# =============================================================================
# COMPETENCY RECORD - Cedric Ancellin's demonstrated engineering skills and
# experience. Third-party technologies and vendors are named as the concrete
# skill surface. Professional background / track record.
# =============================================================================

# ============================================================
# KNOWLEDGE ARCHITECTURE & CURATION - COMPETENCY RECORD
# ============================================================
# Core Purpose:
#   Demonstrated experience designing a single, repeatable
#   knowledge-architecture methodology - machine-readable,
#   AI-first structured knowledge bases with an explicit rules
#   layer, audience-scoped disclosure, and a standard serving
#   protocol - and proving it, unchanged in essentials, across
#   radically different domains: software, long-form fiction,
#   and professional/business knowledge.
#
#   Described in abstract archetypes - a transferable methodology,
#   not any single implementation.
# ============================================================

summary: >
  Designed a domain-agnostic knowledge-architecture methodology -
  a structured, machine-readable knowledge base authored for AI
  agents first, organized as one walkable tree, cross-linked by
  formal pointers, governed by an explicit rules layer, and served
  to AI over a standard protocol (MCP). Personally devised from
  first principles and hardened across several unrelated production
  systems, then generalized into transferable IP with conventions,
  serving tiers, disclosure governance, and an adoption playbook.
  This competency record is itself authored in that discipline.


# ==================================================================
# 1. THE METHODOLOGY
# ==================================================================

methodology:
  thesis: >
    Traditional documentation is written for humans to skim and
    rots the moment it leaves an author's head. This inverts that:
    knowledge authored for machines first, so any AI session can
    start cold and build complete, correct context in minutes -
    then act on it safely, because the rules and rationale travel
    with the facts.
  principles:
    - "Structured over narrative, keyed YAML, not prose."
    - "One walkable tree, any tool can traverse it and assemble a complete picture; no hidden index."
    - "Formal cross-references ($ref pointers) linking concepts across files."
    - "An explicit rules layer, 'laws' and conventions carried alongside the facts they govern."
    - "Incremental and self-similar, a file works whether the base has ten files or ten thousand."
    - "A founding law: 'if it is not in the knowledge base, it does not exist.'"
  serving: "Exposed to AI assistants over the Model Context Protocol (see COMP-11), so the same base drives coding agents, review, and Q&A without bespoke integrations."


# ==================================================================
# 2. DOMAIN-AGNOSTIC PROOF
# ==================================================================

domain_agnostic:
  claim: >
    The same methodology, same spine, same conventions, same
    serving model, has been applied, essentially unchanged, to
    domains with nothing in common. That breadth is the evidence it
    is a general discipline, not a one-off convention.
  archetypes_covered:
    - "A large multi-repository software platform."
    - "A single AI-built web application."
    - "A multi-service consumer product (a backend plus multiple frontends)."
    - "A multi-tenant documentation host serving several unrelated projects."
    - "A work of long-form creative fiction, a world bible with characters, factions, locations, a timeline, a constructed language (phonotactics + lexicon), enforced prose style, and reader-scope/spoiler governance."
    - "A professional/advisory knowledge base representing a person and their career (this résumé is itself an instance)."
  what_stays_fixed: "The structure, the cross-reference discipline, the rules layer, the serving protocol, and the 'machine-first' authoring stance."
  what_varies: "The domain vocabulary and the file taxonomy, the nouns change; the architecture does not."


# ==================================================================
# 3. AUTHORING RIGOR & GOVERNANCE
# ==================================================================

authoring_governance:
  description: >
    Treats the knowledge base as engineered infrastructure, not
    freeform notes, with the same quality gates as code.
  practices:
    - "A formal schema defining required fields and types per file, so structure is explicit and onboardable."
    - "Machine validation (a validator plus git pre-commit hooks) that rejects malformed or drifting entries before they land."
    - "A standardized file taxonomy and a written authoring guide applied consistently across every instance."
    - "A reusable day-0 governance skeleton (overview, manifest/registry, guide, architecture, laws) that every new knowledge base bootstraps from."
    - "Depth tiers (skeleton → vocabulary → flows → complete) so a base is useful early and deepens incrementally."


# ==================================================================
# 4. AUDIENCE-SCOPED DISCLOSURE & INFORMATION FIREWALLING
# ==================================================================

disclosure_control:
  description: >
    A first-class capability: the same base can serve different
    audiences with different truths, and can be bound by an explicit
    disclosure firewall, what may and may not be said, that
    overrides everything else.
  mechanisms:
    - "A binding firewall document declaring present / exclude / never boundaries, read first on every session and enforced ahead of any other instruction."
    - "A hard anti-fabrication rule: in a forbidden domain the correct, complete answer is 'I don't have that information' plus a redirect, never infer, speculate, or invent."
    - "Deliberate, by-design gaps: forbidden material is simply absent rather than redacted in place."
    - "Audience scopes at the serving layer (e.g. a public/consult view vs. a full/contract view) selected per client, so one base safely serves multiple trust levels."
  significance: "Makes structured knowledge safe to expose externally, the difference between an internal wiki and a base you can put in front of a client, a partner, or the public."


# ==================================================================
# 5. SERVING-TIER DESIGN
# ==================================================================

serving_tiers:
  description: >
    Designed a catalog of serving models so the same knowledge base
    can be delivered at the right level of capability and lock-in.
  spectrum: "From file-only static serving (T0) through to a data-connected tier (T4) that can answer against live system state, with an explicit guide for choosing a tier by need."
  low_lock_in: "The substrate is plain YAML in version control, so the adopting party owns the asset outright, the value is the methodology and the judgment to adapt it, not a proprietary tool."


# ==================================================================
# 6. METHODOLOGY AS TRANSFERABLE IP
# ==================================================================

transferable_ip:
  description: >
    Generalized the practice into something an organization can
    adopt, not just an artifact, but a repeatable capability.
  components:
    - "An adoption playbook: bootstrap steps, a maturity model, operating rituals, and adoption anti-patterns."
    - "An enterprise-readiness case: governance, security, scale, AI-operability, and durability, with an honest comparison to alternatives."
    - "A canonical lexicon so the vocabulary is shared and teachable."
  provenance: "Designed and developed from first principles across many stacks and problem domains, each build refined the whole; every element was proven in real, tested production usage before being generalized."


cross_reference:
  ai_knowledge_engineering:
    $ref: "https://zc8.com/competencies/11_ai_knowledge_engineering"
    note: "The software-domain instances (platform knowledge base + from-scratch MCP servers) and the AI-tool serving layer."
  ai_orchestrated_delivery:
    $ref: "https://zc8.com/competencies/20_ai_orchestrated_delivery"
    note: "Where a per-repo knowledge base becomes the institutional memory a memoryless AI author operates from."