files)
• JSON‑LD, Turtle, RDF/XML Import (via cks convert)
• JSON‑LD, Turtle, RDF/XML Export (via cks export)
• Strict Plugin Validation (--strict)
• Static Type Checking (mypy)
• Optional Extension Constraints (opt‑in validators for specialised knowledge types)
• Merkle‑tree based structural hashing for O(1) comparison and diff computation
• Three‑way merge (base‑branch‑branch) with conflict detection and structured error reporting
• In‑place object updates via UpdateObject (merge and replace modes)
• RenameObject operator — changes an object's identity.name without invalidating relations or cascading deletions
• Public operator properties — safe, documented introspection of structural operators (.obj, .object_id, .relation_id, .structure_patch, .mode, .new_name)
• k‑hop subgraph extraction (query_subgraph) with optional budget and type‑weighted ranking
• Partial three‑way merge with per‑identity conflict resolution (resolutions parameter)

Design Goals

CKS is designed to be:

• deterministic
• immutable
• observationally pure
• representation-independent
• implementation-independent
• language-independent
• suitable for formal verification

Current Repository

This repository contains the official Python Reference Implementation of the Canonical Knowledge Structure specifications.

Currently implemented:

• ✅ Canonical Knowledge Objects
• ✅ Canonical Relations
• ✅ Canonical Knowledge Structures
• ✅ Canonical Serialization
• ✅ Validation Pipeline
• ✅ Diagnostic System
• ✅ Reference Engine
• ✅ Canonical Public Interface
• ✅ Command-Line Interface
• ✅ Structural Evolution (CKS‑004)
• ✅ Reference Knowledge Corpus
• ✅ Conformance Test Suite (114 tests)
• ✅ PyPI Publication
• ✅ Import/Export Adapters (JSON‑LD, Turtle, RDF/XML)
• ✅ Modular CLI (commands refactored into separate handlers)
• ✅ Contract Documentation (docs/contracts.md)
• ✅ Static Type Checking (mypy)

Planned:

• Additional language implementations (Rust, TypeScript)

Installation

From PyPI:

pip install cks-core

Or from source:

git clone https://github.com/Deus-corp/cks-core.git
cd CKS
pip install -e .

Quick Example

from cks import (
    construct,
    validate,
    serialize,
)

from cks.core import (
    KnowledgeObject,
    ObjectIdentity,
)

obj = KnowledgeObject(
    identity=ObjectIdentity(
        id="obj-1",
        type="Definition",
        name="Knowledge",
    )
)

structure = construct([obj])

result = validate(structure)

print(result.is_valid)

print(serialize(structure))

Or use the command line:

# Validate a knowledge structure
cks validate examples/corpus/valid_theory_example.json

# Evolve a structure by adding an object
cks evolve examples/corpus/valid_theory_example.json examples/corpus/evolve_add.json

# Three-way merge of diverged structures
from cks import merge, MergeConflictError

base = construct([obj1, obj2])
branch_a = construct([obj1, obj3])
branch_b = construct([obj2, obj4])

try:
    merged = merge(base, branch_a, branch_b)
except MergeConflictError as e:
    for conflict in e.conflicts:
        print(f"Conflict on {conflict.object_id}")

# Partial three-way merge with conflict resolution
base = construct([obj1, obj2])
branch_a = construct([obj1, obj3])
branch_b = construct([obj2, obj4])

try:
    merged = merge(base, branch_a, branch_b, resolutions={"obj-1": "branch_a"})
except MergeConflictError as e:
    for conflict in e.conflicts:
        print(f"Unresolved conflict on {conflict.object_id}")

Or convert between formats:

# Convert JSON‑LD to CKS
cks convert examples/corpus/person.jsonld --format json-ld --output person.cks.json

# Export CKS to Turtle
cks export examples/corpus/valid_theory_example.json --format turtle --output theory.ttl

Testing

Run the complete conformance suite:

python -m pytest -v

Current status:

• 354+ tests
• all passing

The test suite verifies:

• deterministic behaviour
• immutability
• observational purity
• canonical serialization
• validation correctness
• public API conformance
• structural equivalence

Documentation

The complete specification is published separately.

Core specifications:

• CKS-000 — Foundations
• CKS-001 — Core Specification
• CKS-002 — Construction
• CKS-003 — Serialization
• CKS-004 — Evolution
• CKS-005 — Validator
• CKS-006 — Reference Engine
• CKS-007 — Canonical Knowledge Interface
• CKS-008 — Conformance

DOI:

Project Status

Current implementation status:

The current implementation serves as the reference implementation of the existing CKS specifications.

Future work focuses primarily on expanding the specification rather than redesigning the implemented components.

Vision

CKS aims to establish a universal semantic foundation for knowledge exchange between:

• humans
• software
• databases
• distributed systems
• artificial intelligence

through a single canonical representation of knowledge that is independent of every concrete implementation.

License

MIT