research-document

Metadata Standard

Metadata Standard

Purpose

The repository needs one canonical front matter schema for every authored Markdown artifact.

This schema must support:

  • human navigation
  • faceted search
  • AI retrieval
  • lifecycle governance
  • supersession
  • graph generation

Required Fields

id: aie-handbook-foundations
title: Part 1 - Foundations
abstract: Short canonical summary of the artifact.
author:
  - Kevin Miller
date: 2026-07-20
discipline: ai-engineering
project: ai-engineering-course
research_area: foundations
document_type: handbook-part
status: draft
confidence: 0.88
evidence_level: synthesized
canonical: true
tags:
  - context-engineering
  - durable-artifacts
keywords:
  - validated progress
  - retrieval
related: []
supersedes: []
superseded_by: []
reading_time_minutes: 4
summary: One-sentence display summary.
chapter: 1
lesson: 1
sequence: 1
reviewed_on: 2026-07-21
reviewers:
  - research-reviewer
provenance:
  source_type: authored
  derived_from: []
  generated_by: null
audience:
  - researcher
  - engineer
concepts:
  - context-engineering
  - validated-progress
registries:
  evidence: []
  hypotheses: []
  experiments: []

Controlled Vocabularies

discipline

  • ai-engineering
  • software-engineering
  • design-research
  • cognitive-science
  • systems-thinking

document_type

  • concept-page
  • handbook-part
  • lesson
  • workbook
  • cheat-sheet
  • research-brief
  • research-notes
  • evidence-record
  • hypothesis-record
  • experiment-report
  • decision-record
  • roadmap
  • governance
  • notes
  • packaged-archive

status

  • seed
  • draft
  • review
  • accepted
  • superseded
  • archived

evidence_level

  • observation
  • evidence
  • synthesis
  • hypothesis
  • theory
  • accepted-knowledge
  • deprecated-knowledge
  • open-question

Canonical Rules

  • id must be globally unique and stable.
  • title is human-facing and may change.
  • id and path slugs should not depend on titles staying constant.
  • canonical: true is reserved for source artifacts that should receive backlinks and citation priority.
  • Derived artifacts must set canonical: false and include derived_from.

AI Retrieval Rules

  • abstract is mandatory because it is the safest chunk-level summary seed.
  • confidence and evidence_level should weight retrieval and ranking.
  • related, supersedes, and superseded_by should be used to expand retrieval neighborhoods.
  • concepts should map documents into the Knowledge Genome.

Validation Rules

Reject build-time publication when:

  • required metadata is missing
  • status is invalid
  • id collides with an existing artifact
  • supersedes points to a missing artifact
  • derived artifacts lack provenance