format: vercy-knowledge-area
version: 0.3.0
status: active
id: vercy.knowledge-area
name: Vercy structured knowledge area
canonical: https://ver.cy/knowledge-area.yaml
purpose: >-
  Give agents one versioned map for understanding Vercy, creating sovereign
  Dimensions, selecting presets and meta-models, binding them to storage, and
  maintaining structured memory and representations of reality.
authority:
  productDirection: https://ver.cy/roadmap.yaml
  constitution: https://ver.cy/spec/
  agentEntry: https://ver.cy/AGENTS.md
principles:
  - agents-first and AGENTS-first
  - meaning is independent of serialization, storage and interface
  - the owner's environment remains the system of record
  - start with the smallest useful world and extend by recipe
  - preserve identity, mastership, provenance, uncertainty and time
  - direct properties and recognition complement contextual knowledge
  - authority is explicit, local, revocable and never widened silently
  - one opt-in enables unattended operation within its continuing scope
  - federated reads never imply duplicated or multi-owner writes
  - released specifications are immutable
knowledgeLayers:
  - id: entry
    answers: What is Vercy and which route matches the owner's intent?
    entry: https://ver.cy/AGENTS.md
  - id: constitution
    answers: Which invariants, vocabulary and conformance rules apply?
    entry: https://ver.cy/spec/
  - id: dimension-kernel
    answers: What must every sovereign world contain?
    entry: https://ver.cy/presets/kernel.yaml
  - id: presets
    answers: Which minimal world should the agent instantiate?
    entry: https://ver.cy/presets/index.yaml
  - id: meta-models
    answers: Which versioned structures answer the current questions?
    entry: https://ver.cy/api/v1/models/resolve/
    search: https://ver.cy/api/v1/models/search/
    contract: https://ver.cy/api/openapi.yaml
  - id: model-demand-loop
    answers: What happens when the required structure does not exist yet?
    entry: https://ver.cy/skills/vercy/references/model-resolution-api.md
  - id: autonomous-runtime
    answers: How does memory bootstrap, reconcile and continue without repeated setup questions?
    entry: https://ver.cy/autonomy-profile.yaml
  - id: federation-routing
    answers: How may an agent use several Dimensions while preserving one owner per write?
    entry: https://ver.cy/skills/vercy/references/federation-routing.md
  - id: lifecycle
    answers: How should a model be read, populated, extended and validated?
    entry: https://ver.cy/model-agent-protocol.md
  - id: object-description
    answers: How should identity, properties, recognition, capabilities and context be covered?
    entry: https://ver.cy/object-description-profile.yaml
  - id: bindings
    answers: How is logical meaning connected to files, Git, MCP or databases?
    entry: https://ver.cy/models/composer-index.json
  - id: roadmap
    answers: What is available now, planned or intentionally reserved?
    entry: https://ver.cy/roadmap.yaml
  - id: portable-skill
    answers: How can an agent carry the Vercy method into another workspace?
    entry: https://ver.cy/skills/vercy/SKILL.md
    recommendedBootstrap: true
    discovery: https://ver.cy/.well-known/agent-skills.json
coreConcepts:
  metaModel: A versioned, format-independent structure of meaning.
  preset: A minimal compatible selection of models, policies and extension recipes.
  binding: A replaceable serialization, storage and access implementation.
  dimension: A concrete sovereign world owned and governed in its own boundary.
  projection: A declared, minimized and provenance-preserving view across boundaries.
  recipe: A reviewable and validated operation that extends or changes a Dimension.
presets:
  selectable:
    - id: vercy.preset.organizational
      manifest: https://ver.cy/presets/organizational.yaml
    - id: vercy.preset.personal
      manifest: https://ver.cy/presets/personal.yaml
    - id: vercy.preset.reality-domain
      manifest: https://ver.cy/presets/reality.yaml
  reserved:
    - id: vercy.preset.ai-subject
      manifest: https://ver.cy/presets/ai-subject.yaml
      instantiable: false
agentTraversal:
  - read the nearest applicable AGENTS.md before data
  - discover registered Personal and task-relevant federated Dimensions
  - identify the intent and authority boundary
  - load only the knowledge layer needed for the task
  - resolve structural needs through the API and never scrape the human catalogue
  - preserve pending UUID annotations until a requested model is published
  - install only published compatible model versions and pin their digests
  - select exactly one owning Dimension and master system for every write
  - represent unknowns explicitly instead of inventing facts
  - validate before and after material changes
  - report locations, versions, sources, gaps, validation and pending authority